Oxford Studies in Ancient Philosophy volume 39

OX FORD STUDIES I N A N C I E N T P H I L O S O P H Y

This page intentionally left blank

OXFORD S T U D I E S IN A N C I E N T PHIL O S O P H Y EDITO R : B R A D I N W O O D

VOL U M E X X X I X         

3

3

Great Clarendon Street, Oxford   Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide in Oxford New York Auckland Cape Town Dar es Salaam Hong Kong Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore South Korea Switzerland Thailand Turkey Ukraine Vietnam Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries Published in the United States by Oxford University Press Inc., New York © Except where otherwise stated, Oxford University Press,  The moral rights of the authors have been asserted Database right Oxford University Press (maker) First published  All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this book in any other binding or cover and you must impose the same condition on any acquirer British Library Cataloguing in Publication Data Data available Library of Congress Cataloging in Publication Data Oxford studies in ancient philosophy.— Vol. xxxix ().—Oxford: Clarendon Press; New York: Oxford University Press, – v.;  cm. Annual. . Philosophy, Ancient—Periodicals. B.O .′—dc. – AACR  MARC-S Typeset by John Waś, Oxford Printed in Great Britain on acid-free paper by MPG Books Group, Bodmin and King’s Lynn ISBN –––– ISBN –––– (Pbk.)          

ADVI S O R Y B O A R D Professor Julia Annas, University of Arizona Professor Jonathan Barnes Professor A. A. Long, University of California, Berkeley Professor Martha Nussbaum, University of Chicago Professor David Sedley, University of Cambridge Professor Richard Sorabji, King’s College, University of London, and Wolfson College, Oxford Professor Gisela Striker, Harvard University Professor Christopher Taylor, Corpus Christi College, Oxford Contributions and books for review should be sent to the Editor, Professor Brad Inwood, Department of Classics, University of Toronto,  Queen’s Park, Toronto  , Canada (e-mail [email protected]). Contributors are asked to observe the ‘Notes for Contributors to Oxford Studies in Ancient Philosophy’, printed at the end of this volume. Up-to-date contact details, the latest version of Notes for Contributors, and publication schedules can be checked on the Oxford Studies in Ancient Philosophy website: www.oup.co.uk/philosophy/series/osap

This page intentionally left blank

CON T E N T S Plato on the Pleasures and Pains of Knowing



JAMES WARREN

Change in Aristotle’s Physics 



ANDREAS ANAGNOSTOP O U L O S

Separability vs. Difference: Parts and Capacities of the Soul in Aristotle



KLAUS CORCILIUS AND P A V E L G R E G O R I C

Essence and per se Predication in Aristotle’s Metaphysics Ζ



MICHAIL M. PERAMATZ I S

Form and Inheritance in Aristotle’s Embryology



JESSICA GELBER

Aristotelian Philia, Modern Friendship?



ALEXANDER NEHAMAS

Sensation and Scepticism in Plotinus



SARA MAGRIN

Aiming and Determining: A Discussion of Iakovos Vasiliou, Aiming at Virtue in Plato



C. C. W. TAYLOR

Index Locorum



This page intentionally left blank

PLATO ON T H E P L E A S U R E S AND PAIN S O F K N O W I N G JA M E S W A R R E N

P often assures us that it is pleasant to acquire knowledge. In the Republic the philosopher is said to live the most pleasant life because only he experiences the true and pure pleasures to be had from acquiring knowledge of the special objects that are the Forms. Similarly, in the Philebus Socrates claims that learning is a good example of a pure pleasure, namely a pleasure that is essentially neither preceded nor followed by pain. But Plato is also well aware that the process of coming to know something is not always pleasant. Indeed, in matters that would seem to be for Plato of the utmost importance, he is quite clear that our progress towards knowledge can be accompanied by a variety of affective experiences, and it can often be difficult and painful. The claim in Republic  that the philosopher’s life is the most pleasant possible has often been thought problematic, not least because of the various passages which appear to depict philosophical life and philosophical progress as painful. I investigate this problem first by considering a stretch of argument at Philebus  –  in which Socrates tries to give an account of the nature of the pleasures of learning and which includes a specification of the conditions under which certain kinds of learning might be painful or a mixture of pleasure and pain (Section I). Teasing out the precise implications of what is said there will allow us to reconsider the pleasures and pains of the philosopher’s life outlined in the Republic, since Protarchus’ suggestion of the conditions under which learning might not be a pure pleasure but will instead be a relief from pain turns out to be directly applicable to the experience of © James Warren  I would like to thank Brad Inwood, David Sedley, Frisbee Sheffield, and Georgia Moroutsou for comments on an earlier version of this essay.



James Warren

the prisoner released from the cave in the allegory in Republic  (Section II). However, there remain some important obstacles in the way of producing a fully satisfying account of the hedonic life of the philosopher. One of these problems stems from an objection sometimes raised against the portrayal in book . This objection holds that the understanding of the nature of pleasure presumed in that argument should force Socrates to claim that only the acquisition of new philosophical knowledge and not the continued possession and exercise of philosophical knowledge is wonderfully pleasant (Section III). I canvass some possible answers to this problem (Section IV) and conclude that the analysis of various pleasures of learning in the Philebus can usefully be brought to bear on this question (Section V).

I At Philebus   –   Socrates and Protarchus discuss the pleasures associated with learning and try to give an account of their nature:    . Then let us also add to these the pleasures of learning, if indeed we are agreed that there is no such thing as hunger for learning connected with them, nor any pains that have their source in a hunger for learning.    . Here, too, I agree with you.    . Well, then, if after such filling with knowledge, people lose it again through forgetting, do you notice any kinds of pain?    . None that could be called inherent by nature, but in our reflections on what we undergo whenever, deprived of something, we are pained because of the usefulness of what was lost [οὔ τι ϕύσει γε, ἀλλ᾿ ἔν τισι λογισµοῖς τοῦ παθήµατος, ὅταν τις στερηθεὶς λυπηθῇ διὰ τὴν χρείαν].    . But, my dear, we are here concerned only with the natural affections themselves, independently of our reflection on them [χωρὶς τοῦ λογισµοῦ].    . Then you are right in saying that the lapse of knowledge never causes us any pain [χωρὶς λύπης ἡµῖν λήθη γίγνεται].    . Then we may say that the pleasures of learning are unmixed with pain and belong, not to the masses, but only to a few?   . How could one fail to agree? (trans. D. Frede, modified)

Socrates is looking for another example of a pure pleasure, that is a pleasure which is neither necessarily preceded nor necessarily followed by a pain. His first example was the pleasure of smell. The

Plato on the Pleasures and Pains of Knowing



pleasure of learning is the second example. It too, Socrates thinks, is a process of filling a lack, but since simply not-knowing-X is not painful and having-forgotten-X is not painful, then the pleasure of learning X is a pure pleasure. A brief comment specifies that these pleasures of learning that are unmixed with pains belong to ‘the few and not the many’ (  – ), which suggests that Socrates has in mind cases of learning that are not mundane examples of simply coming-to-know something. Most likely, the sort of learning intended is to be connected with the dialogue’s later discussions of the various special epistēmai. There are, of course, important differences between how the Philebus and the Republic imagine epistēmē and its objects. Nevertheless, in both dialogues there is an evident commitment to the idea that certain kinds of special cognitive achievement are to be associated with a particular and superior form of pleasure. Furthermore, in both cases the dominant model for understanding the pleasure of this form of achievement is the filling of some kind of lack which may or may not be recognized or painful. However, it seems quite implausible to think that a philosopher’s cognitive progress is unaccompanied by pains, frustrations, and the like, which are connected with the fact that there is a conscious desire to know or understand something as yet ungrasped. Plato himself is acutely aware that philosophical understanding is often hard-won. In fact, the Philebus passage is very careful to clarify the precise sense in which the pleasures of learning are unmixed with pain. Protarchus voices an important qualification at   – when he notes that, although the simple fact of forgetting is not itself painful, the fact of having forgotten can perhaps be said to be painful just in cases when a person comes to reflect upon his lack of previous knowledge and on occasions when that knowledge is needed. Socrates swiftly brushes aside Protarchus’ concern as irrelevant to the precise point he wishes to make. As he reminds Protarchus, what they want to grasp is the nature of these experiences in themselves, shorn of any further complicating factors. Socrates is right. There are lots of things I do not know for which it is true that I am entirely indifferent about not knowing them; the fact of my not knowing  Cf. S. Delcomminette, Le Philèbe de Platon: introduction à l’agathologie platonicienne (Leiden, ) [Le Philèbe], .  Cf. D. Frede (trans. and comm.), Platon: Philebos [Philebos] (Göttingen, ), –; Delcomminette, Le Philèbe,  and –.



James Warren

them causes me no distress. There are lots of things I did know and no longer know for which it is also true that I am entirely indifferent about not knowing them. To be sure, if I think that something I do not know (or used to know) is something that I ought to know or ought still to know, then that secondary thought might be something that causes me distress. But the first-order fact of simply not knowing is not painful. So learning something need not be a relief from something painful. And yet, Protarchus has pointed to something important. He has given an important set of conditions under which a lack of knowledge (whether the result of forgetting something previously known or, we might add, the simple lack of a piece of knowledge never previously possessed) might be rightly thought to be painful. The conditions are twofold: (i) the lack of knowledge must be noticed or reflected upon, and (ii) the knowledge that is lacking must be recognized as needed or necessary in some way. Each of the two is necessary but insufficient for the state to be painful: I might recognize I do not know the capital city of Botswana but feel no pain at that realization so long as I have no need to know it. Similarly, I might have a genuine need to know some important philosophical truths in the sense that my life will be miserable unless I come to acquire that knowledge. But so long as I remain unaware of this need, the simple fact of not knowing will not of itself be painful to me. When combined, however, the two conditions will be sufficient to generate pain attendant upon a desire to know. While the first of these conditions is often noted, the second is often missed. Yet both are obviously necessary since it is the second which is required to generate in the person concerned a desire to know whatever it is that he does not know and it is crucial for the presence of some kind of negative affective response. The full psychological commitments of Protarchus’ comment at   –  are worth further thought. Clearly, he is distinguishing between ‘something we undergo’ (a pathēma), which we can presume is what is later glossed as ‘a kind of deprivation’ (the state of lacking some piece of knowledge), and something additional, which we have already identified as a further necessary condition for this pathēma to be painful. Protarchus refers to this additional  e.g. Delcomminette, Le Philèbe, : ‘En effet, pour qu’il y ait désir, il faut qu’il y ait non seulement manque, mais encore manque conscient, si du moins le désir doit avoir une direction, un objet.’

Plato on the Pleasures and Pains of Knowing





factor as ‘logismoi’. The Philebus can provide a satisfying account of what logismos amounts to in this context if we look back to its initial stipulation that the best human life must consist in some kind of combination of both pleasure and reason ( – ). Socrates and Protarchus consider two extreme cases: on the one hand, a life which contains pleasure but is devoid of any cognitive capacities such as memory, knowledge, opinion, and wisdom and, on the other hand, a life which retains all those capacities but is without even the least experience of pleasure. Neither alternative seems to them to be choiceworthy, and the remainder of the dialogue proceeds with this conclusion taken as its basis. In outlining the life of a mollusc, the life of pleasure without reasoning, Socrates explains his point as follows: ‘Having no true opinion, nor believing that he is experiencing pleasure when he does so and, being deprived of reasoning [λογισµοῦ δὲ στερόµενον,   ], not being able to reason how he will experience pleasure at some later time, he lives not a human life but the life of some jellyfish or crustacean’ (  –). From this it seems that logismos is, first of all, something that is an essential pre-requisite for living a recognizable human life and, more specifically, is related to what we might call a capacity for self-awareness and for considering one’s well-being or hedonic state at non-present times. Such a capacity might not exhaust the range of what logismos may do, but it is the important characteristic for present purposes. In Protarchus’ proposal at  – too, an important condition of feeling the pain of an absence of understanding is the human capacity to reflect upon or notice that condition and perhaps also to compare it with some previous or hoped-for future state. It is possible, in that case, to give an account of the conditions under which an absence of knowledge is painful by making use of a distinction between first- and second-order knowledge according to which the presence or absence of the first-order knowledge can be the object of a second-order form of knowledge and in which this second-order knowledge will be the exercise of the human capacity here referred to as logismos. Take a case in which I come to know that I do not know X. Imagine also that coming to know that I do not know X is painful to me. It is true that I do not know X, of course, so what I have acquired in coming to know that I do not know is a different  The plural form is clearly not significant since Socrates’ immediate reply replaces it with the singular logismos ( ) with no apology; the replacement does not seem to bother Protarchus.



James Warren

truth. I know more than I did when I simply did not know X and did not know that I did not know X. We noted, remember, that for such a second-order knowledge of an absence of first-order knowledge to be registered as painful there would need in addition to be some awareness of the first-order knowledge that is lacking being something worth having. There must, in other words, be a recognized need for that first-order knowledge. The Philebus’ analysis of human psychological capacities can also supply that additional requirement, once again by referring to the capacity of logismos. The prospective and retrospective faculties associated with logismos at   are not only stressed as essential characteristics of human psychology; they are both also involved in what might at a cursory glance seem to be solely future-directed attitudes such as desire. Later in the dialogue, Socrates is preoccupied with arguing for a division between the roles of the body and of the soul in desire, but while he is doing so he states clearly that he thinks all desires and impulses which initiate a drive for the removal or replenishment involve some sort of memory ( –). Specifically, the memory involved in desire is a memory of the opposite state to that in which the desirer currently finds himself. The desire involved when a person is thirsty, for example, involves the memory of the state of not being thirsty which supplies the drive and impulse towards finding something to drink. Presumably, the drive to find a drink to remove a thirst involves the conjuring from memory of some appropriate representation of the proper state of that desire being fulfilled. Socrates then goes on to distinguish two cases involving a person who is in pain but can remember the pleasant things he lacks. In the first, he has a ‘clear hope or expectation’ of attaining what he lacks. In that case, the memory provides some pleasure while he is also experiencing pain ( –). In the second, he is both in pain and also aware that there is no hope of replenishment. In that case his suffering is twofold ( –). We should note, then, that hopes and desires all involve some activity of memory since it is memory which provides the store of experiences that can be drawn upon to generate the appropriate objects of pursuit in any given situation and which allows the animal to bring to mind some state (which it has experienced in the past) which is the opposite of its present condition.  See D. Frede, ‘Rumpelstiltskins’s Pleasures: True and False Pleasures in Plato’s Philebus’, Phronesis,  (), – at –; D. Russell, Plato on Pleasure and the Good Life [Plato on Pleasure] (Oxford, ), –. Delcomminette, Le Philèbe,

Plato on the Pleasures and Pains of Knowing



We can now offer a full account of the painful cases of coming-toknow which Protarchus mentions at  –: these are cases in which a first-order ignorance is recognized as a result of second-order reflection on a person’s own cognitive state. This ignorance might be simply something that the person has never known or it might be the result of a loss of memory. The same capacity for second-order reflection that can recognize present ignorance, logismos, is also responsible for the person being able either to reflect upon a prior state of knowledge or imagine a future state of comprehension, and in cases where the possession of the relevant piece of knowledge would serve some recognized end, this will generate a desire to know. That desire can be painful. Indeed, if it is to motivate the person sufficiently, its painful nature might itself be something instrumentally useful. In such a way we can imagine the possibility of knowledge causing pain. This is a possibility which might initially be surprising but which on reflection is something that is only to be expected, particularly when the knowledge concerned is of a certain sort, namely the knowledge of an important personal failing.

II Republic  contains the longest sustained account of the pleasure associated with a life of philosophy and also presents the most difficult problems for anyone trying to claim that the life of a fully fledged philosopher is pleasant while holding on firmly to the analysis of pleasure—even an intellectual pleasure—as the process of filling some kind of lack. Before we apply to this problem the analysis in the Philebus of the pleasures of learning and the pains of some kinds of ignorance, we should first consider the most famous Platonic account of the experience of radical and transformative cognitive progress, namely the story of the prisoner’s release from bondage and ascent from the cave into the sunlight at the beginning of Republic . The description of the ascent from the cave emphasizes not the pleasures of discovery and the satisfaction of intellectual lack but quite the opposite: the dizzying and startling effect produced by the taxing and disorienting acquisition of a new perspective on –, has a good account of the sense in which philosophy itself in the Philebus is imagined as a kind of desire (see   –,   –), an image familiar from other dialogues such as the Symposium or Phaedrus but present also in the Republic.



James Warren

reality and value. Indeed, Socrates repeatedly notes the pain and discomfort felt by the student on his way up out of the cave as the bright light and the journey take their toll. We might also relate his experience to the analysis offered by Protarchus. The release of the prisoner from his bonds and his ignorance (ἀϕροσύνη,   ) is painful perhaps because it makes that ignorance obvious to him. The first stage of his education reveals to him the truth of his situation: although he previously thought that he was viewing real objects, in fact he was viewing only shadows cast by the fire behind the puppeteers. Such a realization is hard to endure and the prisoner may well prefer to return to his previous comfortable acceptance of mere shadows. Indeed, the prisoner will be confused if he is told that, despite his struggles to cope with the glaring light, his eyesight is in fact now working better (  –). Socrates notes that when presented with new and more real objects for consideration the prisoner will become confused or at a loss and will perhaps even initially refuse to consider them, preferring instead to turn back towards the objects with which he is more familiar; a degree of compulsion is therefore needed to force him to persist through the uncomfortable—indeed painful—initial transition. We might also note that the freed prisoner feels pain not only when he emerges from underground into the light outside but also when he first turns round and looks away from the shadows to the fire within the cave. In that case, if the first stage of the conversion might plausibly be likened to the unsettling effects of a Socratic elenchus and the undermining of the passive acceptance of mere cultural norms, then this too—as well as the eventual first encounter with the dazzling realities of genuinely intelligible objects—is said to be a painful process. The prisoner is confronted with his own ignorance about things which he previously thought that he knew, but also, we are to assume, acquires a need or desire to know something of which he now realizes he is ignorant: just the two conditions  Cf. ἀλγοῖ,   ; ἀλγεῖν,   ; ὀδυνᾶσθαι,   . There is a helpful account of the experience of the freed prisoner in M. Schofield, ‘Metaspeleology’, in D. Scott (ed.), Maieusis: Essays in Ancient Philosophy in Honour of Myles Burnyeat (Oxford, ), – at –, which does not, however, ask specifically why it is painful.  A similar phenomenon is illustrated by the case of what Socrates calls ‘summoners’ of thought (παρακαλοῦντα,   ; cf. παρακλητικά,   ). Faced with conflicting appearances, the soul is forced into an aporia and is compelled to find a resolution to its confusion by summoning the intellect (  –  ). Socrates makes no reference there to the possibility that the confusion might be painful.

Plato on the Pleasures and Pains of Knowing



noted by Protarchus as sufficient to make a case of acquiring knowledge only a mixed pleasure. The overall portrayal of the prisoner’s experience might therefore be thought to pose a problem for what Socrates will eventually claim for the great intellectual pleasures of philosophical enlightenment. The budding philosopher-ruler will certainly turn his gaze towards new and more knowable objects, and he too might have to come to realize a prior ignorance. In some passages any pleasure that the philosopher will eventually experience from finally acquiring the truth does indeed seem to be connected to a kind of pain, presumably closely linked to the philosopher’s tremendous desire to acquire the truth. Socrates refers, for example, to the philosopher’s ‘birth pangs’ as he struggles to grasp each thing’s nature ( –) and, once the philosopher has achieved the goal of his intellectual desire, Socrates says that he then ‘would understand and truly live and be nourished and, in this way, be relieved of his pain’ (  –). Such comments invite us to think that any pleasure involved is mixed rather than pure. Perhaps the student’s intense desire to know that is often associated with the life of a philosopher, coupled with the realization that there are some terribly important things that he does not know, will always make philosophical progress a rather mixed affair in hedonic terms; the final hoped-for understanding will then be experienced not merely as a great pleasure, but also as a kind of relief. Such comments might be combined with the account of the prisoner’s difficult ascent from the cave to raise concerns about the plausibility of the claim in book  that the philosophical life is most pleasant in so far as it contains episodes in which the philosopher learns important truths. There are evidently cases in which learning the truth can also be associated with significant pain.  B. Gibbs, ‘Pleasure, Pain and Rhetoric in Republic ’ [‘Rhetoric’], in D. Baltzly, D. Blyth, and H. Tarrant (eds.), Power and Pleasure, Virtues and Vices (Prudentia suppl.; Auckland, ), –, comments at : ‘In Bk  Socrates appears to have forgotten his own warnings about the toils and pains and hardships involved in becoming a philosopher and living the philosophical life.’ I see no reason to think there is an inconsistency.  The use of psychic pregnancy, labour, and birth as a metaphor for intellectual progress and production is prominent also in the Symposium. At  – Diotima associates any pain that might be felt on this account specifically with the experience of those who are pregnant in the soul but faced with ugliness. When such people manage to associate with beauty instead, their pains recede, and they can produce their offspring. The message seems to be that intellectual progress (here: the bringing to



James Warren

Those concerns can be set aside, fortunately, once we understand properly the reasons for the prisoner’s pain. The prisoner is pained at being forced suddenly to view objects of increasing brightness. We can distinguish three aspects here: (i) the glare of the new objects of his sight, (ii) the fact of his being forced to view them, and (iii) the fact of this being a sudden turn from familiar to unfamiliar objects. The first aspect is presumably part of Socrates’ demonstration that the prisoner is being asked to turn his cognitive apparatus to objects that are more and more real—that is, have a greater share of ‘being’, are more purely ‘just’, ‘beautiful’, and so on—and are therefore more and more knowable. The cognitive apparatus, the ‘eye of the soul’, that had previously been dealing only with the dimmest objects is now being presented with objects that activate its powers of cognition more and more effectively. But such things take some getting used to, particularly when they occur by compulsion: it is difficult to adjust when moving from a dark room out into the light even though it is true to say that out in the daylight is where a person’s powers of sight work best. It is not therefore simply the fact of being faced with these more knowable objects that generates the pain; rather, the prisoner is pained at being compelled all of a sudden to turn from his previous and familiar objects of attention— the shadows—and being forced to keep his gaze on these new and surprising things. A life of philosophical progress and understanding is not per se painful, but it is so in the case of the prisoner in the cave because of the necessary compulsion and the shocking revelation involved in effecting a rapid transition from the prisoner’s dreadful initial state. When Socrates goes on to describe the education of the budding philosophers, in contrast, he makes it clear that they have to undergo a lengthy process of careful preparation that begins very fulfilment of psychic potential) is not per se painful, but can be so if undertaken in the wrong circumstances or for the sake of the wrong kind of object. See further F. C. C. Sheffield, Plato’s Symposium: The Ethics of Desire (Oxford, ), –. Socrates’ special form of midwifery described in the Theaetetus is also dedicated to first bringing on birth pangs and then, ideally, allaying them (Theaet.  – ). Pain, in that case, is associated with the initial possibly confused or inchoate state of a person’s thoughts before Socrates can coax out a viable intellectual offspring. The pain of this intellectual labour may be generated by a kind of aporia, perhaps similar to the prisoner’s confusion at Rep.   – .  For more discussion of Plato’s use of imagery and metaphor in describing the philosophical life see A. Nightingale, Spectacles of Truth in Classical Greek Philosophy (Cambridge, ), –.

Plato on the Pleasures and Pains of Knowing



early in life (see e.g.   – ). We can therefore be more optimistic about the experience of philosophical students in the ideal city, since there is a significant difference between the tremendous involuntary cognitive upheavals experienced by someone plucked out of the cave and dragged into the light and the altogether less horrific experience of a young person educated in the ideally organized city and led willingly and slowly through a carefully constructed programme of philosophical education which has an assured level of success. Unlike the various people who complain of distress as a result of talking with Socrates and unlike the dazzled and pained prisoner escaping the cave at the beginning of Republic , a philosopher-in-training in the ideal city will be making intellectual progress in maximally beneficial circumstances. As the Philebus notes, there is a great difference between cases in which a desire— including presumably a desire to know something—is coupled with the realization that its satisfaction is extremely unlikely and a desire accompanied by the assurance that it will be fulfilled ( –). The painful experience of the prisoner may resemble the discomfort felt by people in Socrates’ own Athens struggling to make intellectual headway, but that should not generate a general pessimism about intellectual progress itself nor about the great pleasures which it will ideally produce. In short, philosophical progress may never be entirely straightforward, but we should be able to grant to Socrates the concession that, under ideal circumstances, the pain involved will be, at the very least, significantly lessened. And, in any case, elsewhere in the dialogue Socrates is often very upbeat about the pleasures of intellectual discovery. Consider, for example, his description of the ‘philosophical natures’ at   ff., especially   – . These fortunate people, fitted with all the traits of character necessary to allow them to be potential philosopher-rulers, desire ‘the pleasure of the soul itself by itself’ (τὴν τῆς ψυχῆς ἡδονὴν αὐτῆς καθ᾿ αὑτήν), a description very reminiscent of book ’s characterization of the pure and true pleasures at   ff. There is no mention here of the ‘pleasure of the soul by itself’ always being accompanied by pain, nor is there any need for such qualifications.  See M. Lane, ‘Virtue as the Love of Knowledge in Plato’s Symposium and Republic’, in D. Scott (ed.), Maieusis: Essays on Ancient Philosophy in Honour of Myles Burnyeat (Oxford, ), –, esp. –.



James Warren III

There are also, no doubt, distinctions to be drawn between the experiences of someone progressing towards philosophical understanding and a fully qualified philosopher-ruler, and those distinctions will be important in what follows. Still, Socrates is clearly interested in explaining the affective aspects of the philosophical life as a whole, and is also interested in explaining them in part by reference to the specific kinds of knowledge and ignorance—including knowing that one is ignorant or that one knows—that are involved in acquiring and possessing philosophical understanding. We can now approach directly the most significant difficulty which has been raised both for the characterization of the pleasures of learning in the Philebus and also for the account of the philosopher’s pleasures in Republic . In both works the emphasis is squarely on the pleasures of the process of coming-to-know something previously unknown or previously known but now forgotten. In that case it might remain mysterious how the philosopher might be said to continue to live a pleasant life once the necessary and previously lacking knowledge has been acquired. The difficulty begins with the closest Socrates comes in the Republic to an explicit statement of what he thinks pleasure and pain are. In the course of an argument intended to secure the conclusion that pleasure and pain are both to be distinguished from an intermediate state of calm or rest (ἡσυχία), he clearly states that pleasure and pain are both changes or motions: kinēseis (κίνησίς τις ἀµϕοτέρω ἐστόν,   –). That comment is left without further expansion until he comes some two pages later to give a more elaborate account of the different pleasures of the body and the soul. At   Socrates begins a new argument for the superiority of the philosopher’s life by offering two premisses. They deal with first the body and then the soul and assert an analogous relationship between their respective states of need. (i) Hunger, thirst, and the like are ‘emptyings’ (κενώσεις) of the state (ἕξις) of the body (  – ). (ii) Ignorance (ἄγνοια) and foolishness (ἀϕρονύνη) are ‘emptyings’ of the state of the soul (  –). He then infers:

Plato on the Pleasures and Pains of Knowing



(iii) Someone taking in nutrition (ὁ τῆς τροϕῆς µεταλαβάνων) and someone having understanding (ὁ νοῦν ἴσχων) would be filled (  –). By this, he presumably means that the ingestion of food and drink would remove the ‘emptying’ identified in (i) and the acquisition of understanding would remove that in (ii). One of the fundamental problems in interpreting this argument is the question whether Socrates exploits an ambiguity in the terms ‘emptying’ (κένωσις) and the associated ‘fulfilment’ (πλήρωσις) since they can both refer both to a state (of being empty, of being fulfilled) and to a process (of emptying, of fulfilling). From what we have seen of the argument so far, it is difficult to think that anything other than the state of ‘being empty’ is intended in (i) and (ii). Certainly, it is not easy to imagine that the ignorance in (ii) is meant to be only a process of becoming less knowledgeable. On the other hand, the present participle µεταλαµβάνων in (iii) might suggest a process of ingestion rather than a state of being free from e.g. hunger, whereas ἴσχων might rightly be thought to suggest a continued possession of understanding. Despite such uncertainties, the most satisfying overall interpretation holds that the states of ignorance or hunger are painful but the processes of eating or learning are pleasant. The question whether pleasures are always kinēseis becomes acute, of course, when we glance forward to the intended conclusion of the argument, which holds that the philosopher is the one most truly fulfilled since he grasps objects which are themselves most pure and true, and ‘are’ without qualification. If this refers merely to the process of acquiring understanding, then we might now agree only that the process of becoming a philosopher is exquisitely pleasant, but also infer that the resulting state of understanding is not. (Much as we might think that the process of eating when hungry is present while the state of feeling no hunger is not.) Socrates does offer some more information about  N. Pappas, Routledge Philosophy Guidebook to Plato and the Republic (London, ), –, mistakenly detects an inconsistency here: ‘[W]hereas the first half of the argument shrank from praising any pleasure that follows from the relief of pain, the second half endorses the relief from ignorance as though it could raise a person higher than the middle state of calm ( ). Nothing in the argument prepares for this claim, which feels like a gratuitous insistence on the pleasure of philosophy.’ The inconsistency disappears when we note that Socrates nowhere claims that ignorance is painful per se. Rather, it is often a painless lack and so the pleasure of learning is not necessarily preceded by pain.



James Warren

how he understands the pleasures of the philosophical life, but when it comes to the specific question whether these pleasures are associated entirely with the process of acquiring knowledge or may also include pleasures associated with the possession of knowledege, there is unfortunately only limited help to be found in the immediate context of this argument. A survey of the various references in the surrounding discussion to the sorts of pleasure said to characterize the philosophical life proves to be frustrating. In the description of the discussion between three spokesmen for the three kinds of life, each dedicated to the cultivation of one of the three parts of the soul, Socrates has various ways of describing the pleasures characteristic of the life dedicated to reason: the life of the lover of wisdom, the philosophos. Sometimes these expressions point in the direction suggested by the argument thus far, namely that intellectual pleasures are associated with the process of acquiring knowledge, that is to say, with learning. For example, when Socrates imagines the attitude of the other two sorts of people—the profit-lover and the victory-lover—to the philosopher’s life, he often puts it emphatically in terms of their attitude to the pleasures of learning (e.g.   , , µανθάνειν;  , µανθάνοντα). This lends support to the conception of philosophical pleasures as primarily—and perhaps exclusively—the pleasures of coming-to-know special objects. But this manner of expression is not applied consistently. Elsewhere, Socrates is prepared to talk about the pleasures of knowing (  , τῆς ἀπὸ τοῦ εἰδέναι ἡδονῆς) or about the pleasures of contemplating what is (  –, τῆς δὲ τοῦ ὄντος θέας, οἵαν ἡδονὴν ἔχει). There are also occasions on which Socrates refers in the same sentence both to the pleasures of learning and also to the pleasure of knowing. For example, at   –  he wonders how the philosopher will think of other pleasures in relation to his own preferred intellectual pleasures. He compares the other pleasures with the pleasure ‘of knowing how the truth stands’ (τὴν [sc. ἡδονὴν] τοῦ εἰδέναι τἀληθὲς ὅπῃ ἔχει) and ‘always being in such a state [sc. of pleasure] when learning’ (καὶ ἐν τοιούτῳ τινὶ ἀεὶ εἶναι µανθάνοντα). It is hard to be sure whether Socrates means in this case to refer  The phrasing echoes an earlier description of the special characteristic of the rational part of the soul, being that ‘with which we learn and which quite evidently is entirely focused upon knowing how the truth lies [πρὸς τὸ εἰδέναι τὴν ἀλήθειαν ὅπῃ ἔχει πᾶν ἀεὶ τέταται], and is least of all of them concerned with money and reputation’ (  –).

Plato on the Pleasures and Pains of Knowing



to two different kinds of pleasure that the philosopher may experience and to contrast both with the pleasures of the spirit or the appetites, and it is unclear whether the adverb ‘always’ (ἀεί) is supposed to show that the philosopher is always learning or that he is always experiencing pleasure when he learns. But it certainly suggests that there is pleasure associated with knowing the truth, of having acquired knowledge, whatever it may or may not then claim about that state. In short, the problem is that much of the argument so far is plausible only on the understanding that pleasure is the replenishing of a desire or lack. On the other hand, Socrates is apparently happy to talk as if there are also pleasures to be had from knowing, rather than learning, the special objects of the philosopher’s expertise. To be sure, we might understand ignorance as a state of cognitive lack much as hunger is a state of bodily lack, but if pleasure is associated with the process of making good that lack, there seems no other conclusion possible than that the pleasures of replenishing the soul—exquisite and intense though they may be since they are trained on pure and true objects—will be experienced only while the philosopher is acquiring knowledge. What pleasures can be left for the philosopher once he has the understanding he requires? If pleasure ceases when the process of replenishing ends, then ‘the more successful a philosopher is, the sooner his life will cease to be pleasant’. It is essential for the overall political project of the Republic that the ruling philosophers take up their role in the possession of a kind of knowledge that makes them experts in the areas relevant for political decision-making. Readers of the Republic are familiar with the concern that once they have acquired the required expertise the philosophers may be made to live a worse life by being obliged to give up their intellectual pursuits, descend back into the cave, and rule. The present worry is that the fully fledged philosophers may also be made to live a less pleasant life simply because the ascent out of the cave comes to an end.  For more on this somewhat opaque and contested sentence see below, pp. –  and n. .  J. C. B. Gosling and C. C. W. Taylor, The Greeks on Pleasure [Pleasure] (Oxford, ), –.  See also C. C. W. Taylor, ‘Platonic Ethics’, in S. Everson (ed.), Ethics (Cambridge, ), –, who objects to Socrates’ argument on the grounds that () ‘no doubt a truth once discovered does not have to be rediscovered, but a meal once eaten does not have to be eaten over again, and an intellectual life will require repeated acts



James Warren

We have already seen that the Republic contains a complex and varied story of the affective aspects of intellectual advancement, beyond the arguments concerning pleasure in book . And we have seen indications that Socrates wants to say that even the accomplished philosopher’s intellectual life would display a similarly complex affective aspect. Such considerations might alleviate some of the worries about the hedonic life of the philosopher-rulers or, less charitably, they might be taken merely to demonstrate a rift between what Socrates evidently wishes to claim about their pleasant lives and the inadequacy of the model of pleasure in Republic  to support such a claim. It would be far more satisfying if we could construct an account that will allow this expanded sense in which the philosopher, even once he has attained the knowledge required for being a ruler, will continue to live a life characterized by great intellectual pleasures and which also remains consistent with Socrates’ explicitly professed account of the nature of those intellectual pleasures in terms of a process of satisfying some kind of cognitive lack. But we are hampered in the construction of such an account by the fact that although Socrates spends rather a lot of time on describing the various epistemological and psychological, not to mention ethical, aspects of someone’s progress towards philosophical understanding and the comprehension of the Good itself, what that person’s life might be like after that point is left relatively underexplored. Perhaps this is excusable in the sense that Socrates’ major task is to persuade us that such an understanding is possible for a human to acquire and that, once properly installed as the rulers of a city, such rulers would set things up so as to be the best they could possibly be. Quite what it would be like to be such a ruler is not such a pressing concern. We are told, of course, that they will desire and endeavour to enact whatever is good and just, and we can extrapolate something about them having no desire for certain things the rest of us might hanker after—money, fame, familiar familial relationships, and the like—but that is about it. of thought (whether new discoveries or the recapitulation of truths already known) no less than a life of bodily satisfactions will require repeated episodes of bodily pleasure’. See also Gibbs, ‘Rhetoric’, –; Russell, Plato on Pleasure,  n. .  For a discussion of the various psychological, epistemological, and ethical aspects of dialectic, see M. M. McCabe, ‘Is Dialectic as Dialectic Does? The Virtue of Philosophical Conversation’, in B. Reis (ed.), The Virtuous Life in Greek Ethics (Cambridge, ), –.

Plato on the Pleasures and Pains of Knowing



IV A recent attempt by Sylvain Delcomminette to resolve the problem seems to me to be ultimately unsatisfactory. Nevertheless, it deserves serious consideration since it points the way to what I think is a more promising solution. Delcomminette’s overall interpretation aims to show that for Plato knowing (‘connaissance’) and learning (‘apprentisage’) are regularly held to be one and the same, or, perhaps better, that for Plato human knowledge always consists in the regular relearning of previously known things. Delcomminette’s principal piece of textual evidence comes from the immediate context of Republic . He notes the following question from earlier in Socrates’ defence of the superior pleasures of the philosopher’s life, a question which we have already considered briefly above: τὸν δὲ ϕιλόσοϕον, ἦν δ᾽ ἐγώ, τί οἰώµεθα τὰς ἄλλας ἡδονὰς νοµίζειν πρὸς τὴν τοῦ εἰδέναι τἀληθὲς ὅπῃ ἔχει καὶ ἐν τοιούτῳ τινὶ ἀεὶ εἶναι µανθάνοντα; οὐ πάνυ πόρρω; (  – ) I said, ‘How are we to think the philosopher considers the other pleasures in comparison with that of knowing how the truth is and always being in such a state when learning? Will he not think them greatly deficient?’

Delcomminette argues that Socrates here refers to the ‘plaisir de connaître le vrai tel qu’il est et d’être toujours dans un tel état en apprenant’ (Le Philèbe, ). If that is indeed how the second half of the sentence must be understood, then it would appear to lend explicit support to his proposal that the philosopher’s life is best understood as a kind of ‘apprentisage permanent’. He further supports this interpretation by appealing to the Symposium’s famous account at  –  of human psychological flux, in which Diotima claims that . . . not only does one branch of knowledge [ἐπιστήµη] come to be in us while another passes away and . . . we are never the same even in respect of our knowledge, but . . . each single piece of knowledge has the same fate. For what we call studying [µελετᾶν] exists because knowledge is leaving us, because forgetting is the departure of knowledge, while studying puts back a fresh memory in place of what went away, thereby preserving a piece of  Delcomminette, Le Philèbe, : ‘En réalité, tant dans la République que dans le Philèbe, le plaisir pur résulte bien du processus d’apprentissage, mais en tant précisément qu’il est identique à la connaissance’ (emphasis original).



James Warren

knowledge, so that it seems to be the same. (Trans. A. Nehamas and P. Woodruff)

I have already noted that the passage at Republic   –  is not entirely clear in its commitments. The text itself is debated, and it is therefore understandable that different translators render the sentence differently. In that case, it is prudent not to rely heavily on a particular interpretation of a controversial passage. In addition, the reference to the Symposium’s notion of psychological flux is not consistent with the most plausible interpretation of the contrast between the pleasures of the body and those of the soul as outlined in Republic . To make that inconsistency clear, it is necessary to return to the argument we left at   with Socrates having just set out an initial analogy between fillings and emptyings of the body and the soul. With a full account of Socrates’ conception of the nature of the philosopher’s pleasures, we might then be able to give an informed answer to the question of the pleasures of a philosopher’s life after he has come to know the Forms. Socrates continues: (iv) A filling with what is to a greater degree is more truly a filling than a filling with what is to a lesser degree (πλήρωσις δὲ ἀληθεστέρα τοῦ ἧττον ἢ τοῦ µᾶλλον ὄντος; δῆλον ὅτι τοῦ µᾶλλον,   –). The central difficulty here is in making good sense of the notion of degrees of being and then applying it to the intended analogue of de J. Adam (ed. and comm.), The Republic of Plato,  vols. (Cambridge, ), devotes appendix  of his commentary on book  to the discussion of how to construe these lines and, in particular, whether they contain one or two questions. I have cited them, following Slings’s Oxford text, with two questions. Burnet punctuates similarly, bracketing τῆς ἡδονῆς, which appears in some manuscripts after µανθάνοντα. Adam thinks there is only one question and retains τῆς ἡδονῆς, to read: καὶ ἐν τοιούτῳ τινὶ ἀεὶ εἶναι µανθάνοντα τῆς ἡδονῆς; οὐ πάνυ πόρρω καὶ καλεῖν . . ., translating: ‘compared with that [pleasure] of knowing how the truth stands and always enjoying a kindred sort of pleasure while he learns? Will he not think them very far away and . . .?’ Recent translations reflect the difficulty of fixing the precise intended meaning. For example, Griffith’s translation has ‘the pleasure of knowing where the truth lies and always enjoying some similar sort of pleasure while he is learning it? Won’t he regard them as far inferior?’; Grube’s translation has ‘[pleasure] of knowing where the truth lies and always being in some such pleasant condition while learning’.  Literally, the filling with what is is a ‘truer’ filling. The idea is presumably that filling a body with food is less of a filling than filling a soul with knowledge. J. Annas, An Introduction to Plato’s Republic [Introduction] (Oxford, ), , complains of an illegitimate slide from ‘being filled with what really/truly is’ to ‘being really/truly filled’.

Plato on the Pleasures and Pains of Knowing





grees of filling. Socrates himself helps only a little with the first of these problems, since he merely reminds Glaucon in a brisk fashion of a previously agreed distinction between things which share in ‘pure being’ and those which do not. Even so, there is enough spelt out in these lines for us to be fairly confident about Socrates’ view. The general contrast he invokes is between bodily nutrition and the means of caring for the soul (  –); the former obviously makes use of food, drink, and the like, while the latter makes use of true opinion, knowledge, understanding, and every virtue (  –). The question of degrees or categories of ‘what is’ is then explained by a contrast between two kinds of filling, their objects, and their proper location, which is spelt out in the next few exchanges between Socrates and Glaucon. The contrast is complex, but worth exploring carefully because it holds the key to the remaining argument. There is both a kind of filling related to what is always the same, what is immortal, and the truth, which is itself of such a kind and comes to be in such a thing, and, in contrast, another which is related to what is never the same, is mortal, is itself of such a kind, and comes to be in such a thing. It emerges, therefore, that there are three variables involved in the complex set of associations which Socrates wishes to use. There are what we might call (a) the subject of the filling or thing to be filled, (b) the method of filling, and (c) the object of filling, viz. whatever is used to fill (a). Learning, for example, is a method of filling which is taken to be a means of seeing to the care of the soul, and knowledge is necessarily related to objects which are changeless and true. Through learning, we fill the soul with knowledge of these changeless and true objects. Eating, on the other hand, is a means of seeing to the care of the body but is related to objects which are changeable and inconsistent. Socrates insists that the character of the kind of filling is determined by the character of its object, so learning itself is of a kind with its objects. He also insists that the kind of filling comes to be in something which is also of such a kind as it and its objects, so knowledge—which is stable and unchanging—comes to be in a  Cf. S. Rosen, Plato’s Republic: A Study (New Haven and London, ), – . Annas, Introduction, –, is unhappy with this section. She wonders () ‘how what is changeless can come about in what is changeless’ and is also concerned because () ‘it is not clear how this passage should be related to claims elsewhere about the Forms. For the contrast drawn here is not one between Forms and other things, since it has as much application to soul and body as to other things ( ), and the soul is not a Form.’



James Warren

soul which is also immortal and, in important ways, unchanging. The fulfilment of the body’s needs, in contrast, has as its objects perceptible items, bits of food and so on, is itself only temporary— because it has to be constantly repeated using always new items— and comes to be in something equally temporary and changeable, namely the body. A chain of explanation is set in place. The important determining factor is the nature of the ultimate object used for the filling in each case. The nature of the object then determines the nature of the filling itself, which must in turn be related to an appropriate subject to be filled. It remains only for Socrates to spell out the distinctions between the two sets of relations and to rank them. Unfortunately, the text of    has been transmitted in a corrupt state, so it is not easy to see how the argument begins. The conclusion, however, at   – is what we might have expected, namely that the forms of care for the body have a lesser share in being and truth than the forms of care for the soul. And at    it is added that the body differs from the soul in the same way, namely because the body shares less in being and truth than the soul. Something that is always the same shares more in being and truth than something which is not always the same. And if that is the case for the objects of the fillings, then it must also be the case for the fillings themselves. It should be clear why the thrust of this argument sits poorly with the idea that Socrates here holds the view that the philosopher’s soul is in a state of permanent learning of the kind suggested by Diotima at Symposium  – . The argument as a whole rests on the assumption that the filling appropriate for the soul is the filling of something that is always alike, immortal, and true with something that shares those characteristics. The central contrast is between the stability and permanence of the filling appropriate to the soul and the impermanence and changeability of the body and the objects in which it takes pleasure. The upshot of this argument in book  is that filling a bodily need is less truly a filling than filling a psychic need. The subject being filled (the body), the means of filling (eating), and the items used for the filling (food) are all changeable and inconstant. Hunger is sated only temporarily. The body and the food used to feed it are such that the filling cannot be permanent and is at best only ever partial. As he later comments, those who are try Slings ad loc. comments: ‘locus desperatus’. See Adam ad loc. and his appendix  to book  for further discussion and for his own preferred solution.

Plato on the Pleasures and Pains of Knowing



ing to satisfy their bodily desires fail to do so because they are filling something ‘which neither is, nor is water-tight, with things which are not’ (  –). Socrates tellingly compares their state to that of the Danaids of myth, who were condemned to toil fruitlessly trying to fill a leaky vessel by carrying water in a sieve, reusing an image he exploits to good effect in his conversation with Callicles at Gorgias   – . It would be very surprising—not to say unhelpful to his argument—if Socrates simultaneously holds that the intellectual pleasures he is praising for the stability of their objects and the stability and permanence of the soul which they fulfil in fact also display a similar kind of impermanence. And, what is more, Socrates stipulated back at  – that a philosophical nature would have to display an excellent memory. It is therefore very unlikely that the kind of psychological fluidity emphasized in the Symposium is something we are invited to bring to bear on the understanding of intellectual pleasures in the Republic. As the discussion progresses, there are more reasons offered in support of the view that the intellectual pleasures are thought of as being provoked by a change that has a permanent and stable result and, moreover, that they are associated with that part of us that is also permanent and unchanging. At    ff. Socrates brings all of the complicated discussion about different kinds of filling of different kinds of vessel with different kinds of object finally to bear on the question of pleasure. (v) Fulfilment by what is appropriate to our nature is pleasant (  ). (vi) That which is to a greater degree filled really and with things that are generates to a greater degree the enjoyment of true pleasure really and more truly. That which receives things that are to a lesser degree would be filled less truly and securely and would receive more untrustworthy and less true pleasure (  –). Critical attention has focused on (vi), but premiss (v) is undoubtedly just as important. When Socrates considers the pleasures enjoyed by those who are focused on bodily delights, it is  Cf. Gosling and Taylor, Pleasure, : ‘The thought seems to be that a firm lasting container filled with firm lasting contents can truly be said to be filled, whereas when one has a non-stable container and volatile contents it is only in a dubious sense to be called a filling at all: can one fill a hair-sieve with liquid?’; see also C. Bobonich, Plato’s Utopia Recast (Oxford, ), –.



James Warren

not coincidental that he casts such people and their pleasures in decidedly bestial terms. They are ‘like cattle, always looking down and bent over towards the ground, feeding at the table, growing fat, and mounting one another’ (  –). The metaphor of rutting herd animals continues as these people are described as butting one another with ‘iron horns and weapons’ (  ). Clearly, Socrates is encouraging us to disown such behaviour as not appropriate to our proper, human and rational, nature. It is a mere bestial nature which such pleasures fulfil, to the extent to which they can fulfil anything at all. The strong implication is that this vignette sketches the state of people who are focused on enjoying the pleasures produced when they try to satisfy the desires of the appetitive part of the soul. They have become misled by these impure and false pleasures and have created for themselves a misinformed conception of the good life. Tragically, the subsequent constant pandering to the desires of the appetitive part of the soul merely compounds their misfortune and further distorts their conceptions of value. Socrates goes on to refer explicitly to the elements familiar from the tripartition of the soul when he turns at    ff. to consider the pleasures of the spirited, moneyand victory-loving part of the soul. The discussion of pleasure is, after all, part of a much more extensive discussion of the relative happiness of different character types, an enquiry which has taken up much of this and the previous book and in which Socrates has made extensive use of the three parts of the soul to explain the origin and nature of various kinds of life. The question of the precise account of our human nature offered by Socrates in the Republic is complicated and controversial. But, in general terms, Socrates appears to be committed to an account of our nature which encourages us to identify ourselves, first and foremost, with the rational part of our soul which should take care of the other two parts. That is the overall message of the concluding sections of book  and their depiction of a person as composed of a human, a lion, and a many-headed beast (see  – ). The fulfilment of the needs of the rational soul is what best fulfils the best part of our nature and produces the most and finest pleasure. (It  Gibbs, ‘Rhetoric’, , cannot be correct in reading this passage as merely rhetorical.  For discussion see Annas, Introduction, –; D. Scott, ‘Plato’s Critique of the Democratic Character’, Phronesis,  (), –.

Plato on the Pleasures and Pains of Knowing



is impossible, at least while the soul is incarnated, to rid ourselves entirely of the desires and associated pleasures of the appetites and of spirit, but they ought at least to be controlled and reined in as far as possible:  – .) Such an identification with the rational part of the soul is necessary for the proper harmony of the soul’s parts and also, apparently, for the proper functioning of each individual part of the soul. Certainly, in the coda to this argument which once again surveys the various character types distinguished by the prominence of each one of the three parts of the soul, Socrates notes that in the absence of proper guidance by reason even the pleasures of the spirited or appetitive parts are not maximized. Only the philosophical and just soul, ruled by reason, properly enjoys the pleasures of the appetitive and spirited parts, since only with the guidance of reason will each enjoy ‘the best and truest of its own pleasures, in so far as it is possible’ (  –  ). A glutton, for example, will not enjoy the pleasures of the appetite as much as the philosopher, since he is not controlled by reason. Socrates spells out this view in the case of the money- and victoryloving character: his constant irascibility and overwhelming desire for victory prevent the successful satisfaction of his predominant desires (  – ). Premiss (vi) says very little that has not been explained or, at the very least, discussed already; it adds only the association of pleasure with the degree of fulfilment attained and the kind of object being used for the fulfilment. If we have by now accepted the notion outlined and explained at   –  that bodily fulfilment is less a fulfilment than proper intellectual fulfilment, then this new point follows without much trouble. We might still imagine a staunch supporter of the pleasures of eating and drinking objecting that he sees no particular reason to think that his preferred pleasures are any less intense than those of his more intellectually inclined counterpart. And, indeed, perhaps Socrates would agree with him; the problem with bodily pleasures, after all, is that they are based in such violent fluctuations and contrasts between satiety and emptiness that they can mislead people into concentrating on them to the detriment of the health of their souls (  – ). The obstinate hedonist might also claim that his preferred pleasures are no  Cf. Russell, Plato on Pleasure, –. Compare also Socrates’ diagnosis of the constant futile toils of the tyrannical man, trying desperately to satisfy his uncontrolled and changing appetites:  – .



James Warren

less truly pleasures than the more intellectual varieties. Socrates’ response to this objection is not so clear. He describes the pleasures in dispute here not only as mixed with pain but also as ‘copies and shadow-pictures’ of true pleasure (  –). It is not immediately clear what the precise connotations are of such a metaphor, which is evidently meant to resonate with other related passages of general epistemological and metaphysical importance elsewhere in the work. It might be wondered whether Socrates means to say that such ‘shadow-pictures’ of true pleasure are not really pleasures at all. Alternatively, he might mean only that they are pleasures of a deficient and unsatisfying sort: pale imitations of the rich and developed true intellectual pleasures. It is hard to be sure because, throughout this section of the book, Socrates sets out to distinguish between kinds of pleasure using a variety of criteria without pausing always to make clear their precise significance or the precise connections between them, although they are evidently connected. We hear, for example, about pleasures which are ‘pure’ (καθαραί), ‘true’ (ἀληθεῖς), or ‘really’ (ὄντως) pleasures and those which are not. We are also told that the more ‘true’ a pleasure is, the more ‘real’ it is (e.g.   ), and Socrates is prepared to praise those pleasures whose objects ‘share in pure being’ (καθαρᾶς οὐσίας µετέχειν,   ). It is unsurprising, therefore, that it is sometimes suggested that in the Republic Plato fails to distinguish satisfactorily between claiming that some pleasures are not genuine pleasures at all and that some pleasures, although still pleasant, are ‘false’ in the sense that their object or the content of the pleasure is somehow false. That particular problem can be left aside for our present pur ἆρ᾿ οὖν οὐκ ἀνάγκη καὶ ἡδοναῖς συνεῖναι µεµειγµέναις λύπαις, εἰδώλοις τῆς ἀληθοῦς ἡδονῆς καὶ ἐσκιαγραϕηµέναις; Compare shadows as an object of εἰκασία ( – ); shadows in the cave ( –); painters as imitators of things that are themselves only εἴδωλα ( –); σκιαγραϕία as a form of deceptive appearance ( –).  See D. Frede, ‘Disintegration and Restoration: Pleasure and Pain in Plato’s Philebus’, in R. Kraut (ed.), The Cambridge Companion to Plato (Cambridge, ), – at –; and cf. Gosling and Taylor, Pleasure, – and –. It is also sometimes claimed that the necessary distinction is found more properly articulated in the Philebus. (Much discussion of the Philebus has concentrated on the proper understanding of the characterization of a pleasure as ‘true’ or ‘false’, but on the question of the relationship between the categories of true/false and pure/impure pleasures in that dialogue see Frede, this note, and also J. Cooper, ‘Plato’s Theory of Human Good in the Philebus’, Journal of Philosophy,  (), –. Note the close assimilation of the purity and truth of pleasure at Phileb.  –.) There is a good discussion of the ‘shadow-pictures’ of pleasure of Republic  in M. M. Erginel,

Plato on the Pleasures and Pains of Knowing



poses since, whatever we finally decide about the precise nature of these ‘shadow-pictures’ of true pleasure, we are still faced with the problem of what to say about the hedonic life of the fully fledged philosopher and his true and genuine pleasures. The argument so far, after all, strongly implies that the true pleasure to be had is associated with the kinēsis that is learning, filling up the cognitive lack that is ignorance, and that this filling is something which takes place in a stable and everlasting container, uses stable and everlasting objects, and therefore does not have to be repeated. Indeed, the fact that, unlike the bodily pleasures with which it is contrasted, such true pleasure is not in constant need of repetition is one of the reasons why Socrates thinks it is a superior form of pleasure. There are a number of ways in which Socrates can respond to the concern that the philosophical life will contain great and exquisite pleasures while the philosopher is in the process of acquiring knowledge but, after that point, will seem to have many fewer opportunities for continued enjoyment of those same pleasures. A first general point to bear in mind is that Socrates nowhere promised to show that the philosopher is at every moment of his life experiencing the greatest pleasures; we are not to imagine him in a constant state of intellectual ecstasy. Rather, the demonstrandum is that the philosopher’s life, taken as a whole, is most pleasant. This lessens the need for us to show that the philosopher is at all times experiencing the greatest pleasures, since we might well agree that the philosopher’s life will contain at some point in it the greatest, most true, and purest pleasures. Second, the life of a fully fledged philosopher will nevertheless contain a great variety of pleasures and, moreover, will contain pleasures which are still superior to those found in any other possible life. Socrates asserts at   –   that only in the light of the rule of reason in the soul is a person able to experience appetitive and spirited pleasures of the best and truest variety available. Of course, these pleasures are never going to be pure and true in the sense that the intellectual pleasures are, but nevertheless this passage serves as an important reminder that the philosopher will also continue to enjoy the pleasures of eating and so on and, more to the point, we are assured that because of the harmonious arrangement of his soul, free from internal conflict (  ), and the fact that ‘Pleasures in Republic IX’ [‘Pleasures’] (Ph.D. diss., University of Texas at Austin, ), ch. .



James Warren

therefore his desires are all marshalled and arranged by reason sub specie boni, he will be able to do so to the greatest extent possible for any person. In contrast, when one of the other parts of the soul is dominant, it forces its fellow soul-parts to pursue pleasures which are alien to them (  –). This observation remains unsatisfying to the extent that it concerns pleasures that are not related directly to the philosopher’s special emphasis on living a life identified with the activities of reason. A philosopher might well take great pleasure from eating his healthy diet, perhaps even more pleasure than the glutton or gourmand take from theirs, because he eats in a way that is ultimately guided by a conception of the good. But that still falls short of the hoped-for account of why a philosophical life remains most pleasant, and Socrates himself seems most interested in locating the superiority of the philosopher’s hedonic life in its being related closely to the experience of pleasures that are both true and also—as we have seen— appropriate to our best nature. Another possibility is that Socrates has in mind a wider conception of intellectual pleasures than just those concerned with Forms. Perhaps the fully qualified philosopher will continue his intellectual development by acquiring various true beliefs, finding out various facts about the world, reading literature or history, even doing some mathematics or revisiting his old harmonic theory textbooks and trying out some new problems. There is some textual support for such a view since at   –  Socrates groups not only the pleasures of knowledge and understanding but also true beliefs and generally all virtue against those concerned with food,  This is important additional support for the earlier contention ( – ) that the philosopher’s life is the most pleasant because only he has experienced true and pure intellectual pleasures and, as an expert in all pleasures, he would judge his life to be the most pleasant. C. C. W. Taylor, ‘Plato and Aristotle on the Criterion of Real Pleasures’, in his Pleasure, Mind, and Soul: Selected Papers in Ancient Philosophy (Oxford, ), –, does not note this important point and perhaps as a result rejects Socrates’ claim that the philosopher excels other types of men in his experience of pleasure. Erginel, ‘Pleasures’, ch. , has a wide-ranging discussion of this argument.  An interpretation of this kind is developed by Erginel, ‘Pleasures’, ch. , which he further supports by relying on a scalar interpretation of the being and truth of various objects of pleasure: there are, in other words, objects that stand between those that are ‘always the same, immortal, and the truth’ and those that are ‘never alike and mortal’ (  –). In J. Warren, ‘Pleasure, Plutarch’s Non posse, and Plato’s Republic’, forthcoming in Classical Quarterly, I argue that Plutarch uses a similarly expanded notion of the pleasures of reason in his criticism of Epicurean hedonism.

Plato on the Pleasures and Pains of Knowing



drink, and nutrition as a whole. All of the former types, it seems, will produce pleasures that are superior to those of the latter type. This is a further important reminder that the philosopher-ruler will not be a disembodied soul; he will continue to live and take enjoyment in various pursuits and activities beyond the special case of acquiring knowledge of perfect, intelligible, and everlasting Forms. But yet again, the proposed pleasures that are said to characterize the philosopher’s life are not obviously of a kind that cannot also be enjoyed by those less fortunate people who cannot be said to live a philosophical life. A rather wide group of people, we might imagine, can come to acquire and perhaps enjoy acquiring true beliefs or a grasp of empirical facts, even if we grant the possibility that such learning would be transformed significantly by a proper grasp of the nature of the good.

V A better answer can be given if we allow ourselves to work with a richer conception of the workings of the reasoning part of the soul. We have already seen signs in the discussion of the pleasures and pains involved in the philosopher’s ascent from the cave that the Republic must be using something like the conception of first- and second-order knowledge that Protarchus expresses in more explicit terms in the Philebus and that in both cases there is an evident interest in the pleasure and pain to be associated with a kind of reflexive knowledge. My further claim is that the analysis taken from the Philebus can be used to alleviate the problem of the philosopher’s intellectual pleasures in Republic  by pointing to a set of pleasures that the philosopher will be able to experience after the point of coming to know the Forms and that are not accessible in any way or to any degree by someone who has not similarly come to know the Forms. It is certainly wrong to say that the philosopher, once he has acquired knowledge of the Forms, will continue to experience the pleasures of that initial and extremely satisfying discovery because he will in one way or another ‘forget’ what he has learnt. It is hard to square such a proposal with the evident emphasis in Republic  on the stability and permanence of not only the object of philosophical knowledge but also the rational soul with which that knowledge is acquired, not to mention the insistence that phi-



James Warren

losophers will have excellent powers of memory ( ). However, Delcomminette’s proposal points in the right direction because it is true that the philosopher’s life will continue to be characterized by various changes in the soul that might reasonably be said to be examples of coming-to-know of the sort that would qualify as potential pleasures. We do not, on the other hand, need to posit some kind of constant state of learning and forgetting of any first-order knowledge, since any psychological changes necessary can be restricted to the second-order kinds of knowing. Protarchus drew our attention to the possibility of there being a second order of reflection on what a person knows and the connections this might have to experiences of pleasure or pain in coming to know something, since it allows a distinction between comingto-know in some cases in which one does and in other cases in which one does not also know that one does not know that something. There are cases in which this second-order knowledge that one does not know something is coupled with the fact that one previously did know that something, in which case we are right to talk in terms of ‘forgetting’ or ‘remembering’. But this is not true of all cases. There is surely, we might insist, an important distinction between having forgotten something and merely not having to mind something that we still know. That distinction is brought out most forcefully by the fact that in the case of something forgotten but now recognized as necessary to know, the previously held piece of knowledge is not easily remembered. Indeed, the difficulty of remembering that previously held piece of knowledge coupled with the recognized need for it is precisely the combination of factors that would make it plausible to say that the experience is a painful one. Once again we can turn to the Philebus for a more explicit expression of an idea that I want to suggest is relevant to Socrates’ claims in Republic . In his discussion with Protarchus, Socrates articulates a distinction between ‘remembering’ something that has been forgotten and ‘calling to mind’ something that has not been forgotten but has simply not been the focus of attention. At  – he distinguishes between two forms of ‘recollection’, anamnēsis: one in which the soul ‘takes up’ (ἀναλαµβάνῃ,   ) a memory, which is some And there is also the further question whether the person concerned knows that he has forgotten, which is a complicated combination of (i) not knowing X, (ii) knowing that he does not know X, and (iii) knowing that he previously did know X.

Plato on the Pleasures and Pains of Knowing



thing originally experienced together with the body, and another in which the soul unearths or recovers (ἀναπολήσῃ,   ) a memory which it previously had lost (ἀπολέσασα,   ) of a perception or a piece of learning. Both, he says, can rightly be called examples of anamnēsis, although it is evident that we are not meant to think on this occasion of the special kind of recollection considered in the Meno and Phaedo: both forms of anamnēsis in the Philebus deal with perceptions or things learnt during a person’s life. This distinction between the two forms is embedded in a longer section that tries to clarify what memory is ( – ), since Socrates wishes to use the pleasures belonging to memory as an example of pleasures which belong only to the soul and not to the soul and body together. His principal concern, therefore, is to show that even in cases where what is being recalled is something that originally involved the body (a perception or some other kind of experience) the recollection of it involves only the soul. But whatever the other subtleties of the passage, it is reasonable to identify here a recognition on Socrates’ part that there is an important difference between the soul remembering something that has been forgotten—that is, a memory that has been lost (  –)—and the soul recovering something stored in the memory. Calling a piece of latent knowledge to mind can hardly be called ‘learning’, of course, nor can it really be called ‘remembering’. But the Socrates of the Philebus apparently thinks it might still be called a case of anamnēsis, and what matters for our purposes is that he does identify a psychological capacity involving the taking up of things stored in the memory. At this point we might be put in mind of not only Aristotle’s discussion of anamnēsis in De memoria , but also his useful distinction between the first and second actualities of knowing. Of course, Aristotle has his own account of how it can be both pleasant to learn and also pleasant to possess and use already learnt knowledge. And that account is in turn related to a more general disagreement between him and Plato on the necessity of thinking of pleasure as a kind of kinēsis. That disagreement is already well known and further consideration of it would be a distraction from the main point at hand. Still, it seems that the distinction between two species of anamnēsis at Philebus   offers something that will do the same job as Aristotle’s useful distinction. It provides a distinction between  There is a helpful discussion of this passage in Delcomminette, Le Philèbe, –.



James Warren

the kinēsis that is the remembrance of knowledge that has been forgotten and a kinēsis that is the bringing to mind of knowledge that has become somehow latent but can be activated at will and without effort when it is found to be necessary. We can now return to the Republic. Philosopher-rulers will not spend all of their time ruling. Indeed, we are told explicitly that for the most part they will be able to spend their time in philosophy ( –). Socrates does not say much about what kind of philosophy a philosopher-ruler will do, nor does he give a detailed account of what a philosopher-ruler will do as he rules, but some of what he does say will allow me to illustrate some of the pleasures which will characterize the fully fledged philosopher-ruler’s life. When the philosopher is not ruling but instead doing philosophy, we can assume that either he is acquiring more philosophical knowledge—which is pleasant in an uncontroversial way—or he is reviewing and revisiting philosophical knowledge he already has. The latter activity is neatly characterized as the first kind of anamnēsis canvassed in the Philebus: the soul takes up something stored in the memory. The philosopher will turn his attention back to this or that Form or consider how the Forms are related to one another. Whatever he does, precisely, it is reasonable to think that it involves a change of a kind in his soul, the bringing to mind of latent knowledge, and is therefore something we can readily classify as an intellectual pleasure. These pleasures are both most plausibly imagined as kinēseis and, furthermore, are related directly to his being a philosopher-ruler. When the philosopher-ruler is actually ruling, although it is evidently not his preferred activity, it too presents opportunities for intellectual pleasures. At   Socrates likens the activity of the philosophers in constructing the ideal city to that of a painter. Just as the painter will work by looking back and forth between his picture and the original that he is attempting to depict, so too the philosophers will turn their attention first to the Forms they are attempting to instantiate as best they can, then to their city, then back to the Form, and so on. Throughout this process they of course know what Justice is, for example, but the constant movement back and forth between the model and the original might rightly be said to correspond to a psychological shift of attention from the perceptible construction to the ideal model and back again. They call to mind the original and then, in the light of that, they turn their at-

Plato on the Pleasures and Pains of Knowing



tention back to the city. And the process goes on, not stopping when the city is complete but also being called into action whenever the philosophers are called to make a judgement about a specific question of the city’s affairs. To borrow the apparatus of Protarchus’ observation, we might say that the philosopher-ruler will continue to use his faculty of logismos and in doing so call to mind and reconsider various things that he knows both when he is doing philosophy and when he is ruling. He needs to do this not because he has in any reasonable sense of the term forgotten the nature of the Just or the Fine but because, although he does know these things, his attention is moving to and from these intelligible objects. It is not at all implausible to imagine that on each occasion when he turns once again to consider, for example, the Just, this will involve a coming-to-know that, while not of the significance of the first time he came to know its nature, will share enough of the characteristics of that first occasion to be thought of as a kinēsis that fills a kind of lack in the soul. And, as such, it can be thought of as a true and pure pleasure. Finally, it is a kind of pleasure that is entirely unavailable to anyone who is not a philosopher-ruler. Corpus Christi College, Cambridge

BIBLIOGRAPHY Adam, J. (ed. and comm.), The Republic of Plato,  vols. (Cambridge, ). Annas, J., An Introduction to Plato’s Republic [Introduction] (Oxford, ). Bobonich, C., Plato’s Utopia Recast (Oxford, ). Cooper, J., ‘Plato’s Theory of Human Good in the Philebus’, Journal of Philosophy,  (), –; repr. in his Reason and Emotion (Princeton, ), –. Delcomminette, S., Le Philèbe de Platon: introduction à l’agathologie platonicienne [Le Philèbe] (Leiden, ). Erginel, M. M., ‘Pleasures in Republic IX’ [‘Pleasures’] (Ph.D. diss., University of Texas at Austin, ). Frede, D., ‘Disintegration and Restoration: Pleasure and Pain in Plato’s Philebus’, in R. Kraut (ed.), The Cambridge Companion to Plato (Cambridge, ), –. (trans. and comm.), Platon: Philebos [Philebos] (Göttingen, ).



James Warren

‘Rumpelstiltskins’s Pleasures: True and False Pleasures in Plato’s Philebus’, Phronesis,  (), –; repr. in G. Fine (ed.), Plato II (Oxford, ), –. Gibbs, B., ‘Pleasure, Pain and Rhetoric in Republic ’ [‘Rhetoric’], in D. Baltzly, D. Blyth, and H. Tarrant (eds.), Power and Pleasure, Virtues and Vices (Prudentia suppl.; Auckland, ), –. Gosling, J. C. B., and Taylor, C. C. W., The Greeks on Pleasure [Pleasure] (Oxford, ). Lane, M., ‘Virtue as the Love of Knowledge in Plato’s Symposium and Republic’, in D. Scott (ed.), Maieusis: Essays on Ancient Philosophy in Honour of Myles Burnyeat (Oxford, ), –. McCabe, M. M., ‘Is Dialectic as Dialectic Does? The Virtue of Philosophical Conversation’, in B. Reis (ed.), The Virtuous Life in Greek Ethics (Cambridge, ), –. Nightingale, A., Spectacles of Truth in Classical Greek Philosophy (Cambridge, ). Pappas, N., Routledge Philosophy Guidebook to Plato and the Republic (London, ). Rosen, S., Plato’s Republic: A Study (New Haven and London, ). Russell, D., Plato on Pleasure and the Good Life [Plato on Pleasure] (Oxford, ). Schofield, M., ‘Metaspeleology’, in D. Scott (ed.), Maieusis: Essays in Ancient Philosophy in Honour of Myles Burnyeat (Oxford, ), –. Scott, D., ‘Plato’s Critique of the Democratic Character’, Phronesis,  (), –. Sheffield, F. C. C., Plato’s Symposium: The Ethics of Desire (Oxford, ). Taylor, C. C. W., ‘Plato and Aristotle on the Criterion of Real Pleasures’, in his Pleasure, Mind, and Soul: Selected Papers in Ancient Philosophy (Oxford, ), –. ‘Platonic Ethics’, in S. Everson (ed.), Ethics (Cambridge, ), –. Warren, J., ‘Pleasure, Plutarch’s Non posse, and Plato’s Republic’, forthcoming in Classical Quarterly.

CH ANGE IN AR I S T O T L E ’ S P H Y S I C S 3 ANDREAS A N A G N O S T O P O U L O S

I Physics .  Aristotle famously defines change (kinēsis) as ‘the entelecheia of the potential being, qua such [potential]’ (a–). Ostensibly, he defines change because the term kinēsis figures in the definition of nature (phusis) given earlier in the Physics, so that we must understand what change is if we are to understand what nature is (b–, –; b–). Aristotle also saw his definition as a major advance on what had been the fundamental topic of natural philosophy for generations and a stumbling-block for his predecessors. But few readers have echoed Aristotle’s own satisfaction with his definition. A primary reason for this dissatisfaction is that the definition appears to pick out the products of change rather than changes themselves. Call this the ‘product puzzle’. For example, the entelecheia—usually understood as ‘actuality’—corresponding to a © Andreas Anagnostopoulos  I am grateful to audiences in Berkeley and Berlin, as well as to the participants in my seminar on these issues in Berlin. Comments from Jonathan Beere, István Bodnár, Alan Code, Dorothea Frede, Brad Inwood, Sean Kelsey, John MacFarlane, Jacob Rosen, and an anonymous referee have improved this essay significantly. I owe a special debt to David Ebrey, Jessica Gelber, and Pieter Sjoerd Hasper, who have helped me enormously in thinking and writing about this topic. Finally, I am grateful for the support of the TOPOI Excellenzcluster in Berlin.  ἡ τοῦ δυνάµει ὄντος ἐντελέχεια, ᾗ τοιοῦτον, κίνησίς ἐστιν (a–). Compare the definition given in the mirror passage in Metaphysics Κ : τὴν τοῦ δυνάµει ᾗ τοιοῦτον ἐστιν ἐνέργεια λέγω κίνησιν (b). I have opted for ‘the potential being’ over both ‘what is potentially’ and ‘that which is potentially’, in order to avoid misunderstanding. Both the latter two options might be thought to pick out that being which enjoys merely potential existence. For example, if we are talking about a pile of bricks that is potentially a house, ‘what is potentially’ might seem to refer to the house (or its form) since the house, one might say, enjoys potential existence. Even if we avoid this construal, perhaps by adding a completing predicate (so: ‘what is potentially a house’ or ‘that which is a house potentially’), these constructions are in a different way too loose. They may suggest some being merely one-in-number with the potential being, rather than that whose very nature is to be potentially something.  Simplicius, however, claims that Aristotle defines change ‘marvellously’ (θαυµασίως) (In Phys. .  Diels).



Andreas Anagnostopoulos

potential house seems to be the house, rather than the process of becoming a house. Some commentators have responded to this puzzle by construing entelecheia as ‘actualization’; others by taking Aristotle to be talking not about a potential being, but about what is potentially becoming, e.g. what is potentially in the process of becoming a house. Both of these options, however, seem to smuggle the notion of change into the terms of the definition. According to Kosman’s seminal account, the solution to this puzzle lies in Aristotle’s addition of the phrase ‘qua such’. Although the product of change is the paradigmatic actuality of the potential being, the change, and it alone, is the actuality of the potential being ‘qua such [potential]’. Thus, while the definition characterizes change as (i) a genuine actuality of (ii) a genuine potential being, and so avoids circularity, (iii) the phrase ‘qua such’ ensures that it picks out changes but not the products of change. I shall refer to this skeletal interpretation as the ‘consensus interpretation’, since it has become entrenched in recent years.  So H. Cherniss, Aristotle’s Criticism of Presocratic Philosophy (Baltimore, ), ; W. D. Ross, Aristotle’s Physics [Physics] (Oxford, ),  and ; J. L. Ackrill, ‘Aristotle’s Distinction between Energeia and Kinesis’, in R. Bambrough (ed.), New Essays on Plato and Aristotle (New York, ), – at –; T. Penner, ‘Verbs and the Identity of Actions’ [‘Verbs’], in G. Pitcher and O. Wood (eds.), Ryle (Garden City, NY, ), – at –; A. L. Peck, ‘Aristotle on Κίνησις’, in J. Anton and G. Kustas (eds.), Essays in Ancient Greek Philosophy (Albany, NY, ), – at ; J. Kostman, ‘Aristotle’s Definition of Change’ [‘Definition’], History of Philosophy Quarterly,  (), –. On Ross and Peck see the following note.  Thus, change could be thought of as the actuality of such a potentiality, in the sense of the goal of that potentiality or what the potentiality is for. Aside from circularity, such an interpretation would violate Aristotle’s dictum that a potentiality is defined by reference to what it is for (Metaph. Θ , b–). See D. Charles, Aristotle’s Philosophy of Action [Action] (Ithaca, NY, ), –, and R. Heinaman, ‘Is Aristotle’s Definition of Change Circular?’, Apeiron,  (), –, for interpretations along these lines. Peck, ‘Aristotle on Κίνησις’, , and Ross, Physics, , suggest such a reading of δυνάµει ὄν, but this makes their kinetic readings of ἐντελέχεια (e.g. ‘actualization’) elsewhere superfluous. On the reading in E. Hussey, Aristotle’s Physics, Books III and IV [III & IV] (Oxford, ), see n.  below. I argue that the ‘potential becomer’ reading is unwarranted in sect. (b) below.  For the charge that the ‘actualization’ reading is circular see L. A. Kosman, ‘Aristotle’s Definition of Motion’ [‘Motion’], Phronesis,  (), – at –; M. L. Gill, ‘Aristotle’s Theory of Causal Action in Physics III. ’ [‘Causal Action’], Phronesis,  (), – at , and Aristotle on Substance: The Paradox of Unity [Paradox] (Princeton, ), ; T. K. Johansen, Aristotle on the SenseOrgans [Sense-Organs] (Cambridge, ), ; U. Coope, ‘Change and its Relation to Actuality and Potentiality’ [‘Change’], in G. Anagnostopoulos (ed.), A Companion to Aristotle (Oxford, ), – at –. See also n.  below.  Kosman, ‘Motion’.  Although it would be an overstatement to say that there is a single agreed in-

Change in Aristotle’s Physics 



In what follows I shall argue that two components of the consensus interpretation should be left aside. First, the idea that change is defined as an actuality involves difficult and obscure metaphysical theses that are unsupported by, and sometimes in conflict with, Aristotelian doctrines (Section ). Second, the idea that the quaphrase neutralizes a subsequent threat of picking out the products of change is in tension with the grammar of Aristotle’s definition and incompatible with his own explanation of the phrase’s function (Section ). I offer a comparatively straightforward interpretation that draws on well-attested Aristotelian theses about the metaphysics of change (Section ). Change is defined as the activity (rather than actuality) of potential being. There is thus no threat of picking out the products of change, and no obstacle to picking out changes. The role of the qua-phrase is to specify that change is the proper rather than accidental activity of a potential being. That is, change is the activity that a potential being engages in precisely because it is a potential being. This analysis facilitates an attractive account (Section ) of the definition’s role in the complex dialectic between Aristotle and those who harboured doubts about change. Consensus interpreters have generally held that Aristotle addresses—indeed must address— such doubts by bringing change into the sphere of actuality. By contrast, I argue that the primary contribution of Aristotle’s definition to this dialectic is not to specify change’s ontological status, but rather to certify the scientific respectability of change—its terpretation, the three components of the consensus reading I identify can be found in Kosman, ‘Motion’, as well as in Gill, ‘Causal Action’ and Paradox, –, and Coope, ‘Change’. Hussey, III & IV, appears to reject the consensus analysis of δυνάµει ὄν (potential being) at – but not always; see n.  below. S. Waterlow, Nature, Change, and Agency in Aristotle’s Physics: A Philosophical Study [Nature] (Oxford, ), Johansen, Sense-Organs, –, and M. F. Burnyeat, ‘De anima II ’, Phronesis,  (), – at , do not commit themselves to the consensus reading of the qua-phrase and Burnyeat has reservations about Kosman’s reliance on certain passages in Metaphysics Θ  and De anima . . C. Freeland, ‘Aristotle on Bodies, Matter, and Potentiality’, in J. Lennox and A. Gotthelf (eds.), Philosophical Issues in Aristotle’s Biology (Cambridge, ), – at , relies on the consensus reading of the qua-phrase. Among detractors, Charles, Action, –, and Kostman, ‘Definition’, cite problems in ascribing to Aristotle the view that change is an actuality, while R. Heinaman, ‘Kosman on Activity and Change’ [‘Kosman’], Oxford Studies in Ancient Philosophy,  (), –, raises a quite compelling objection to a feature of Kosman’s proposal as stated, but which, I shall suggest, Kosman can safely give up; see n.  below.



Andreas Anagnostopoulos

status as an object of scientific investigation and understanding. In characterizing change as the activity of potential being, Aristotle’s definition ties change to a kind of being (i.e. potential being) that falls outside the problematic dichotomy of opposite principles to which many of his predecessors were limited, such that they could not adequately explain change. And in specifying that change is the proper activity of potential being, the definition posits an intrinsic causal connection between potential being and the change that it undergoes. Aristotle’s definition thus shows not merely that change can be coherently characterized, but also how it can be rendered intelligible within the scope of the explanatory science of nature that the Physics sets out to construct. In closing (Section ), I address Aristotle’s use of entelecheia in defining change, which has made an interpretation in terms of actuality seem inevitable, and thus constitutes a serious challenge for my interpretation. For although Aristotle uses entelecheia interchangeably with energeia in this context, there is no independent reason to think that entelecheia can mean ‘activity’. I argue that while reading entelecheia as ‘activity’ is problematic, it is also problematic to read energeia as ‘actuality’ in this context. And given the philosophical advantages of my proposal, we are entitled to focus primarily on the term energeia, a standard connotation of which is activity.

. Change as actuality (a) What is an actuality? In addressing the question whether change is defined as an actuality, we immediately run up against the fact that different commentators use the term ‘actuality’ in somewhat different ways. Waterlow, for example, construes entelecheia and energeia as indicating that change (or the changing subject) is something ‘real, as real as anything else actual is real’, as opposed to something of suspect ontological status. Such glosses, however, might be mis The phrase is from Waterlow, Nature, . See also Johansen, Sense-Organs, –; Burnyeat, ‘De Anima II ’, , and ‘Kinēsis vs. Energeia: A Much-Read Passage in (but not of) Aristotle’s Metaphysics’ [‘Passage’], Oxford Studies in Ancient Philosophy,  (), – at  and ; Coope, ‘Change’, –. Note that Burnyeat reads the τις in ἐνέργειά τις, both at Physics . , b, and in the reference to that passage at DA . , a, as having an ‘alienans function’, indicating that from change we ‘cannot expect everything you would normally expect from an

Change in Aristotle’s Physics 



leading. They might suggest that Aristotle’s terms mean ‘actuality’ in a contemporary connotation, indicating that change ‘actually’ exists in the sense that it is not merely possible, or perhaps in the sense that it is not a fictional or merely imagined entity. Even though change is something ‘actual’ in both of these senses, neither of them captures the meaning of entelecheia and energeia, or the significance of Aristotle’s definition. Perhaps the strongest motivation for reading the definition in terms of actuality is Aristotle’s use of the term entelecheia in defining change, and in fact ‘actuality’ is often employed simply as a stand-in for that term. Whatever its obscurities, the term entelecheia functions primarily as a counterpart to dunamis (potential, capacity), especially in their use as adverbial datives modifying a form of the verb ‘to be’, in phrases such as ‘actual [entelecheiāi] being’ and ‘potential [dunamei] being’. In such constructions entelecheia and dunamis specify a way or mode of being some kind of thing. Something can be a tree, for example, either entelecheiāi (‘actually’) or dunamei (potentially). Entelecheia is connected etymologically to the notion of a goal or end (telos), and perhaps to being complete (entelēs). ‘Actuality’ in this sense thus involves being at a telos, or being a complete and full-fledged being of some kind, and stands in opposition to potential being, for the particular kind of being at issue. We may thus ask: what kind of being is it, such that change is, according to the definition, actually being that? To be sure, not all scholars have explicitly acknowledged the need to answer this energeia’ (‘Passage’, ), so that he might opt for a weaker formulation than Waterlow’s ‘as real as anything else actual is real’.  Against the first suggestion, Aristotle never uses the terms ἐνέργεια and ἐντελέχεια in discussions of possibility and necessity, and he claims that the terms δυνατόν and ἀδύνατον, when they are so used, mean something different from their meaning when said in accordance with having or lacking a δύναµις (Metaph. ∆ , esp. b– a), the correlate to ἐνέργεια and ἐντελέχεια. As for the second, δύναµις and potential being are ‘actual’ in the sense that they ‘in fact’ exist; they are not fictional or merely imagined. Of course, a potential house is not in fact a house, but potential houses in fact exist. One reason that Aristotle’s opponents in Physics .  were unable to see their way out of the dilemma is that they were not in a position to consider the possibility that potential being ‘really’ exists in this sense. In Metaphysics Θ  Aristotle argues that there is unexercised potential being. On the relation between contemporary notions of actuality and Aristotle’s notions of ἐντελέχεια and ἐνέργεια, I am indebted to the discussion in J. Beere, Doing and Being: An Interpretation of Aristotle’s Metaphysics Theta [Doing and Being] (Oxford, ), –, and I am grateful to him for sharing earlier versions of the manuscript.



Andreas Anagnostopoulos 

question. This, I think, is a mistake, one I shall say more about shortly. I begin, however, by considering a position that both acknowledges and answers this question. (b) Constitutive actuality Clearly, change is not the final actuality corresponding to the product; that is, change is not actually being the product (or that in virtue of which something is actually the product). Change is therefore best thought of as a kind of intermediate actuality. Thus one core idea in Kosman’s enormously influential account is that the changing subject is an actual or complete being, but the kind of being it is actually is potential being, the very kind of being specified in the qualification ‘qua such’. The process of becoming a house, for example, is the ‘constitutive actuality’ of a potential house. This means that the bricks, while being built and only then, are an actual potential house. This process of being built is the potential house’s actuality, in the sense of the potential house’s actually being a potential house (or perhaps: that in virtue of which the bricks are an actual potential house). The house (or its form), by contrast, is the ‘deprivative’ actuality of the potential house; it is the actual being (or its form) at which the potential house is directed, and which it is potentially. The deprivative actuality of the potential house is not an actual potential house but an actual house. The term ‘deprivative actuality’ reflects Kosman’s conviction  Coope describes change as a potential’s ‘being actual’ and its ‘being most fully actual as a potential’ (‘Change’, –), suggesting the ‘constitutive actuality’ reading. Gill and Hussey have not, as far as I can tell, clearly acknowledged or answered this question. But while Gill’s account seems to be consistent with holding that change is the ‘constitutive actuality’ of potential being, Hussey at times appears to specify a quite different notion of being the actuality of something; see n.  below. According to Waterlow, change is an ‘active expression’ and ‘manifestation’ of its subject’s nature and is what the subject does ‘on account of’ its nature as potential being (Nature, –).  Whether Kosman consistently holds this view is less clear, and so I hesitate to describe it simply as his position. For as we shall see below, several of the texts he relies on involve actualities that, though they preserve the corresponding potential beings, should not, I think, be identified with being those potential beings in a certain way. When I speak of ‘constitutive actuality’ I shall mean the concept that is described above.  Kostman, ‘Definition’, , helpfully puts the ‘constitutive actuality’ proposal in terms of the iterability of ‘actual’ and ‘potential’.  Kosman tends to describe this distinction as one between kinds of actuality rather than one between ways in which an actuality is related to that of which it is the actuality. I think the distinction is of the latter kind. Moreover, it can be understood

Change in Aristotle’s Physics 



that the potentiality involved in a change is extinguished once the change has run its course. But other aspects of Kosman’s account are sufficient to distinguish the notion of ‘constitutive actuality’ without any particular commitments as to the persistence of potential being into the product. Such neutrality—which I also hope to maintain in my own account—is preferable given the controversy surrounding the difficult question of persistence. Although Kosman sometimes emphasizes that the potential being persists (as potential) into the change, or that it is most manifest as a potential being during the change, these claims alone do not warrant thinking of change as a ‘constitutive actuality’ of potential being. What is required is a difference between the state of the bricks when ‘dormant’ and their state when changing, and that this difference concern the degree to which they are a potential house: while in the process of being built, the bricks are a potential house in a ‘complete’ or ‘full-fledged’ manner, i.e. ‘actually’. This may cause concern at the outset. It is not clear that there is a genuine state of actually being a potential being at all. I shall develop this concern to a significant degree as follows: ‘Actuality of X’ may signify either the actuality at which X is directed (deprivative) or that which constitutes something’s being an actual X (constitutive).  This idea, as well as the correlative idea that the qua-phrase distinguishes between the two types of actuality on this basis, is also found in Themistius, In Phys. .  ff. Schenkl.  If, as I have suggested, the notion of ‘constitutive actuality’ can be made out independently of the extinction thesis, then the core of Kosman’s interpretation can survive the rejection of the extinction thesis. Robert Heinaman, ‘Kosman’, points to a number of texts that appear to contradict the extinction thesis. If the matter for a change should be identified with potential being, we may add to Heinaman’s catalogue those texts in which Aristotle characterizes matter as persisting into the product of change and, in the case of substances, as a subject of which form is predicated. Of course, such texts will hardly be decisive in the light of the fact that Physics .  describes the potential and actual being as exclusive: ‘in some cases the same things are both potentially and actually, however not simultaneously or not according to the same thing, but, for example, hot actually and cold potentially’ (a–).  To see what is at stake, consider a teenager who claims to be an adult because he is a ‘full-grown child’. In this kind of case we might respond by pointing out that one cannot claim to be an adult just by being at the final stage of any (perhaps arbitrarily chosen) period. Adult humans and adult frogs are full-grown human beings and frogs respectively. The concern is that it is inappropriate to speak of a ‘full-grown child’ because there is no genuine stage of being full-grown for a child. Kosman asks us to understand ‘actuality of a potential being’ as we might understand ‘perfection of a stutter’ to refer to the cultivated stutter one develops, for example, for the role of a stuttering character in a play. The cultivated stutter can be called a genuine



Andreas Anagnostopoulos

in the course of this section. The idea that only the changing subject is an ‘actual potential being’ implies that when dormant, that subject is merely potential in a corresponding way, that it is a merely ‘potential potential being’, so to speak. So thinking of change as the ‘constitutive actuality’ of a potential being requires adding not only an intermediate actuality, but also a corresponding potentiality, to the picture we might otherwise have of change. We would expect to find some indication that change is a ‘constitutive actuality’ of potential being in Metaphysics Θ , which contains Aristotle’s only explicit account of ‘when each thing is potentially [dunamei] and when not’ (b). Indeed, Aristotle is there precisely concerned to limit the application of the term ‘potential being’. Yet he never broaches the idea that only the changing subject is fully (‘actually’) a potential being. Instead, as Charles points out (Action, – n. ), Aristotle envisages the label ‘potential F ’ as applying to a ‘dormant’ object, one that is not already in process of becoming F. Moreover, if only an object in the process of becoming F were fully a ‘potential F ’, we would expect Aristotle, in Metaphysics Θ , to concede more to the Megarian position that ‘only when something is active [energēi] is it capable [dunasthai], and when it is not active, it is not capable’ (b–). Kosman appeals to the descriptions of end-containing activities at Metaphysics Θ , b–, and of realizations of dispositions (hexeis) in De anima . , in order to provide evidence for, or illumination of, the idea that change is a ‘constitutive actuality’ of potential being. As for the former passage, what it tells us explicitly about change is that change is exclusive of its end. This would be at best misleading if there were, in addition to the potentiality for ‘perfection’ in a clear enough sense; it involves the full possession of the ability, for example, to stutter more often, audibly, naturally, persuasively, etc. than before. However, it is telling that, as Beere, Doing and Being, – n. , points out, cultivating a stutter in this way does not in general culminate in being rid of the stutter, the other sort of ‘perfection of a stutter’ Kosman asks us to imagine. Indeed, one would ordinarily hesitate to cultivate an unwanted stutter.  Kosman, ‘Motion’, relies only on the De anima .  passage, while in ‘Substance, Being, and Energeia’ [‘Substance’], Oxford Studies in Ancient Philosophy,  (), –, he relies on both texts. Hussey, III & IV, –, relies on the De anima passage, though it is not clear that he does so in support of the ‘constitutive actuality’ proposal. Gill, Paradox, –, appeals to the Metaphysics Θ  passage in order to substantiate an interpretation she claims to find in Kosman, ‘Motion’, but this is an interpretation in terms of ‘activity’, one which ‘avoids circularity because Aristotle carefully distinguishes activities from changes’ and yet ‘captures the dynamic quality of change’ ().

Change in Aristotle’s Physics 



the product, also a potentiality directed at the change itself (e.g. a ‘dormant’ potential house’s potentiality for being an ‘actual’ potential house). For in this case the change would contain this telos; indeed it would be this telos. Thus the characterization of change here, by its silence, tells against the idea that change is a ‘constitutive actuality’ of potential being. The passage from De anima .  seems to give us more. It employs a model on which there are intermediate levels of actuality that are at the same time levels of potentiality. For example, the person who possesses, but is not currently using, knowledge qualifies as an actual knower by actually possessing a disposition (hexis), which is itself a kind of capacity for further exercise. But it is not, at least not primarily, to the possession of a hexis that change is compared, but to the realization of a hexis. For as Kosman claims, the De anima .  passage implies that realizations of hexeis preserve, rather than exclude, their potentialities. (In describing end-containing activity as an ‘active manifesting of a potentiality’ (‘Substance’, ), Kosman presumably means to imply that such activity too preserves the corresponding potentiality.) Thus realizations of hexeis appear to have the preservative structure claimed for change. However, the preservative structure of hexis-realizations implied by the De anima passage is in fact different from that ascribed to change. To say that change is the ‘constitutive actuality’ of potential being is not to say  Kosman, while drawing on the idea that changes (as he sees them), like ἕξεις, combine actuality and potentiality, is cautious not to push this parallel too far; see ‘Motion’, .  I say ‘implies’ because I believe, with Kosman, ‘Motion’, , and Burnyeat, ‘De anima II ’, –, that what this passage describes as a ‘preservation’ and ‘development into itself and ἐντελέχεια’ (b–) is not the exercise of a ἕξις (this exercise is there called an ἐνέργεια), such as continuous contemplation of the Pythagorean theorem, but rather the transition from not exercising to exercising, e.g. from merely knowing the Pythagorean theorem to contemplating the theorem (this transition is not called an ἐνέργεια). However, see R. Heinaman, ‘Activity, Change and De anima II. ’, Phronesis,  (), –, esp. –, for criticism of this reading. If we read the passage instead so that it is describing the relevant ἐνέργεια (e.g. contemplation) as a ‘preservation’ and ‘development into itself and ἐντελέχεια’ (b–), and so a change only in a specialized sense, if at all, it will then be explicitly contrasting change with the kind of ἐνέργεια that is thought ultimately to characterize change, telling against an identification of change with this type of ἐνέργεια on the basis of preservative structure.  In fact, Kosman seems to treat both realizations of ἕξεις and end-containing activities as ‘constitutive actualities’ of potential beings. See ‘Motion’, , and ‘Substance’, –, though the latter does not employ the language of ‘constitutive actuality’.



Andreas Anagnostopoulos

that the changing subject is both potentially and actually the same kind of being. It is rather to say, for example, that the changing subject is actually a potential house and potentially a house. But the De anima passage tells us that someone actively contemplating—an actual contemplator—is still in possession of his knowledge, and so qualifies as a potential contemplator. The person actively contemplating would thereby qualify as an ‘actual potential contemplator’—thus securing the desired parallel in preservative structure with change—if contemplating were the ‘constitutive actuality’ of the potential contemplator. For then contemplating would just amount to actually being a potential contemplator. However, Aristotle nowhere indicates that realizations of hexeis or end-containing activities are ‘constitutive actualities’ of the corresponding potential beings. For example, he does not claim that actively seeing is a way or mode of being a potential seer, namely, being a potential seer—having the power of sight—actually. Actively seeing is rather a matter of actually seeing, of being an actual seer. This is not to deny that seeing is an exercise of one’s power of sight, one that preserves and even expresses or manifests one’s power of sight, one’s being a potential seer. But this falls far short of the claim that seeing is or even involves being a potential seer at a higher degree of actuality. The fact that change too preserves and manifests the corresponding potential being simply does not imply that change is a ‘constitutive actuality’ of potential being. (c) Actuality in another sense? The foregoing might be taken to show that we should discard the notion of ‘constitutive actuality’ as too restrictive, while maintain One of Kosman’s favoured terms, ‘manifestation’, betrays a slide between these two claims, and between the corresponding notions of mere exercise and ‘constitutive actuality’. The manifesting or manifestation of a potentiality might be either (i) the expression of that potentiality in something else, for example, in the corresponding exercise, or (ii) ‘a potentiality [itself] in its full manifestation’ (‘Motion’, ). I suggested earlier that the ‘constitutive’ vs. ‘deprivative’ distinction can be drawn independently of the extinction thesis about potential being. But the extinction thesis may be required to distinguish between a potential being’s preservative exercise (what I have argued is not in general the ‘constitutive actuality’ of the potential being) and the product of that exercise, so long as both are thought of as actualities of it.  It may rather tell against the ‘constitutive actuality’ proposal. For an explanation is needed of why the realizations of hexeis, if structurally similar to changes, are not described as constitutive actualities.

Change in Aristotle’s Physics 



ing that change is an actuality of the potential being in whatever sense can be gleaned from Aristotle’s descriptions of end-containing activities and hexis-realizations in these two passages. These sorts of activity exhibit what at least some scholars appear to have in mind in thinking of change as an actuality of potential being. In looking to these activities for guidance, one must of course be careful not to draw on those of their attributes that are denied to changes. However, these passages give no indication that the features shared by changes and the other sorts of activity can elucidate or legitimate change’s alleged status as an actuality. Note first that Metaphysics Θ , b–, need not, and I think should not, be read in terms of actuality at all, since every energeia that Aristotle mentions there is an activity, and the passage explicitly aims to classify actions (praxeis, b). More importantly, the one hint at the concept of actuality, namely the notion of completeness, is a point of contrast between change, which is exclusive of its end and so ‘incomplete’, and end-containing activity, which is thereby ‘complete’. De anima . , on the other hand, explicitly invokes the concept of entelecheia. Here Aristotle tells us that contemplating is an entelecheia (b, ) and that the person contemplating a being entelecheiāi (a, b). And it is likely that he is telling us that the  Hussey, III & IV, is not careful enough. To secure the parallel with ἕξιςrealization, Hussey thinks, we must ‘take something’s admitting of a particular kind of change as a disposition [hexis] to change in that way, which will be exercised in appropriate circumstances. The exercise of the disposition will then be the changing, not the having changed; and change will then appear, as required, as an actuality, corresponding to the potentiality as [exercising the disposition] to [having the disposition]’ (). Apparently, he takes ἕξις-realization to be an actuality in the following sense: it is that at which the corresponding δύναµις is ultimately directed, what the potentiality is ultimately a potentiality for. And so he suggests construing the definition of change in terms of, for example, a ‘potentiality to break’ rather than a ‘potentiality to be broken’ (). But change surely cannot be like the realization of a ἕξις in this respect. He writes that ‘the only objection to this reading is that the disposition involved is specified in such a way as to obscure the issue’ in so far as Aristotle ‘misleadingly’ describes the former as the latter, e.g. the ‘potentiality to break’ as a ‘potentiality to be broken’ (–). But the position that he takes Aristotle to ‘misleadingly’ suggest (i) makes his analysis of the qua-phrase superfluous, (ii) makes Aristotle’s definition vulnerable to the most straightforward charge of circularity, and (iii) is inconsistent with Hussey’s analysis elsewhere: ‘[“potentially being”] can only be expanded to “potentially being F ” where “F” gives the endstate. And the fact that this is the only plausible reading casts doubt on the reading of “changeable” as “potentially in process of change” . . . so we must understand “changeable” . . . on the model of “potentially having changed”’ (). Some of these concerns about Hussey’s analysis are raised by M. L. Gill, ‘Review of E. Hussey, Aristotle’s Physics, Books III and IV’, Philosophical Review,  (), –.



Andreas Anagnostopoulos

person who possesses knowledge is a being entelecheiāi at b. If he does not explicitly call the possession of knowledge entelecheia in DA . , this is surely only because it is taken for granted. But Aristotle does not characterize the acquisition of knowledge as entelecheia, or the person in the process of acquiring knowledge as a being entelecheiāi. If activities that are neither realizations of hexeis, nor the activities of actual beings, nor yet end-containing activities, can still qualify as actualities, there is no hint of such a view in these texts. And indeed, once we leave aside these features, we are left with the idea that changes, like these other activities, are exercises that preserve and perhaps even express or manifest the corresponding potentialities. But the notion of a preservative exercise is not a notion of actuality, so that these texts do not elucidate any sense in which change is an actuality. The idea that change is a kind of exercise of potential being, I shall argue, comes very close to capturing the force of Aristotle’s definition. But it would be gratuitous to claim, on this basis, that Aristotle defines change as an actuality. Moreover, we shall see that such a claim adds little, if anything, to the philosophical significance of Aristotle’s definition. Although I cannot consider every possible proposal for how to think of change as an actuality, the foregoing discussion reveals a risk for the actuality interpreter who leaves aside the ‘constitutive actuality’ interpretation, namely the risk of relying on a notion other than that of actuality. But this is not the only risk. For Aristotle defines change not just as some entelecheia, but more specifically as the entelecheia of potential being. Thus, quite generally, change’s alleged status as an actuality must correspond tightly enough to potential being for change to be defined as its actuality. This point will be strengthened and made more precise in the analysis of the phrase ‘qua such’, starting in the next section. While the ‘constitutive actuality’ proposal does justice to this tight connection between change and potential being, by identifying change as actually being potential being, it is not clear how a different kind of ‘actuality’ interpretation might do so. For example, one might point out that change  It is helpful to compare two other kinds of case in which such locutions could draw on intrinsic connections between the actuality and what it is the actuality of, even though we are not talking about ‘constitutive actuality’. We might think of the house as the actuality of the potential house or of contemplation as the actuality of what is potentially contemplating. In both cases, we might draw on the fact that the actuality involves the very same kind of being that the potential item enjoys, but at a higher degree of actuality. In the one case the actuality enjoys that kind of being

Change in Aristotle’s Physics 



is something real that exists while the potential being exists—even because the potential being exists—or that change is the counterpart, on the side of actuality, of potentially changing. But neither of these claims would warrant thinking of change as potential being’s actuality. Moreover, the sense in which change is precisely potential being’s actuality must be robust enough to distinguish the change from the product. The product, while presumably an actuality, should not qualify as an actuality of potential being. For, as I shall argue in the next section, if the product is an actuality of potential being, then the definition of change (construed in terms of actuality) will not have the resources to exclude the product from its scope.

. ‘Qua such’ The ‘product puzzle’, recall, begins from the thought that the products of change are entelecheiai and energeiai, perhaps paradigmatic ones, of potential beings, so that the definition threatens to pick them out in addition to, or even instead of, changes themselves. The consensus interpretation faces this threat acutely (by reading the Greek terms as ‘actuality’) but boasts of avoiding it, thanks to its analysis of the phrase ‘qua such’: while the products of change are actualities of potential beings, only changes are actualities of potential beings qua potential. Notably, Kosman relies on the qua(the house is actually a house) while in the other the actuality is that kind of being (contemplating just is actually contemplating).  This is part and parcel of the view that change is a kind of categorial being. See Simpl. In Phys. .  ff. Diels. I cannot fully address this idea here, though I think there are very good reasons for thinking that Aristotle is committed to placing change in a category. However, the Categories does not do so (, a ff.) and I doubt that Aristotle has this idea in mind in Physics . –, not only for the reasons mentioned above. For this proposal is in tension with Aristotle’s claim that ‘these having been distinguished according to each kind [γένος] [of being], [being] actually and [being] potentially, the ἐντελέχεια of the potential being, qua such, is change’ (a–). If he were here thinking of change as a kind of categorial being, it would follow immediately that there is change of change, and so on ad infinitum. Aristotle is here clearly thinking of change as occurring within various categories, in the sense that it is the transition from potential to actual being in each of those categories.  We have seen how problematic dealing with these challenges might be in Hussey’s reasoning (n.  above). He seems to account for the idea that the change alone is the potential being’s actuality by claiming that the relevant potentiality is directed ultimately at it rather than at the product (III & IV, –).  So Kosman: ‘We may now say: the phrase “as such” signals that it is the con-



Andreas Anagnostopoulos

phrase not only to exclude the products from the scope of the definition, but also to include changes, since he takes changes to be actualities of potential beings only in a specialized sense. In this section I shall for convenience employ the term entelecheia predominantly, although much of what I say here, as in the last section, will have consequences for the relative importance of the two terms. I begin with a grammatical point about the qua-phrase, one which creates an initial challenge for the consensus reading. An analysis of Aristotle’s own explanation of what he means by qua (a– b ) will show that the role of the qua-phrase is to specify the proper rather than accidental entelecheia of a potential being, and it will also strengthen the initial challenge. Finally, I argue that the main passage cited in favour of the consensus interpretation of the quaphrase (b–) lends no support to that interpretation and is consistent with mine. My rejection of the consensus reading of the qua-phrase, as well as the account of it that I begin in this section, will be crucial to understanding other aspects of Aristotle’s definition, notably its dialectical significance. (a) Process, product, and the grammar of the definition Let us begin by considering the use of a qua-phrase in the relatively simple proposition that the doctor builds ‘qua builder’ (Phys. . , b–). What exactly does the phrase ‘qua builder’ characterize as a builder? It is helpful to expand the claim as it is sometimes translated: The doctor builds in so far as he is a builder. stitutive and not the deprivative actuality which is referred to in Aristotle’s definition. . . . to speak of the actuality of a potentiality qua potentiality is to signal that the actuality is constitutive and not deprivative’ (‘Motion’, ). Hussey is less clear: ‘the general purpose of the qua-clause is of course to pick out a certain kind of actuality corresponding to a certain kind of potentiality’ (III & IV, ). This makes it sound as if the qua-phrase distinguishes not different kinds of actuality that a single potentiality might have, but rather different potentiality–actuality pairs. He also claims, however, that the qua-phrase is ‘attached’ () to ἐντελέχεια in the definition, suggesting that there is a single potentiality in play. This account of the grammar of the qua-phrase (sect. (a) below) supports his claim that his account ‘largely coincides’ () with that of Kosman. More recently, Coope writes that the qua-phrase ‘distinguish[es] between the process of change and the product of change’, which she sees as two different actualities of, for example, the bronze’s potential to be a statue (‘Change’, ). Even some who read the definition in terms of ‘actualization’ take the qua-phrase to exclude the products of change; see Penner, ‘Verbs’, , and Kostman, ‘Definition’, –.

Change in Aristotle’s Physics 



Clearly, ‘he’ refers to the doctor; the doctor’s being a builder enables him to build. We may thus call the doctor the ‘subject’ of the qua-phrase. Here is Aristotle’s first statement of the definition of change, expanded in the same way: The entelecheia of the potential being in so far as it is such [i.e. potential]. What is the referent of ‘it’, which is characterized as ‘such’, i.e. ‘potential’, or, in later formulations, as ‘changeable’, ‘alterable’, etc.? In other words, what is the ‘subject’ of the qua-phrase? There are two basic options: (a) The entelecheia of a potential being in so far as the entelecheia is such [potential]. (b) The entelecheia of a potential being in so far as the potential being is such [potential]. Reading (a) is amenable to the consensus interpretation, since it suggests a specialized concept of entelecheia—an ‘entelecheia qua potential’—that might apply to changes and only to changes. However, entelecheia cannot be the ‘subject’ of the qua-phrase. For  What I call the ‘subject’ of the qua-phrase need not be the grammatical subject of the sentence. This is evident in a claim such as ‘The doctor cures the sick qua sick.’  Kostman, ‘Definition’, , and Penner, ‘Verbs’, –, offer temporal readings, for example: (c) The entelecheia of the potential being while it [the potential being] is still potential. The idea is that while both the change and the product are ἐντελέχειαι of the potential being, only the former exists while the potential being exists (as potential), so that the qua-phrase, on this reading, excludes products. As I have stated it, this reading treats ‘potential being’ as the subject of the qua-phrase. But note that understanding the qua-phrase in this way seems to require a somewhat more complex rendering of the definition. ‘While it is still potential’ is not, I take it, meant simply to carve out one of two time-segments of the potential being, such that the ἐντελέχεια of one segment is the process, and that of the other segment is the product (in which case it would not be helpfully illustrated by a case in which the potential being (‘the buildable’) does not persist into the product). Rather, what is meant is something like ‘that ἐντελέχεια of the potential being, which [ἐντελέχεια] exists while it [the potential being] is still potential’. This seems to read too much into the few words of Aristotle’s definition. Moreover, it is not merely ‘unnatural’, as Penner puts it, for the qua-phrase to ‘deputize for “while”’ (‘Verbs’, ). Rather, qua (ᾗ) does not have a temporal meaning, and what is uncontroversially Aristotle’s explanation of the meaning of qua employs an example in which such temporal distinctions cannot be drawn (Phys. . , a– b ; see sect. (b) below).  Hussey alone explicitly endorses this position, claiming that the qua-phrase



Andreas Anagnostopoulos

the entelecheia (the change) could not plausibly be characterized as a potential being. Nor, in accordance with various versions of the qua-phrase that Aristotle later offers, could it be characterized as ‘changeable’ (a), as ‘potentially something’ (a), or as ‘capable’ (dunaton) (b). Kosman might respond that the entelecheia is just the potential being’s being most actual, its being a potential being to the fullest extent, so that the actual being in question is in fact a potential being (changeable etc.). Still, on his view, it is not the entelecheia itself (i.e. the change) that is most fully a potential being (changeable etc.), but rather the changing subject. What then of reading (b)? On this reading, the phrase ‘qua such’ characterizes the potential being, where ‘such’ (potential) specifies an aspect or feature, broadly speaking, of the potential being: its being a potential being. Even at this linguistic level, we can appreciate ‘must be attached and rephrased as follows: “The actuality-qua-potentially-being of that which potentially is”’ (III & IV, ). Kosman, in a later essay, distinguishes the ‘actuality of a potential qua potential and the actuality of that potential qua end other than itself to which it is directed’ (‘Substance’, ), and here we may wonder what is being characterized as an ‘end other than itself to which it is directed’, if not the ἐντελέχεια itself. Perhaps Kosman means ‘qua directed at such an end’, but this seems no different from ‘qua potential’. Although the idea that there are two sorts of actuality of a single potential being naturally suggests reading (a), Kosman and Coope, as I understand them, intend to work out this idea along the lines of reading (b).  Nor, drawing on Aristotle’s later specifications of particular species of change, is an alteration itself ‘alterable’ (a), a growth or decrease itself ‘subject to growth or decrease’ (a–), a generation or passing away ‘generable or subject to passing away’ (a–), or a locomotion ‘capable of locomotion’ (a). Finally, the building process is not itself ‘buildable’ (a, b). In support of reading (a) Hussey appeals to a–, which he translates: ‘The actuality, then, of what is potentially, when, being in actuality, it is operating, not qua itself but qua changeable—is change.’ Here, he claims, ‘the qua-phrase is clearly attached to the verb “is operating” [ἐνεργῇ], corresponding to “actuality”’ (III & IV, ). If Hussey means that the subject of the phrase ‘qua changeable’ is the ‘operating’, then he is clearly mistaken. For the operating is not itself changeable.  This issue is perhaps obscured by a free substitution of ‘potential being’ (δυνάµει ὄν) and ‘potentiality’ or ‘potential’ (δύναµις). Thus Kosman sees Aristotle as claiming that change ‘is not the actuality of a potentiality in the sense of the actuality that results from a potentiality, but rather in the sense of an actuality which is a potentiality in its full manifestation’ (‘Motion’, ). Here he claims that the ἐντελέχεια is a potentiality—presumably a δύναµις—in a certain condition. I find this suggestion difficult to understand. But in any case, it is of no help with the present concern. For the problem is that if ἐντελέχεια is the subject of the qua-phrase, then it will be characterized as a potential being (δυνάµει ὄν), as ‘changeable’, as ‘capable’, and so on, not that it will be characterized as a δύναµις.

Change in Aristotle’s Physics 



an intuitive challenge for the view that the qua-phrase is responsible for excluding the products of change: if the qualification ‘qua potential’ serves to exclude the product, then some other qualification, or perhaps a lack of qualification altogether, should give us the product. Qua what, we may then ask, does the potential house have the actual house as its entelecheia? Not ‘qua potential house’, which should give us only the change. Nor ‘qua bricks’, the only kind of alternative Aristotle explicitly considers (a–); for this (unlike ‘qua potential house’) does not indicate any directedness towards the product. Nor can we say ‘qua actual house’, since this could not characterize the potential house, as reading (b) requires. Perhaps not qua anything? But then it seems that we are back to the aspect of the potential house that the qua-phrase specifies: its being a potential house. (b) Proper and accidental Let us step back from the question of whether the qua-phrase might exclude the products of change from the scope of the definition and look at Aristotle’s own explanation of his use of ‘qua’. His explanation involves showing how the addition of the phrase excludes a different entelecheia, in this case, ‘the entelecheia of the bronze, qua bronze’: λέγω δὲ τὸ ᾗ ὡδί. ἔστι γὰρ ὁ χαλκὸς δυνάµει ἀνδριάς, ἀλλ᾿ ὅµως οὐχ ἡ τοῦ χαλκοῦ ἐντελέχεια, ᾗ χαλκός, κίνησίς ἐστιν· οὐ γὰρ τὸ αὐτὸ τὸ χαλκῷ εἶναι καὶ δυνάµει τινί [κινητῷ], ἐπεὶ εἰ ταὐτὸν ἦν ἁπλῶς καὶ κατὰ τὸν λόγον, ἦν ἂν ἡ τοῦ χαλκοῦ, ᾗ χαλκός, ἐντελέχεια κίνησις· οὐκ ἔστιν δὲ ταὐτόν, ὡς εἴρηται. (a–) By ‘qua’ I mean this: although the bronze is potentially a statue, still, change is not the entelecheia of bronze qua bronze. For it is not the same to be bronze and to be potentially something since, if it were the same without qualification and according to definition, then the entelecheia of the bronze, qua bronze, would be change. But it is not the same, as has been said.

As I claimed above, and as reading (b) requires, the qua-phrase 

On this point see the discussion of b– in sect. (d) below. Kosman, like others, suggests that without the qua-phrase the definition would naturally be taken to refer to the product. But see n.  above. Gill suggests that ‘qua potentially a house’ yields the house, while ‘qua potential’ yields the change (‘Causal Action’, ). As far as I can tell, Gill is not drawing on the fact that ‘qua potentially a house’ is more determinate than ‘qua potential’. Indeed, it is hard to see how such a difference would discriminate process from product. But I do not see what other relevant difference she intends. 



Andreas Anagnostopoulos

refers to an aspect or feature of the subject that is to undergo change—its being a potential being. We may now say more precisely that it isolates one among a plurality of merely coinciding beings—beings that are merely one-in-number. The potential statue and the bronze are different in being: ‘it is not the same thing to be bronze and to be potentially something’ (a–). On the other hand, we may presume that in this case the bronze and the potential statue are one-in-number. Aristotle is more explicit in his illustrative example: δῆλον δ᾽ ἐπὶ τῶν ἐναντίων· τὸ µὲν γὰρ δύνασθαι ὑγιαίνειν καὶ δύνασθαι κάµνειν ἕτερον—καὶ γὰρ ἂν τὸ κάµνειν καὶ τὸ ὑγιαίνειν ταὐτὸν ἦν—τὸ δὲ ὑποκείµενον καὶ τὸ ὑγιαῖνον καὶ τὸ νοσοῦν, εἴθ᾿ ὑγρότης εἴθ᾿ αἷµα, ταὐτὸν καὶ ἕν. (a–b) It is clear in the case of opposites. To be able to be healthy and to be able to be sick are different, otherwise to be sick and to be healthy would be the same. But what underlies, i.e. what is being healthy and what is being sick, whether moisture or blood, is the same and one.

In this example it is clear that ‘what is able to be healthy’ is one-in-number with ‘what is able to be sick’, because it is the same underlying thing, whether moisture or blood, that is sometimes healthy and sometimes sick. So, the potential statue and the bronze—the beings, one of which, at the most basic level, the qua-phrase specifies—are merely one-in-number. Aristotle tells us that it is because the potential statue and the bronze are merely one-in-number that the addition of the phrase ‘qua potential statue’ can single out the change (becoming a statue). That is, it is because the bronze and the potential statue are merely one-in-number that ‘the entelecheia of the bronze qua bronze’ is distinct from the entelecheia of the potential statue qua potential statue (i.e. the change). On the other hand, if the potential statue and the bronze were not merely one-in-number but one-in-being, then the addition of the qua-phrase could not discriminate between the two entelecheiai. In order to understand these claims, it is helpful once again to consider the simpler example of the doctor. The doctor cures ‘qua doctor’, but builds a house ‘not qua doctor but qua builder’ (Phys. . , b–). Aristotle implies that the doctor builds only ‘according to what coincides’ (b–). This means that it is in virtue of being a doctor that the doctor can be said to cure, but it is in vir-

Change in Aristotle’s Physics 



tue of being one-in-number with something else (what coincides), specifically a builder, that the doctor can be said to build. Applying this distinction to the case at hand, since it is one thing to be a potential statue, and another to be bronze, a single item that is both bronze and a potential statue will have one thing as its entelecheia in virtue of being a potential statue (i.e. qua potential statue), and another thing as its entelecheia in virtue of being bronze (i.e. qua bronze). Still, even the latter entelecheia is in a looser sense an entelecheia of the potential statue; it qualifies as an entelecheia of the potential statue in virtue of the fact that the potential statue is also something else, bronze. But if being bronze and being a potential statue were the same (if the potential statue and the bronze were ‘the same without qualification and according to definition’, a–), then the entelecheia that belongs to something in virtue of its being bronze would also belong to it in virtue of its being a potential statue. That entelecheia too would then qualify as a change according to Aristotle’s definition. While both curing and building could be denoted by the description ‘activity of the doctor’ taken loosely, adding the phrase ‘qua doctor’ will narrow the extension of the description to just one of these, namely curing. But it will narrow the extension in this way only because, in a stricter sense, curing is the only activity of the doctor. It is the only activity that belongs to the doctor in virtue  Here I am assuming that the ἐντελέχεια of X qua X, given that X is also Y, is also the ἐντελέχεια of Y qua X. Although Aristotle in the current passage uses only the former sort of description, it is clear from the fact that he takes himself to be elucidating the function of the qua-phrase that he intends his remarks to apply to the latter sort of description. That is, he is explaining why adding ‘qua potential statue’ to ‘ἐντελέχεια of the potential statue’ yields a different extension from adding ‘qua bronze’, or even not adding anything at all.  Similarly, the fact that what is capable of health and what is capable of sickness are merely one-in-number allows that being healthy and being sick are different (‘otherwise to be sick and to be healthy would be the same’, b–). Presumably we are to think of being healthy and being sick as the ἐντελέχειαι of what is able to be healthy qua able to be healthy, and of what is able to be sick qua able to be sick, respectively. If what is able to be healthy and what is able to be sick were the same not only in number but also in being, we could not distinguish the two ἐντελέχειαι in this way: being sick, for example, would also be the ἐντελέχεια of what is able to be healthy that belongs to it in virtue of its being able to be healthy. We already know that this ἐντελέχεια is being healthy, so that ‘to be sick and to be healthy would be the same’. But since being able to be healthy and being able to be sick are distinct, the ἐντελέχεια something has in virtue of the former need not be an ἐντελέχεια it has in virtue of the latter.  Physics .  does not use such a phrase. But it does characterize all these activities as things the doctor does: ‘the doctor cures’ and ‘the doctor builds’ (b–).



Andreas Anagnostopoulos

of his being a doctor. We may thus call curing the proper activity of the doctor. The activities now excluded, such as building, are in this stricter sense activities of some other being with which the doctor coincides, and so activities of the doctor only ‘qua something else’. We may thus call them accidental activities of the doctor. Similarly, both being built and the entelecheia of the bricks qua bricks (whatever this might be) could be denoted by the description ‘entelecheia of the potential house’. The phrase ‘qua such [potential]’ will narrow the scope of the definition to one among what are, in this loose sense, several entelecheiai of the potential being. But it will do so only because being built (the change) is in a stricter sense the only entelecheia of the potential house; it is the only proper entelecheia of the potential house. The entelecheiai that the addition of the qua-phrase excludes are in this stricter sense entelecheiai of some being with which the potential being coincides; they are merely accidental entelecheiai of the potential being. In this way, the qua-phrase does not add anything that is not implicit in the strictest reading of the remainder of the definition. Nevertheless, adding ‘qua potential’ guards against a less strict understanding of the description ‘entelecheia of a potential being’, on which it has a broader extension. This accords precisely with the use of such phrases (as well as some alternatives) in Physics . . Aristotle’s use of the qua-phrase is neither superfluous nor obscure, but something for which we are well prepared. (c) ‘Qua such’ and the products of change We are now in a better position to understand why the qua-phrase cannot do what consensus interpreters ask of it. If the qua-phrase were to exclude the product of a change from the scope of the definition, the product would have to be a merely accidental entelecheia of the potential being. But this is implausible. The house, for example, seems to be a proper entelecheia of the potential house, if it is an entelecheia of the potential house at all; it is precisely because something is a potential house that, when actualized, it becomes a house.  Kosman claims that the ‘the most serious difficulty’ with what he calls the ‘actualization’ reading is that it makes the qua-phrase superfluous (‘Motion’, ). By ‘actualization of X’ he appears to mean ‘becoming an actual X’, so that it is not clear whether he thinks the same objection holds against my ‘activity’ reading. Hussey makes the stronger claim that, as he puts it, ‘attaching’ the qua-phrase to ‘potential being’ (which I take to be reading (b) above) ‘give[s] no promise of sense’ (III & IV, ).

Change in Aristotle’s Physics 



And it is not clear what other being, merely one-in-number with the potential house, has the finished house as its (proper) entelecheia. Neither the bricks (which, though one-in-number with the potential house, do not have the house as their proper entelecheia) nor the actual house (which is not one-in-number with the potential house) will do. And even consensus interpreters tend to suggest that the house is instead a proper actuality of the potential house. This is suggested by their engagement with the product puzzle and by the common idea that the product is the paradigmatic actuality of the potential being. But in this case the qua-phrase, properly interpreted, could not exclude the product from the scope of the definition. In particular, the qua-phrase cannot discriminate between two different concepts or kinds of (proper) entelecheia of something, as Kosman’s ‘constitutive’ and ‘deprivative’ actuality appear to be. And the qua-phrase certainly cannot add changes to the extension  It must be admitted that scholars are not always explicit on these issues. Coope, ‘Change’, takes Aristotle to exclude the entelecheia of bronze, qua bronze, by specifying that he means the entelecheia proper to the potential house, not bronze, in the way I have explained. But she then claims that this does not suffice to exclude the state of the finished product from the scope of the definition: ‘There are two ways in which the bronze’s potential to be a statue might, in this sense, be actual. One is for the bronze to be a statue, the other is for the bronze to be in the process of becoming a statue. For all that has been said so far, either of these could count as the actuality of the bronze’s potential to be a statue, qua potential’ (). This suggests that the product too is a proper actuality of the potential being. But she then relies on an additional function of the qua-phrase, ‘to emphasize that the actuality in question is the actuality of something insofar as it is merely potentially F ’, where ‘being potentially a house, in the sense [Aristotle] means here, is incompatible with actually being a house’ so that the potential house, i.e. the buildable, could not ‘be actual’ while still buildable (). This suggests that the product is not an actuality of the potential being at all, since the product is no longer a potential being.  Kosman’s remarks in ‘Motion’ about the example of the stutter do not come down clearly on the question whether we should not understand ‘constitutive’ and ‘deprivative’ as specifying two different actualities of the same entity. He suggests both (i) that there are two kinds of perfection of something, deprivative (being rid of something) and constitutive (something’s flourishing), which can be applied to the stutter. But he also claims that (ii) the two perfections are of the stutter ‘qua stutter’, and of the stutter ‘qua speech’ (or ‘as the privation from which’, ) respectively, and it seems here that these qua-phrases have the stutter as their subject, indicating two aspects of the stutter. But if (ii) the qua-phrases indicate two different aspects of the stutter, then we do not need to (i) posit two notions of perfection. For the flourishing of the stutter and the flourishing of speech (of the ‘stutter qua speech’) are both perfections in the same sense. On the other hand, if (i) we are employing two notions of perfection, then there is no need for (ii) the qua-phrases to distinguish aspects of the stutter. Being rid of a stutter is different from the flourishing of that same stutter. Despite this unclarity, the general thrust of Kosman’s account suggests that (i) rather than (ii) represents his view.



Andreas Anagnostopoulos

of a definition that would otherwise pick out only the products of change, as Kosman suggests. (d) A difficult passage The consensus interpreter will object that in the immediately following passage Aristotle explains how the qua-phrase excludes the products of change. This passage, in fact, is thought to show that Aristotle saw a genuine threat of including the products of change in the definition, since they are energeiai and entelecheiai of potential beings, so that the qua-phrase, as the only available option, must exclude them. I consider the passage in two steps, beginning with the main argument: ὅτι µὲν οὖν ἐστιν αὕτη, καὶ ὅτι συµβαίνει τότε κινεῖσθαι ὅταν ἡ ἐντελέχεια ᾖ αὐτή, καὶ οὔτε πρότερον οὔτε ὕστερον, δῆλον· ἐνδέχεται γὰρ ἕκαστον ὁτὲ µὲν ἐνεργεῖν ὁτὲ δὲ µή, οἷον τὸ οἰκοδοµητόν. καὶ ἡ τοῦ οἰκοδοµητοῦ ἐνέργεια, ᾗ οἰκοδοµητόν, οἰκοδόµησίς ἐστιν (ἢ γὰρ οἰκοδόµησις ἡ ἐνέργεια [τοῦ οἰκοδοµητοῦ] ἢ ἡ οἰκία· ἀλλ᾽ ὅταν οἰκία ᾖ, οὐκέτ᾽ οἰκοδοµητὸν ἔστιν· οἰκοδοµεῖται δὲ τὸ οἰκοδοµητόν· ἀνάγκη οὖν οἰκοδόµησιν τὴν ἐνέργειαν εἶναι)· ἡ δ᾽ οἰκοδόµησις κίνησίς τις. (b–) That change is this, and that it happens that something is being changed when there is this entelecheia, and neither before nor after, is clear. For each thing admits at one time of energein and at another time not, for example the buildable; and the energeia of the buildable, qua buildable, is the oikodomēsis . . . and the oikodomēsis is a kind of change.

Here Aristotle announces his aim: to confirm his definition by showing that it picks out change, or at least that what it picks out is simultaneous with change. And he will do this by focusing on the case of the buildable. The entelecheia of the buildable is identical to, or at least simultaneous with, the relevant change. Aristotle’s argument for this conclusion, at least in outline, is straightforward. He identifies the building process (oikodomēsis) both with the entelecheia/energeia specified by his definition (Aristotle switches terms within this passage) and also with what is intuitively the relevant change. So much is relatively clear and uncontroversial. But Aristotle’s support for the identification of the oikodomēsis with the  See Kosman, ‘Motion’, , and Coope, ‘Change’, . This passage is the basis of the temporal reading offered by Penner, ‘Verbs’, –, and Kostman, ‘Definition’, ; see n.  above.  In the translation I have left out the parenthetical remark.

Change in Aristotle’s Physics 



b

relevant energeia at  –, omitted in the above translation, is grammatically ambiguous. According to the standard grammatical reading, it is as follows: For either the oikodomēsis is the energeia, or the house [is the energeia]. But when the house is, the buildable no longer is; the buildable is being built. So the oikodomēsis must be the energeia. (b–)

This standard reading of the passage appears to support the idea that the house is an energeia of the buildable (‘for either the oikodomēsis is the energeia, or the house [is the energeia]’, b– ) and so an entity that Aristotle’s definition runs the risk of including. His subsequent exclusion of the house is then taken to explicate the function of the qua-phrase, in line with the consensus interpretation. Granting the standard grammatical reading for now, let us look more closely at how Aristotle excludes the house. He places a constraint on when the relevant energeia can exist and points out that the house does not meet this constraint. In general, ‘each thing admits at one time of energein and at another time not’ (b–). He then applies this principle to the particular case of ‘the buildable’ in order to show that his definition excludes the house. In accordance with an earlier claim that potential and actual being are exclusive (a–), he claims that once there is a house, the buildable is no longer (b), and excludes the house on that basis. The constraint is thus that the relevant energeia can exist only when the buildable exists; the house does not meet this constraint. The passage at b– is usually taken to provide a constraint on being the energeia of the buildable qua buildable, leaving open the possibility that some other energeia of the buildable—an energeia of the buildable not qua buildable—might exist even when there is no longer anything buildable. But Aristotle simply does not say this. Rather, I claim, he provides a constraint on being any energeia of the 

The Greek text is given above; see further n.  below. I am, however, reluctant to ascribe to Aristotle the view that potential being never persists through a change. The reading of b– above, inspired by Waterlow’s discussion (Nature, –), depends only on the thesis that potential being is extinguished by the end of the change in this particular case, not on a general extinction thesis. Aristotle’s treatment of potential and actual being as exclusive here and earlier in the chapter may be a heuristically useful simplification, in much the same way that, according to D. Ebrey (‘Why Aristotle Needs Matter’, unpublished manuscript), Aristotle relies on the persistence of matter in Physics . . 



Andreas Anagnostopoulos

buildable, whether ‘qua buildable’ or not. Aristotle’s exclusion of the house does not illustrate the function of the qua-phrase. In support of this claim, note that taking the qua-phrase to exclude the house in this passage does not sit well with his explanation of the phrase’s function in the preceding lines. There he explained that the qua-phrase distinguishes energeiai of two beings that are one-in-number, as one might distinguish between the entelecheiai of what is capable of health and what is capable of sickness, which are one-in-number by sharing an underlying subject (whether moisture or blood). In the present text, however, we are not asked to distinguish between two candidate energeiai that are proper energeiai of two coinciding beings. It is rather that once the house exists, we are not left with anything buildable. Whatever we are left with— whatever the house might be the energeia of, if it is the energeia of anything—is not one-in-number with the buildable. Moreover, Aristotle’s use of energein (b) in stating the principle by which he excludes the house is difficult to square with the consensus reading of the qua-phrase. For first, energein is a substitute for energeia (and presumably, entelecheia) but the house does not seem to be an instance of something’s energein at all. Aristotle appears to be drawing on the plausible idea that something cannot engage in activity when it does not exist. Second, even if it is granted that being a house is an instance of something’s energein, still it is not an instance of the potential house’s energein—even accidentally—so long as the potential house does not persist into the finished house, as Aristotle here maintains. One might object that Aristotle considers the house an energeia of the buildable when he considers its candidacy for what the definition picks out: ‘either the oikodomēsis is the energeia, or the house [is the energeia]’ (b–). If so, one might reason, the qua-phrase must be responsible for excluding the house. But the fact that Aristotle brings up the house as a candidate for the energeia of the buildable qua buildable does not imply that he thinks it is or has any plausible claim to be that energeia (or any other energeia). In par Note that my reading alone can accommodate τοῦ οἰκοδοµητοῦ at line b, found in MS E (tenth cent.) and in Themistius. With these words included, the text reads (again, on the standard grammatical reading): ‘For either the οἰκοδόµησις is the ἐνέργεια of the buildable, or the house [is the ἐνέργεια]’ (b–). If Aristotle meant to specify the ἐνέργεια of the buildable qua buildable, as opposed to that of the buildable (but not qua buildable), he would not characterize the former as ‘the ἐνέργεια of the buildable’.

Change in Aristotle’s Physics 



ticular, he simply does not say that the house is an energeia of the buildable, or that his definition, without the qua-phrase, would pick it out. Moreover, there are several plausible explanations for his bringing up the house. He might do so because the house, though not the energeia of the buildable, is an energeia, perhaps in some extended sense (see Section  below); or because others had difficulty defining change in a way that excluded the products of change; or because only the oikodomēsis and the house bear any kind of per se connection to the potential house. Whatever the explanation for the house’s candidacy, the contrast between the house and the oikodomēsis allows Aristotle to illustrate how his definition picks out the change by specifying the subject that undergoes it. So, even on the standard grammatical reading of this passage, it should not be taken to support the consensus reading of the quaphrase. And on the alternative grammatical reading of the passage, the proposition that the house is a candidate for the relevant energeia is not so much as entertained: For either the oikodomēsis is the energeia or [the oikodomēsis is] the house. But when the house is, the buildable no longer is; the buildable is being built. So the oikodomēsis must be the energeia. (b–)

On this reading, the passage considers two candidates for the oikodomēsis, the energeia and the house, and rules out the latter, thus supporting the earlier claim that ‘the energeia of the buildable, qua buildable, is the oikodomēsis’ (b–). One advantage of this reading is that on it Aristotle focuses on limiting the duration of change (the oikodomēsis, which is assumed to be a change at b) so that change occurs only when the relevant energeia occurs, in line with his explicit aim, which is to argue that ‘something is being changed when there is this entelecheia, and neither before nor after’ (b–). On the standard reading, Aristotle instead limits the duration of the energeia to that of the relevant change (i.e. oikodomēsis).  Pieter Sjoerd Hasper has persuaded me of the plausibility of the alternative reading and pointed out the advantage of it mentioned above. However, I find it very difficult to understand why Aristotle should consider the possibility that the οἰκοδόµησις is the finished house. My worry is not that it is obviously, perhaps analytically, false that the οἰκοδόµησις is the house. Aristotle need not be taken to be responding to a plausible alternative view. My worry is that on this reading the extension of Aristotle’s definition, which is supposed to be puzzling and difficult to understand, is treated as given. On the standard reading, there is something in the way of an explanation of how the definition manages to pick out changes.



Andreas Anagnostopoulos

(e) ‘Qua such’, the product puzzle, and actuality I have argued that the qua-phrase cannot be responsible for excluding the products of change from the scope of the definition, as the consensus interpretation claims. It follows, I think, that the products of change are not in fact entelecheiai and energeiai of potential beings, as they are widely assumed to be. For there is no other mechanism that could then exclude the products of change from the scope of the definition. Once this widespread assumption, that the products of change are entelecheiai and energeiai of potential beings, is given up, the product puzzle can be rejected out of hand. At the very least, Aristotle does not recognize a risk of picking out the products. And without the assumption that the products of change are entelecheia and energeiai of potential beings, the sometimes elaborate accounts of how the qua-phrase distinguishes products from changes become superfluous. The problematic assumption that the products of change are entelecheiai and energeiai of potential beings is entailed by the conjunction of two more basic theses: (i) that entelecheia and energeia mean ‘actuality’, and (ii) that the products of change are actualities of potential beings. It should be noted that commentators have  This assumption is to be distinguished from the similar assumptions that the house (or its form) is an ἐνέργεια and ἐντελέχεια (full stop), or an ἐνέργεια and ἐντελέχεια corresponding to a certain potentiality, or an ἐνέργεια and ἐντελέχεια that some potential is for or directed towards. In view of how often this assumption is made, it is surprising that there is little direct textual evidence in its favour. Aristotle’s use of ‘ἐνέργεια of X’ almost always refers to the activity that X engages in. As far as I know, the only place where Aristotle clearly characterizes the product of a change (or its resulting form) as the ἐνέργεια of what is capable of becoming it is at DA . , a–, where Aristotle claims that health, which is in one sense that by which we are healthy (ὑγιαίνοµεν), is ‘a sort of shape and form, and account and as it were energeia of what is receptive . . . [i.e.] of the curable [ὑγιαστοῦ]’. Note that Aristotle is here talking about health (ὑγίεια) as an ἐνέργεια, whereas Phys. . , a–b, is about being healthy (ὑγιαίνειν), which is in the DA passage something we do by virtue of health. The situation with ἐντελέχεια is more difficult. In general, ‘ἐντελέχεια of X’ seems to refer to that in virtue of which something is an X ἐντελεχείᾳ, or if not that, at least to that in virtue of which an X (while still X) is the kind of being an X is directed at ἐντελεχείᾳ. A full defence of this point would require analysis of several quite difficult and controversial passages; for example, the soul is said to be the ἐντελέχεια of an instrumental body, of a body potentially having life (DA . , a ff.); light is the ἐντελέχεια of the transparent (DA . , a); the pilot is the ἐντελέχεια of the ship (DA . , a–). I am not convinced that there is a standard use of ‘ἐντελέχεια of X’ on which the existence of the ἐντελέχεια requires the extinction of X.  For these reasons, among others, I think the most plausible ‘actuality’ interpretation will read ἐντελέχεια as ‘constitutive actuality’ unambiguously and independently of the qua-phrase, so that the products of change will not qualify as ἐντελέχεια

Change in Aristotle’s Physics 



overwhelmingly endorsed the second thesis. In fact, they have often suggested that the products of change are proper or paradigmatic actualities of potential beings, and for this reason they have looked to—and in my view misconstrued—the qua-phrase in order to exclude them. In any case, my rejection of the consensus reading of the qua-phrase provides an additional, albeit conditional objection to the consensus reading of entelecheia and energeia as ‘actuality’— conditional on the idea that the products of change are actualities of potential beings.

. Change as the proper activity of potential being In this section I present my interpretation of each part of the definition, drawing on the account of the qua-phrase already begun, and explain how the definition manages to pick out changes without circularity. (a) Activity Aristotle defines change as a kind of activity, rather than actuality. I reject much of the framework for thinking about Aristotle’s definition that has provided support for the latter reading and believe that my reading can avoid many of the problems that we have encountered with it. As far as the text of Physics . – is concerned, my reading would be the most natural by far had Aristotle used only the term energeia in characterizing change. I address his use of the of potential being at all. As I mentioned earlier, a looser conception of the sense in which change is an ‘actuality of’ potential being—or worse yet, a failure to specify such a sense precisely—may make it more difficult to deny that the products of change qualify as actualities of potential being, and so increase the risk of picking out the products of change.  There is another way in which my analysis of the qua-phrase so far, as specifying that Aristotle means the proper rather than accidental ἐντελέχεια and ἐνέργεια of the potential being, is difficult to square with a reading in terms of actuality. To speak of an ‘accidental actuality’—especially an ‘accidental constitutive actuality’—of something is awkward in the way that it would be awkward to speak of something’s ‘accidental οὐσία’ or ‘accidental matter’. To be sure, one could, if necessary, find sense to assign to such phrases. Still, talk of the actuality of a thing is naturally and immediately understood to refer to its proper actuality, so that Aristotle’s emphasis on the qua-phrase seems misplaced. Versions of this point were brought to my attention on separate occasions by Klaus Corcilius and Christoph Helmig, although they might not see it as supporting my reading of ἐντελέχεια and ἐνέργεια.



Andreas Anagnostopoulos

term entelecheia, which does not elsewhere mean ‘activity’, in Section  below. Menn shows decisively that the Protrepticus, Eudemian Ethics, Topics, and Magna Moralia use energeia, often interchangeably with chrēsis (use), to refer to the exercise of a dunamis or hexis. The paradigmatic case is that of using or exercising knowledge as opposed to merely having it, and Aristotle applies the same language to sense perception, living, and virtue. It should be noted, however, that energeia applies indiscriminately to end-containing activities, the realizations of hexeis (such as contemplation and seeing), and ‘ordinary’ changes. The distinctions of De anima .  and Metaphysics Θ  between types or levels of energeia are not in evidence in these works. The concept of activity, as the standard meaning of energeia, is also indiscriminate between agency, i.e. acting on something, and ‘patiency’, i.e. suffering or being acted on. In particular, when the patient of change is changing, it is exercising a potential. It is in this sense engaged in activity. This point is crucial, since I take Aristotle to define change as the patient’s activity of suffering. Thus Aristotle has a concept of energeia as activity that uncontroversially covers changes but is not limited to them, so that he may use it to define change without circularity. And in employing this concept in his definition of change, Aristotle is not characterizing change in terms of some specialized kind of energeia, such as hexis-realization or end-containing activity, with which change is otherwise contrasted. In this regard, my interpretation draws on the well-attested Aristotelian thesis that all, but not only, changes are activities. It does not require attributing to him stronger and more questionable theses about the metaphysics of change. It is extremely important to distinguish this notion of activity from that of actualization. Actualization, at least as I understand it, involves reference to an initial state before the actualization and to a state of being actual that the actualization aims to reach. Actualization is a transition from potentiality to actuality. This is the source of a circularity charge I mentioned earlier. But the concept  S. Menn, ‘The Origins of Aristotle’s Concept of Ἐνέργεια: Ἐνέργεια and ∆ύναµις’ [‘Origins’], Ancient Philosophy,  (), – at –.  Waterlow, Nature, –, points this out.  Whether a definition in terms of ‘actualization’ is in fact circular depends on whether the relevant notion of transition is identical to the notion of change being defined. Thus, one might take ‘actualization’ to apply also, for example, to the transition from merely having knowledge to actively contemplating, while denying that

Change in Aristotle’s Physics 



of activity is, as I mentioned, wider than, and thus distinct from, that of change. Moreover, activity is not essentially a transition to actuality, since there are activities that are their own ends, and so already involve being in the relevant state of actuality. Finally, reading entelecheia and energeia as ‘activity’ brings to the fore a threat to Aristotle’s definition: that it will also pick out activities that are not changes. Exactly which activities these are remains controversial. My own view is that Aristotle’s definition should exclude (i) the realizations of hexeis discussed in De anima . , such as seeing and contemplating, (ii) end-containing activities as specified at Metaph. Θ , b–, and (iii) transitive agency—an agent’s acting on (even changing) a patient (leaving open how these three classes intersect). But it is not obvious how these activities should be excluded from the scope of the definition. Note that most if not all of them are activities of potential beings. For example, contemplation (of the human variety) is the activity of a ‘potential knower’ in the sense specified at DA . , a–; in fact, it is the activity that a potential knower engages in, as Aristotle might put it, ‘qua potential knower’. Here I would like to point out only that excluding these activities is no more a challenge for my interpretation than for an interpretation in terms of actuality. For if changes are actualities of potential beings, then surely some of these activities are as well; indeed, realizations of hexeis and end-containing activities are often taken as models for thinking of change as actuality. this transition is a change. In my view, the most decisive objection to the ‘actualization’ reading is that neither ἐντελέχεια nor ἐνέργεια means ‘actualization’. This is especially clear if ‘actualization’ is understood broadly as suggested above. For while the continuous realizations of ἕξεις (e.g. seeing, contemplating) are characterized as ἐνέργειαι, the transitions to those activities are merely transitions ‘to ἐνέργεια’ (DA . , b) but not themselves called ἐνέργειαι.  Kosman’s answer to this challenge, as far as I can tell, is that in the case of change alone there is an ‘ultimate’ actuality distinct from it, which is deprivative (‘Motion’, ), or that ‘for an energeia [in the sense of end-containing activity] there is no difference between the acting out of a potentiality qua potentiality and the acting out of that potentiality simpliciter’ (‘Substance’, ). These claims, however, imply that the non-change activity at issue is still a ‘constitutive actuality’ and ‘the acting out of a potentiality qua potentiality’ (whatever else it might be in addition), and so do not explain why it should be excluded from the definition’s scope. See above, n. . Commentators often rely on the distinction between end-containing and end-exclusive activities at Metaph. Θ , b–, in order to exclude certain activities from the scope of Aristotle’s definition. See Penner, ‘Verbs’, , Gill, ‘Causal Action’, , and Waterlow, Nature, . However, it is not clear whether the account of change in Physics . – involves or implies that distinction.



Andreas Anagnostopoulos

(b) Potential being I take Aristotle to be talking about potential being rather than potential becoming. The kinds of potential being Aristotle has in mind are more specifically potential being in the four categories of substance, quality, quantity, and place. Also, I mean potential being (dunamei on) rather than the power or capacity (dunamis) in virtue of which something is a potential being. For example, I mean an acorn that is a potential tree rather than the acorn’s capacity to become a tree. Thus, for example, alteration is defined as the activity of what is potentially qualified, not of what is potentially in process of alteration, nor of the power or capacity (dunamis) to be or become qualified. Regarding the choice between potential being and potential becoming, it must be admitted that Aristotle repeatedly, and more often than not, refers to the subjects of change as the ‘changeable’, the ‘alterable’, etc., and this preponderance suggests that he means a potential ‘becomer’ after all. I believe, however, that such an interpretation is unwarranted, though Aristotle’s use of such terms cannot be fully explained away. First, as several scholars have pointed out, an interpretation in terms of potential becoming would make the definition vulnerable to a straightforward charge of circularity, and is thus extremely uncharitable. Second, when Aristotle introduces the relevant concept of potentiality, it is clear that he has in mind potential being rather than becoming. Aristotle begins the discussion of  This is important in so far as one might naturally take ‘ἐνέργεια of a δύναµις’ to mean something like ‘exercise of a potential’, which would not obviously make sense if applied instead to some actual being (e.g. ‘exercise of bronze’). This is not to deny that the ἐνέργεια of a potential being is in general the exercise of the corresponding δύναµις. The present passage offers no indication that talk of δυνάµει ὄντα can be reduced to talk of δυνάµεις, or of how such a reduction might go, except for the interpretative desideratum that the relevant sense of δύναµις not be itself defined in terms of change.  See Kosman, ‘Motion’, ; Waterlow, Nature, ; and Coope, ‘Change’, . I am not convinced by Charles’s attempt to defuse the charge by claiming that phrases such as ‘changeable’ leave ‘a gap for a positive characterization of the basis and nature of the relevant capacity’ (Action, –). For aside from telling us that changes issue from capacities for change, this account, as Charles admits, ‘rests on the assumption that there is a distinctive type of capacity which is actualised in all cases of change’ (). The problem is that we have no independent grasp of what this ‘positive characterization’ is, and so of what the capacity is, and so we have gained no insight into what change in general or each kind of change is. This will not help us understand what ϕύσις (nature) is, which is the explicit purpose of the definition of change.

Change in Aristotle’s Physics 



change by introducing the concepts of being potentially (dunamei) and being entelecheiāi, which comprise one of the four divisions of being that are candidates for separate study in the Metaphysics. The adverbial dative uses of dunamis and entelecheia here (b–) are meant to modify kinds of (categorial) being. That Aristotle intends to rely on a distinction between potential and actual categorial being is evident also from the very sentence that offers the first statement of his definition, where that distinction is repeated: ‘these having been distinguished according to each kind [genos] [of being], [being] actually and [being] potentially, the entelecheia of the potential being, qua such, is change’ (a–). In addition, since this is the first statement of the definition, it should be given significant weight. Finally, when Aristotle uses phrases of the form ‘what is F potentially [dunamei]’ in this chapter to talk about particular examples, he talks only about beings in the categories of quality and substance: ‘what is potentially cold’ (a–) and ‘what is potentially a statue’ (a). Nothing I have said above explains away Aristotle’s employment of the troublesome phrases such as ‘changeable’ and ‘alterable’. Perhaps, having given his general definition of change and made clear that it is intended to employ the concept of potential being, Aristotle turns his attention at least partly to classifying and distinguishing the different kinds of kinēsis in the way that his general definition suggests: different types of change will be proper activities of different types of potential being. That this is an important task can be seen, for example, in the opening chapters (especially . ) of On Generation and Corruption, where one of the framing questions is whether generation and alteration are the same or different, or in Physics  and , where various relations, dependencies, and contrasts are posited between different kinds of change. A word such as ‘alterable’, although it does not mention the category of being which the object is potentially, characterizes the subject of a certain kind of change in a way that invites one to correlate it to the particular kind of change it undergoes. Talk of ‘the alterable’, while strictly inappropriate in a context of defining alteration, is thus understandable if Aristotle’s interest has shifted partly to showing how the structure of his definition allows him to classify and distinguish types of change. In addition, there is no single word that means ‘potentially 

See Metaph. ∆  and Ε , a–b.



Andreas Anagnostopoulos

qualified’, and so on, for each of the four relevant categories, making talk of potential being rather cumbersome. So far I have argued that Aristotle intends to characterize change in terms of potential being rather than potential becoming—that terms such as ‘potential being’, rather than terms such as ‘changeable’, represent his considered view. But even if this is granted, a deeper concern remains, namely, that even talk of potential being (dunamei on), for Aristotle, can ultimately be understood only in terms of change. For there is only one analysis of the term dunamis that could plausibly be taken to be a definition of the term, and it characterizes dunamis in terms of change (Metaph. Θ , b– a; ∆ ). I cannot here address this deeper concern. (c) ‘Qua such’ With the phrase ‘qua such’ Aristotle makes it clear that he means the proper activity of a potential being, rather than just any, even  The reasons Gill, Paradox, , cites for Aristotle’s use of ‘changeable’ etc. are of a similar nature, but she also suggests that these phrases are intended to capture the idea that the notion of potential being at issue is privative. I do not find either of our explanations fully satisfying because Aristotle insists that, just as there is no being apart from and common to the various categories of being, since change is always change according to some category, neither is there change apart from (and presumably common to) the particular species of change (b–a). This suggests that Aristotle needs to define the species of change in addition to (if not instead of) change in general. But he nowhere gives a definition of a species of change in terms of potential being in the relevant category. For terms such as ‘alterable’ infect every characterization of an individual species of change.  There are, I think, (at least) three ways in which one might reasonably resist the force of these considerations. First, one might claim that in an adverbial dative construction such as δυνάµει ὄν, which appears in the definition of change (‘potential being’), the term δύναµις takes on a meaning different from, and independent of, the notion of δύναµις as capacity. That the adverbial dative construction involves a different meaning of δύναµις (‘potentiality’ rather than ‘capacity’ or ‘power’) is suggested by many recent studies of Metaphysics Θ: M. Frede, ‘Aristotle’s Notion of Potentiality in Metaphysics Θ’, in T. Scaltsas, D. Charles and M. L. Gill (eds.), Unity, Identity, and Explanation in Aristotle’s Metaphysics (Oxford, ), –; C. Witt, Ways of Being: Potentiality and Actuality in Aristotle’s Metaphysics (Ithaca, NY, ); S. Makin, Aristotle: Metaphysics Book Theta (Oxford, ). These studies do not, as far as I can tell, take a stand on whether the notion of potentiality is definitionally independent of the notion of capacity. They also do not explicitly address the circularity problem arising from the use of δύναµις in the definition of change. A second strategy is to accept that talk of δυνάµει ὄντα is to be understood in terms of capacities but to insist that these are capacities for being rather than the capacities characterized in terms of change in Metaphysics Θ  and ∆ . Third, one could accept that talk of δυνάµει ὄντα is to be understood in terms of those capacities characterized in terms of change, but insist that those capacities can also be understood without reference to change, as capacities for being. See Waterlow, Nature, –.

Change in Aristotle’s Physics 



accidental, activity of that being. The qua-phrase thus enables the definition to exclude activities that, strictly speaking, are activities of a being with which the potential being merely coincides. The phrase cannot do what the consensus interpretation requires of it— to exclude the products of change—but no such need arises for my interpretation, since the products of change are not activities of potential beings. Up to now I have focused on the extensional work of the quaphrase, but the phrase’s function does not consist entirely or even primarily in excluding some other activity it might otherwise be taken to include. More generally, the purpose of the definition is not merely to specify the right extension. Indeed, if we focus only on the extensional work of this aspect of Aristotle’s definition, his choice of examples is perplexing. What Aristotle does with the case of the bronze that is also a potential statue (a–) is straightforward enough, but we may wonder what the activity of bronze qua bronze amounts to. Is there a proper activity of being bronze—something perhaps easier to accept in the case of living organisms—or perhaps no such proper activity at all? The extensional function of the quaphrase might be better illustrated by a case of two uncontroversial energeiai belonging, the one properly, the other accidentally, to the same being. If, for example, an object grows and alters at the same time, we can distinguish these energeiai as different kinds of change by noting that they are properly speaking energeiai of different beings, the one of what is potentially so-qualified, the other of what is potentially large. Turning away from extensional concerns, let me now say more about what the qua-phrase contributes to the meaning of the definition. Earlier, I said that to be the proper entelecheia (or energeia)  At least there is no general need. But perhaps there are cases in which being is an activity—the activity, for example, of being human. My position is consistent with at least a version of this Thomistic view. So long as these actus essendi are proper activities of actual rather than potential beings, they will be not be picked out by Aristotle’s definition, as I understand it. See Kosman, ‘Substance’, n. , and Menn, ‘Origins’, –, esp. n. , on the Thomistic view.  On this reading, Aristotle does not presuppose that there is such an activity, but only points out that change, as he defines it, is not such an activity, as the current Spanish king might say, ‘Of France, I am not the king’.  As for activities that are not changes of any kind, e.g. contemplation, the quaphrase may well distinguish the proper subject of change from the proper subject of contemplation. This alone would not explain why, for example, contemplation does not satisfy the definition by being the proper activity of its proper subject.



Andreas Anagnostopoulos

of some being is to qualify as that being’s entelecheia in virtue of that being’s being what it is, as opposed to its being something else. This cumbersome phrasing reflects a desire to remain neutral on the interpretation of entelecheia and energeia. We may now ask, more specifically, what is meant by calling change the proper activity of some being, i.e. that being’s proper energeia in the broad sense I have mentioned. To start, the definition clearly picks out two entities that are for Aristotle related as (material) cause and thing caused—the proper subject of change and the change itself that this subject undergoes. The fact that the definition picks out causally related entities does not yet imply that the definition posits such a causal relation. However, it seems to me that to call change the proper activity of something is to posit a causal relation between change and that whose activity it is. We may speak loosely of any activity a thing engages in as an instance of its energein. But to specify the activity as a being’s proper activity—as the activity of that being ‘qua such’—is for Aristotle to specify the activity as causally explained by that being. If so, we may reformulate the definition as follows: change is the activity a potential being engages in because it is a potential being. Aristotle’s aim is to say what change is, and this requires saying what being it is whose proper activity is change, even if that being always, even necessarily, coincides with some other being. To bring out this point, consider the view that the privation and the relevant potential being (which I take to be matter) are necessarily onein-number during the change. For example, the unmusical and the potentially musical (the man) are necessarily one-in-number during the man’s becoming musical. Thus, becoming musical is in some sense the activity of the unmusical. But Aristotle would forcefully deny that becoming musical is the proper energeia of the unmusical—that something becomes musical because it is un However, not every being—not even every potential being—is the material cause of its proper activity. For example, an agent (e.g. a housebuilder) is the efficient, not material, cause of its own proper activity. If such an agent is also to be described as a potential being, then the proper activity of a ‘potential being’ in this sense need not have that potential being as its material cause. I have here benefited from a related suggestion made by Christian Pfeiffer that Aristotle’s use of δύναµις and ἐνέργεια can often be seen as indicating that the phenomenon they describe is a causal structure of one sort or another.  This view seems to me implied by the present text in combination with Physics . On this necessity see S. Kelsey, ‘Aristotle Physics I. ’, Phronesis,  (), –.

Change in Aristotle’s Physics 



musical, just as he denies that what is musical comes-to-be from what is not musical, qua not musical.

. Aristotle and his predecessors One might worry that the account I have offered leaves Aristotle without a compelling response to his predecessors—Parmenidean and others—who harboured doubts about the possibility or ontological status of change. Faced with such worries, scholars contend, Aristotle responds, as he must, by defining change as an actuality. (The unremarkable idea that change is an activity is comparatively impotent in this regard.) But it is rather the rest of the definition, ‘of a potential being, qua such’, that is crucial in responding to such doubts. This part of the definition enables Aristotle to address the scientific respectability of change in a way that is deeply connected to his discussion of change in Physics , especially to his treatment of the so-called Parmenidean dilemma in chapter . As for the problematic previous accounts of change that he mentions in Physics . , Aristotle’s central innovation is to see that the notion of potential being is needed, and how it should be employed, in giving sufficient conditions for change. The suggestion that Aristotle should define change as an actuality in order to respond to his predecessors is implausible, both in general and with regard to the particular sorts of doubt at issue.  I take Aristotle and his Parmenidean opponent to agree that this is impossible in Physics . . See a– for the statement of the opponent’s position and b–  for Aristotle’s agreement on the version of that position mentioned above. Because Aristotle distinguishes matter and privation in Physics , and denies to the latter the status of that out of which, as such, things come-to-be, I am reluctant to have the definition depend on ‘potential being’ implying the corresponding privation. See Gill, ‘Causal Action’, , and Paradox, –; Waterlow, Nature, –; and Coope, ‘Change’, .  Waterlow writes of the ‘actualization’ reading: ‘to offer such a definition would be to give up the fight to show that process and change are themselves real and actual. For if one says only that a change tends towards some eventually actual end-state, one is left with no basis for maintaining that the tending itself is real while it continues and of an ontological status commensurate with that of the actuality brought about’ (Nature, ). Surely she would say the same about my ‘activity’ reading, except that it does not even have the virtue of implying that change ‘tends towards some eventually actual end-state’. See also Johansen, Sense-Organs, –, and Coope, ‘Change’, .  See Waterlow, Nature, –; Johansen, Sense-Organs, –; Burnyeat, ‘De anima II ’, , and ‘Passage’,  and ; and Coope, ‘Change’, –.



Andreas Anagnostopoulos

First and in general, Aristotle’s definitions are intended to structure knowledge, and so we should hold them not to standards determined by the dialectical context in which he finds himself, but to standards of truth. This does not mean that definitions can be arrived at independently of consideration of the received views, or even that such definitions are useless in responding to his opponents. But if they are useful, presumably it will be because his opponents have overlooked some part of the nature of a thing that Aristotle’s definition makes clear, or that they have failed to see how to define something without contradiction, or some such error. Second, with regard to doubts about the very possibility and existence of change, even if Aristotle’s opponents were to accept that change is by definition a kind of actuality, they might insist that the concept of change is not and perhaps cannot be instantiated for other reasons. And in fact, many of them found the very concept of change problematic in ways that are simply not addressed by characterizing change as entelecheia or energeia, however these are understood. Aristotle’s Parmenidean predecessors, for example, had an enormously influential argument behind their admittedly counter-intuitive conclusion, an argument that works by showing that there is nothing out of which things could intelligibly come-to-be. To insist, against their conclusion, that change is something actual or otherwise ontologically respectable would be futile. Third, with regard to doubts about the ontological status of change—for example, whether change is ‘as real as anything else actual is real’—I have already argued that entelecheia and energeia do not mean ‘actuality’ in the connotations that would be most helpful here, namely, the senses of ‘actuality’ that are opposed to the merely possible or to the fictional, imaginary, or otherwise unreal. But if we draw more cautiously on the notion of entelecheia as a mode of being some kind of thing fully or completely, the  The same goes for Kosman’s suggestion in a later paper that Aristotle’s definition ‘is aimed precisely at explaining the respect in which becoming is a “kind” of being’ (‘Substance’, ), though he does not suggest that this might have any value against those who have doubts about the existence or ontological status of change. My claims here accord with what Aristotle says in Posterior Analytics .  and .  about definitions that are ‘demonstrative’ in that they are deductions that differ in arrangement. For the demonstration will depend on the existence of the prior terms as a premiss: in this case, for example, ‘potential being’.  Plato’s Parmenides ( –) represents Zeno as defending ‘ridiculous’ Parmenidean positions by showing that the alternatives are yet more ridiculous.

Change in Aristotle’s Physics 



definition may not provide a characterization of change robust enough to counter doubts about its ontological status. On the ‘constitutive actuality’ proposal, for example, the relevant kind of being is potential being, so that the definition will say that a change is ‘actually being a potential being’, thus ascribing to change only the level of actuality that potential beings can achieve. The ontological status of change will be left no better off than that of potential being, and an opponent with doubts about the ontological status of change might well have similar doubts about that of potential being. And even for Aristotle, it is precisely in comparison with potential being that entelecheia as actuality is thought of as having a higher ontological status. We shall see that while some thinkers relegated change to a lesser ontological status, this does not lie at the forefront of Aristotle’s dispute with them. I now offer an account of how the definition is useful in responding to two kinds of doubts mentioned above, taken in turn. As I understand it, Aristotle’s dispute with the Parmenidean opponent in Physics .  involves not ontology but the structure of explanation. The Parmenidean is faulted for his failure to grasp a formal principle: that causal explanation is sensitive to the differences between merely coinciding beings. Aristotle seeks that out of which, qua itself, things come-to-be. Seeing that ‘everything is or is not’ (b–), the Parmenidean opponent thinks that ‘what is’ and ‘what is not’ are the only available candidates for this per se source of coming-to-be. He is unable to isolate the causal contribution of some third being (e.g. an acorn) since it would have to coincide with ‘what is’ or with ‘what is not’ (e.g. what is a tree or what is not a tree). That is, the opponent is unable to conceive of there being a third explanatory principle of change in addition to the agreed-upon opposites. Aristotle’s resolution of the dilemma in Physics .  commits him to the idea that that out of which, qua itself, things come-tobe—matter—must be one-in-number with, but distinct-in-being  More generally, there is a tension between the idea that ἐντελέχεια should be understood as a mode of being some kind of being actually and the idea that the definition bolsters the ontological status of change by characterizing it as ἐντελέχεια. For if the first idea is correct, then simply specifying that change is an actuality will not necessarily imply that changes enjoy some unitary degree of actuality shared by all actual beings, but only that they enjoy the degree of actuality appropriate to the particular kind of being at issue.  I hope to develop this interpretation elsewhere.



Andreas Anagnostopoulos

from, the privation. However, aside from some hints in Physics .  (a–), there is no indication in the first book of the Physics that matter essentially involves potentiality. In Physics .  Aristotle’s allowance for potential being within his ontology is explicit. Note that Aristotle’s definition of change exploits the same formal principle in its use of the qua-phrase. For to say that change is the activity of a potential being ‘qua such’ is to distinguish change from accidental activities of the potential being. These other activities will be proper activities of some other being with which the potential being coincides—perhaps even must coincide if it is to undergo change—such as the corresponding privation. Moreover, such an intrinsic relation between a potential being and the change it undergoes is required for potential being to play the (unmistakably causal) role of matter in a change, i.e. as that out of which, qua such, something comes-to-be. To this extent, the definition of change provides the conceptual underpinnings for the idea that change has a material cause distinct from the two opposites, form and privation. Note that this point holds not only if the definition posits a causal relation between change and its subject, as I suggested earlier, but also if it posits only a weaker logical or conceptual relation. Finally, this intrinsic connection between change and what undergoes it—a relation without which change will be left unintelligible—is here codified into an account of what change is, rather than a statement of the causes and principles of change. In sum, while Aristotle’s treatment of the Parmenidean dilemma makes room for the requisite material cause of change by removing an obstacle to there being an explanatory principle of change in addition to the opposites, the definition of change specifies change as intrinsically tied to, and essentially explained by, potential being. The definition of change thus legitimates the enterprise of natural science in so far as it ensures that one of its primary objects of study is subject to scientific treatment. One might think that the account of the dialectical significance of the definition given so far is equally available to an interpretation in terms of actuality. After all, if Aristotle defines change as the actuality of potential being, surely he means the proper actuality, so that the definition, so understood, will posit an intrinsic connection between the change and the potential being that is its material cause. However, an entity is not in general the material cause of its

Change in Aristotle’s Physics 



own ‘constitutive actuality’. For example, a man is not a cause of his constitutive actuality, in the sense of his actually being a man or that in virtue of which he is an actual man—his form. The actuality we are talking about here is already a component or aspect of the man. Similarly, the potential house is not a material cause of the alleged ‘constitutive actuality’ of the potential house; for this actuality (the being built) is on this view already a component or aspect of the potential house. And even if the proper activity of potential being can in some other way be understood as an actuality of that potential being—what I have expressed doubts about in Section (c)—this certainly does not add anything with respect to the causal underpinnings of change. Aristotle does not mention a Parmenidean opponent in Physics . But he does compare his own definition favourably to others’ attempts to understand change: That this has been said correctly is clear both from what others say about change and from the fact that it is not easy to define it in another way. For one would not be able to put change and metabolē into another genus. This is clear if we examine how some people assign it, claiming that change is difference and inequality and what is not. None of these necessarily changes— neither when things are different nor when unequal nor when they are not; and metabolē is not even into or from these things any more than [into or] from their opposites. The reason for assigning it to these is that change is thought to be something indefinite, and the principles of the second column are indefinite because they are privative; none of them is a ‘this’ or a ‘such’, or belongs to any of the other categories. And the reason why change is thought to be indefinite is that it is not possible to assign it either to dunamis of things that are or to [their] energeia; for neither what is able [dunaton] to be of some quantity nor what is of some quantity in energeia [energeiāi] necessarily changes. (b–)

I shall not give a full analysis of these views and their shortcom It may seem that, in insisting on this point, I am asking for too much precision in Kosman’s (and perhaps Aristotle’s) phrasing. Can we not discern, in the phrase ‘ἐντελέχεια of the potential house’, a reference to the bricks and planks in abstraction from, or even before, their being built? I think we cannot, because the definition would then refer not to the ‘potential house’, some component of which is the being built, but instead to the potential house considered merely as a ‘potential potential house’, that is, in a way we could equally consider the dormant bricks and planks. But if this ‘potential potential house’ is the entity the definition refers to as ‘potential being’, then we are no longer talking about that of which change is the constitutive actuality. For the change does not constitute actually being a ‘potential potential house’ any more than the form of man constitutes actually being a potential man.



Andreas Anagnostopoulos

ings. However, we may note the following: Aristotle certainly does consider views that define change as ‘difference and inequality and what is not’ and in general as indefinite and privative. He is thus concerned to counter views that, in these ways, relegate change to a lesser ontological status. However, first, Aristotle does not fault them for assigning to change a lesser status any more than he faults the view that change is (unqualified) energeia for assigning to it a greater status. His basic objection to these accounts of change is that they do not provide necessary and sufficient conditions for change, whether they are understood as attempting to specify the poles of change or the changing subject. And this form of objection applies equally to accounts of change as ‘difference, inequality and what is not’ and to accounts of change as dunamis or energeia of ‘things that are’. Different and unequal things, as well as ‘what is not’, neither necessarily change nor have a better claim than their opposites to being the poles of change (b–). Similarly, ‘things that are’ do not necessarily change, either when they are merely potentially or when they are ‘in energeia’ (b–). Second, although the first set of views Aristotle considers (that change is ‘difference, inequality, and what is not’) arise from a more basic thought that change is something indefinite (b–), this thought in turn arises from the thought that ‘it is not possible to assign it either to dunamis of things that are or to [their] energeia’ (b–). Thus, these theorists do not need to be told that change is something real. They need to be shown how to give sufficient conditions for change (how to ‘assign it’), so that they will not need to relegate change to the indefinite.

. Energeia and entelecheia Had Aristotle consistently used the term energeia, my position would be quite obvious and natural, and I mean to draw heavily on my understanding of that term. But his use of entelecheia is problematic for my interpretation since this term does not elsewhere have a connotation of activity. And the term energeia, used interchangeably here, appears to have a well-attested meaning of actuality in addition to its original one. However, before appraising this challenge, I would like to review some of the conclusions reached so far concerning the philosophi-

Change in Aristotle’s Physics 



cal support for, and ramifications of, the ‘actuality’ and ‘activity’ readings of these terms. I do so not only for the sake of summary, but also and more importantly in order to clarify the relative importance of Aristotle’s use of these terms in assessing the two readings. () The ‘actuality’ reading is not necessary in order to avoid circularity in the definition, since the ‘activity’ reading does not make the definition circular. () While the idea that changes are activities of potential beings draws on well-attested Aristotelian doctrine, the same cannot be said of the idea that changes are actualities of potential beings. The ‘constitutive actuality’ proposal is relatively obscure and in tension with Aristotle’s other stated views regarding changes and actuality or completeness. And it is not clear how else change should qualify as the actuality ‘of the potential being, qua such’. () Conditional on a widely held assumption—that the products of change are actualities of the corresponding potential beings—the ‘actuality’ reading is vulnerable to the threat of picking out the products of change, a threat that cannot be avoided by the addition of the qua-phrase. Moreover, while the products of change are not ‘constitutive actualities’ of potential beings, the products may well be actualities of potential beings if ‘actuality’ is construed in some other way. On the other hand, there is no obvious reason to think that the products of change or their forms are activities of potential beings. () Defining change as an actuality would not in any significant way enable Aristotle to address his predecessors’ doubts about change. In fact, we have seen that the ‘constitutive actuality’ proposal precludes taking the definition to have the consequences I have claimed it does for the causal explicability of change, and so for legitimating the change’s central role within natural science. For these reasons, reading the definition in terms of actuality does not make the definition more plausible or add to its philosophical value. On the contrary, much is to be gained by rejecting this reading in favour of the ‘activity’ reading. Thus, whatever advantage the ‘actuality’ reading has over the ‘activity’ reading—at least with respect to the considerations explored in this essay—will lie in its abi-



Andreas Anagnostopoulos

lity better to accommodate Aristotle’s use of the terms entelecheia and energeia in characterizing change. But the extent to which the ‘actuality’ reading has even this advantage is not as straightforward as one might think. For although entelecheia does not elsewhere mean ‘activity’, and although, in some contexts, energeia has a meaning that either is or approaches ‘actuality’, still, I shall argue, Aristotle’s remarks about the term energeia tell against its having such a meaning here. I first offer a brief overview of Aristotle’s use of the two terms in Physics . – and Metaphysics Κ , then point out a passage in Physics .  that is difficult to square with the ‘actuality’ reading, and finally, argue that the only explicit guidelines Aristotle gives us for using energeia with a meaning that approaches ‘actuality’ make it unlikely that it is so used in the definition of change. Within the texts of Physics . – and Metaphysics Κ , entelecheia is not favoured by the number or prominence of its occurrences. While Aristotle’s first statement of the definition in Physics .  and the discussion of the next thirty or so lines use only entelecheia, Aristotle switches to energeia within the argument about the buildable (b), and the latter term predominates thereafter up to the end of Physics . . Metaphysics Κ  first gives the definition in terms of energeia (b), which is thereafter used more often. In fact, Simplicius claims to have found energeia in Physics . , noting only that others claimed to find, apparently as exceptions to the rule, entelecheia in certain manuscripts. At b–, as we saw, Aristotle uses the verb energein in work ‘It is worth noting that Aristotle, defining change, at first said that it is the ἐνέργεια of the changeable qua changeable, but Alexander, Porphyry, Themistius, and others explicating the definition, reading Aristotle a bit later calling it ἐντελέχεια, and having found, in certain manuscripts, the text “the ἐντελέχεια of the potential being, qua potential”, substituted ἐντελέχεια for ἐνέργεια—as being equivalent for Aristotle—in the definition of change’ (Simpl. In Phys . – Diels). His remark, in fact, implies that ἐνέργεια enjoyed better support in the manuscripts available to him, although he does not elaborate, and the currently recorded manuscripts do not support his claim. If ἐνέργεια is the only correct term, we might explain the presence of ἐντελέχεια by citing the fact that Aristotle begins the sentence in which the definition of change occurs with the distinction between being δυνάµει and being ἐντελεχείᾳ (a–). This might lead one to ‘correct’ ἐνέργεια to ἐντελέχεια on the grounds that the definition is meant to draw on the concept of ἐντελέχεια here specified. Such an emendation might seem warranted by the thoughts—all of them still current—that there is a close relation between the two terms ἐνέργεια and ἐντελέχεια, that the standard meaning of ἐνέργεια would invite a circularity charge, and that Aristotle can respond to sceptics about change only by characterizing change as ἐντελέχεια.

Change in Aristotle’s Physics 



ing out the extension of the definition: ‘For each thing admits at one time of energein and at another time not.’ Energein here cannot easily be understood in terms of actuality and cannot be replaced by a cognate of entelecheia. The verb energein here does not mean ‘to be actual’. And Aristotle gives no indication that with this term he is employing a different concept from the one he has been using up to now with entelecheia. In fact, the principle expressed in this line, as we saw, is meant precisely to help us understand when the entelecheia mentioned in the previous line (b–) exists. So, in this passage, the standard meaning of entelecheia will not do. Recall that Aristotle’s early concept of energeia—which I take him to employ here—is that of activity, broadly understood. Since this primary and original use of energeia does not mean ‘actuality’, we must ask how a term that originally means ‘activity’ might take on a quite different connotation of actuality. Energeia is often thought to signify ‘actuality’ (and to approach the standard meaning of entelecheia) in adverbial dative constructions, such as energeiāi on, which generally pick out actual and complete beings. But it would be presumptuous to infer from this usage that the word energeia means ‘actuality’. For we could analyse such constructions as ‘being in activity’, ‘being in virtue of activity’ or ‘being according to activity’. Such constructions might pick out actual beings even if the term energeia itself does not refer to an actual being or an actuality. But energeia in the nominative case sometimes picks out actual beings or their forms (by virtue of which they are actual). Aristotle 

The same clause appears at Metaph. Κ , a In fact, the translation ‘actuality’ for ἐνέργεια (as well as ‘actually’ for the adverbial dative) derives from the Latin actualitas, itself derived from in actu, which translates the adverbial dative ἐνεργείᾳ. But the connection between actus and ἐνέργεια is plausible because actus expresses doing, not because it expresses actuality.  Here I follow Kostman, ‘Definition’, –. See also the next note.  The formulation ‘in activity’ suggests the Thomistic corollary that being is an activity, as in the cases of, for example, being a builder, knower, or perceiver ἐνεργείᾳ. The Thomistic strategy is less straightforward in the cases of, for example, being red, or being a knower ἐνεργείᾳ in the sense of having knowledge. But the alternative formulations, ‘according to activity’ and ‘in virtue of activity’, leave open whether the ἐνέργεια at issue is an activity that constitutes being ἐνεργείᾳ (as e.g. the activity of seeing constitutes being a seer ἐνεργείᾳ) or an activity that brings about the being ἐνεργείᾳ (as e.g. the activity of being built brings about a house). The latter option could then be cashed out in terms of the activity of the being that is now actual (Beere, Doing and Being, –) or (also) in terms of the activity of an external agent (Menn, ‘Origins’, –, ). 



Andreas Anagnostopoulos

indicates, in one of the so-called ‘etymological’ remarks in Metaphysics Θ, that this usage involves an extension of the term energeia from its original one: ‘The term energeia, which is tied together with entelecheia, has been derived from changes [kinēseis] to apply also to other things. For energeia seems most of all to be change [kinēsis]’ (Θ , a–). The ‘derived’ usage enables energeia to apply to other things. There is no indication that the derived usage will also apply to changes, part of the original extension of the term. As Menn points out, there is only one passage in which Aristotle offers any kind of justification for this usage. It implies, I think, that this usage involves an extended sense of energeia that cannot easily be taken to characterize change. This passage appears within a broader argument for the priority in being of energeia on the grounds that ‘energeia is a telos, for whose sake the dunamis is acquired’ (Metaph. Θ , a–). In general, capacities for activities (such as seeing, housebuilding, and contemplating) are acquired for the sake of their exercise, and not the other way around (a–). This implies that in those cases in which ‘the last thing is the use [chrēsis] (for example of sight, seeing, and no other ergon beyond this comes-to-be from sight)’ (a–), the activity (energeia) is the telos. Aristotle wants to extend the connection between energeia and telos (and so the priority of energeia) to the case of substantial form. And so he claims, in the second etymological remark, that ‘the ergon (work, function) is a telos, while the energeia is an ergon. And so even the word energeia is said in accordance with ergon and tends towards entelecheia’ (a–). This remark facilitates the extension of his thesis that energeia is a telos to the case of a product that exists beyond the mere exercise or use of some ability (a–). And even in such a case, the energeia that is the process of becoming is ‘more telos than dunamis’, since it comes-to-be at the same time as the product and takes place in the product (a–). Presumably the change itself is less of  The analysis here of Aristotle’s extension of the term ἐνέργεια is heavily indebted to Menn’s discussion in ‘Origins’, especially at –. One might think that Metaphysics Θ ’s analogical extension of the terms δύναµις and ἐνέργεια in the first half of the chapter is another place where ἐνέργεια is extended from the process of change to the product, or the form of the product. I think that Aristotle commits himself only to the claim that a substance is an ἐνεργείᾳ ὄν, and believe that this can be analysed along the lines suggested in the previous note.  Or alternatively: ‘the ἐνέργεια is more of a τέλος than the δύναµις is’.  ἐνέργεια here refers only to the process, not to the product; for it is counted as a τέλος only in a measured sense and is interchanged with χρῆσις in the next sentence.

Change in Aristotle’s Physics 



a telos than the product. Aristotle concludes that it is ‘evident that the ousia and the form is energeia’ (b–). The only ground Aristotle offers for calling the product an energeia is that it, as well as the process that produces it, is an ergon, and of course ergon can mean ‘work’ both in the sense of ‘task’ and in the sense of ‘finished product’. Add to that the further idea (implicit in the thought that the change’s status as a ‘more of a telos’ derives from its connection to the product) that the product is the primary telos, and Aristotle’s conclusion, that the product-energeia (being the primary telos), like a process-energeia when it is the final telos, is prior to the potential being, follows. Note that while the product is an ergon and thus an energeia in an extended sense, which has to be argued for, the process is assumed uncontroversially to be an energeia in the original use throughout the passage. It seems, then, that (i) energeia in the extended sense has the connotation not of actuality or completeness but rather of ergon (even though finished erga are in fact complete, and all erga are actually existent), (ii) if energeia in the extended sense were somehow to pick out a change, it would have to treat the change as finished product rather than as task-ergon, which is absurd, especially since, (iii) in the current passage, the change is assumed to be an energeia and an ergon in the sense of a task, and the finished ergon the result of such an energeia. Thus, (iv) if the word energeia is used to describe a change, as it is in Physics , it is far more likely to do so because change is an exercise or an activity—an ergon in the sense of function or task rather than (in some peculiar way) in the sense of a product. That is, it is far more likely to be employed in its original sense, or perhaps in a broad sense that spans erga of both kinds. Given that energeia and entelecheia are used interchangeably— and Simplicius indicates that energeia might have been more prevalent in earlier manuscripts than it is now—we must rely on the standard meaning of one term at the relative expense of the other.  One may wonder whether the argument extracted from Θ  is behind every instance of using ἐνέργεια to refer to an actual substance or a substantial form. Again, this is, as far as I know, the only justification for such a use that Aristotle offers. One might alternatively see Aristotle as sloppily using the nominative ἐνέργεια for the dative adverbial construction ἐνεργείᾳ ὄν here, as he may do elsewhere (e.g. Metaph. Θ , b). This would mirror his occasional use of δύναµις where he should use δυνάµει ὄν (potential being). At DA . , a, and . , a–, and Metaph. Η , a, and Θ , b, Aristotle characterizes matter as δύναµις; δύναµις also stands in, I think, for δυνάµει ὄν at Θ , b. Such passages do not show that δύναµις can also mean ‘potential being’.



Andreas Anagnostopoulos

I accept that my interpretation comes at a steep price in so far as entelecheia does not elsewhere mean ‘activity’. I have argued, however, that an interpretation in terms of actuality also comes at a significant price, since it is difficult to read energeia in that way here. And I hope that once the philosophical advantages of my interpretation are taken into account, it will be seen as a relative bargain. Ludwig-Maximilians-Universität, Munich

BIBLIOGRAPHY Ackrill, J. L., ‘Aristotle’s Distinction between Energeia and Kinesis’, in R. Bambrough (ed.), New Essays on Plato and Aristotle (New York, ), –. Beere, J., Doing and Being: An Interpretation of Aristotle’s Metaphysics Theta [Doing and Being] (Oxford, ). Burnyeat, M. F., ‘De anima II ’, Phronesis,  (), –. ‘Kinēsis vs. Energeia: A Much-Read Passage in (but not of) Aristotle’s Metaphysics’ [‘Passage’], Oxford Studies in Ancient Philosophy,  (), –. Charles, D., Aristotle’s Philosophy of Action [Action] (Ithaca, NY, ). Cherniss, H., Aristotle’s Criticism of Presocratic Philosophy (Baltimore, ). Coope, U., ‘Change and its Relation to Actuality and Potentiality’ [‘Change’], in G. Anagnostopoulos (ed.), A Companion to Aristotle (Oxford, ), –. Ebrey, D., ‘Why Aristotle Needs Matter’, unpublished manuscript. Frede, M., ‘Aristotle’s Notion of Potentiality in Metaphysics Θ’, in T. Scaltsas, D. Charles, and M. L. Gill (eds.), Unity, Identity, and Explanation in Aristotle’s Metaphysics (Oxford, ), –. Freeland, C., ‘Aristotle on Bodies, Matter, and Potentiality’, in J. Lennox and A. Gotthelf (eds.), Philosophical Issues in Aristotle’s Biology (Cambridge, ), –. Gill, M. L., Aristotle on Substance: The Paradox of Unity [Paradox] (Princeton, ). ‘Aristotle’s Theory of Causal Action in Physics III. ’ [‘Causal Action’], Phronesis,  (), –. ‘Review of E. Hussey, Aristotle’s Physics, Books III and IV’, Philosophical Review,  (), –. Heinaman, R., ‘Activity, Change and De anima II. ’, Phronesis,  (), –.

Change in Aristotle’s Physics 



‘Is Aristotle’s Definition of Change Circular?’, Apeiron,  (), –. ‘Kosman on Activity and Change’ [‘Kosman’], Oxford Studies in Ancient Philosophy,  (), –. Hussey, E., Aristotle’s Physics, Books III and IV [III & IV] (Oxford, ). Johansen, T. K., Aristotle on the Sense-Organs [Sense-Organs] (Cambridge, ). Kelsey, S., ‘Aristotle Physics I. ’, Phronesis,  (), –. Kosman, L. A., ‘Aristotle’s Definition of Motion’ [‘Motion’], Phronesis,  (), –. ‘Substance, Being, and Energeia’ [‘Substance’], Oxford Studies in Ancient Philosophy,  (), –. Kostman, J., ‘Aristotle’s Definition of Change’ [‘Definition’], History of Philosophy Quarterly,  (), –. Makin, S., Aristotle: Metaphysics Book Theta (Oxford, ). Menn, S., ‘The Origins of Aristotle’s Concept of Ἐνέργεια: Ἐνέργεια and ∆ύναµις’ [‘Origins’], Ancient Philosophy,  (), –. Peck, A. L., ‘Aristotle on Κίνησις’, in J. Anton and G. Kustas (eds.), Essays in Ancient Greek Philosophy (Albany, NY, ), –. Penner, T., ‘Verbs and the Identity of Actions’ [‘Verbs’], in G. Pitcher and O. Wood (eds.), Ryle (Garden City, NY, ), –. Ross, W. D., Aristotle’s Physics [Physics] (Oxford, ). Waterlow, S., Nature, Change, and Agency in Aristotle’s Physics: A Philosophical Study [Nature] (Oxford, ). Witt, C., Ways of Being: Potentiality and Actuality in Aristotle’s Metaphysics (Ithaca, NY, ).

This page intentionally left blank

SEPARABILI T Y   . D I F F E R E N C E : PARTS AN D C A P A C I T I E S O F THE SOU L I N A R I S T O T L E KLAUS CORCIL I U S A N D P A V E L G R E G O R I C

The ancients are divided . . . about the parts of the soul, and in general what is a part and what is a capacity, and wherein their difference lies. (Stobaeus . . a = Porphyry fr.  Smith)

. Introduction I the opening chapter of the De anima Aristotle provides a list of methodological questions that a systematic enquiry into the soul has to address. The question whether the soul has parts or not opens up a series of related questions and problems, and it seems that the way to answer them is to assume that the soul does have parts. However, if the soul has parts, two further methodological questions suggest themselves: () ‘should one first investigate the whole soul or parts of the soul’, and () how should one ‘determine which parts are by their nature different from one another’ (b–). Question () has to do with the status of the whole soul in relation to its parts, whereas question () has to do with the status of parts of the soul in relation to one another. These two questions seem to have differ© Klaus Corcilius and Pavel Gregoric  This paper is a result of the collaboration made possible by the TOPOI Excellenzcluster, Research Group D-III-E-II- (‘Mapping Body and Soul’). We would like to thank our colleagues who commented on earlier drafts of this paper: Andreas Anagnostopoulos, István Bodnár, Pieter Sjoerd Hasper, and Thomas Johansen with whom we had several stimulating exchanges. Thanks to Victor Caston for discussing the main ideas from this paper with us, and to the participants at the TOPOI workshop ‘Part of the Soul and Methodology in Aristotle’ (Berlin, – November ) who made comments on our paper, in particular Jim Lennox and Gábor Bétegh. Last but not least, we are grateful to the Editor, Brad Inwood, who read this paper in very short time and made valuable suggestions.



Klaus Corcilius and Pavel Gregoric

ent underpinnings, the first one dealing with the ‘vertical’ whole– part relation, and the second dealing with the ‘horizontal’ part–part relation. Although the two questions are connected and require answers that cohere well with one another, they seem to be sufficiently distinct to be investigated independently. In this paper we shall investigate how Aristotle answers the second question. The main difficulty for the view that the soul has parts, as we learn from the last sections of DA . , is the concern for the unity of the soul. How can the soul, which is supposed to be the principle of unity, account for the unity of the living being if it itself has parts? This seems to explain Aristotle’s occasional expressions of reservation regarding the talk of parts of the soul (e.g. DA . , b–; . , a–b; Juv. , b–). On the other hand, Aristotle himself often speaks of parts of the soul in putting forward his own views (e.g. DA . , b, ; . , a; PA . , a–b). Moreover, the way he sets out his positive account of the soul in books  and  of the De anima strongly suggests that Aristotle’s answer to the aforementioned methodological question is that the soul does have parts and that he has succeeded in ‘determining which parts are by their nature different from one another’. The standard way of resolving this tension is to say that the soul has parts only in a very special or loose sense. Material objects have parts in the strict sense, parts which can be detached from the whole and which can exist separately from one another and from the whole. Not so with the soul. The soul can be divided only conceptually, namely by distinguishing various capacities of the soul. Each capacity of the soul enables a living being to perform one activity, and thus each has an account which is different from the account of the other capacities. As such, the capacities are merely logical parts or aspects of the soul, which does not imply that they can be detached from the whole so as to exist separately from one another or from the whole. Hence, Aristotle can talk of parts of the soul and yet not be worried about compromising the unity of the soul, for he is talking of parts only in a special sense. On this view, any capacity of the soul can be called a ‘part of the soul’, for it is conceptually distinct from every other capacity of the soul. For example, Richard Sorabji claims that ‘Aristotle’s statement that the most appropriate account of the soul is the one which picks out these capacities, already suggests the thought that perhaps the soul just is these capacities. This thought is confirmed when we

Separability vs. Difference



notice that Aristotle speaks of the capacities as parts of the soul.’ This view seems to be adopted also by Jonathan Barnes: ‘The language of parts need not trouble us: morion is used interchangeably with dunamis and archē, without any substantialist implications.’ More recently, Ronald Polansky writes of ‘the soul’s faculties, that is, its “parts”’. On the standard view, then, every part of the soul is a capacity of the soul, and every capacity of the soul is a part of the soul. There is much to be said in favour of this view. It explains Aristotle’s uneasiness about the idea that the soul has parts, and it intelligibly shows how Aristotle can nevertheless talk about parts of the soul without jeopardizing the soul’s unity. Moreover, this view sits well with passages in which the main capacities of the soul, as they are distinguished in the De anima, are called ‘capacities’ and ‘parts’ interchangeably (e.g. DA . , a–b, b–; . , a–). In spite of its initial plausibility, we think that the standard view fails to do justice to some passages that seem to speak against the identification of capacities with parts of the soul. Let us briefly review two such passages. T(a) At present we must confine ourselves to saying that the soul is the principle of those [i.e. the activities of living beings mentioned in a– b ] and is divided into these, viz. threptikon, aisthētikon, dianoētikon, kinēsis. (b) But whether each one of these is a soul or a part of the soul, and if a part of the soul, whether in such a way that it is separable in account only or also in place, in some cases it is not difficult to see, whereas in others there is a problem. (DA . , b–)

Here Aristotle asks whether each of the capacities he has introduced in the immediately preceding passage (a–b) is a soul or a part of the soul, and if it is a part, what that involves. Then he adds a comment that for some capacities it is easier to answer these ques

R. Sorabji, ‘Body and Soul in Aristotle’, Philosophy,  (), – at . J. Barnes, ‘Aristotle’s Concept of Mind’ [‘Concept’], Proceedings of the Aristotelian Society,  (–), – at .  R. Polansky, Aristotle’s De anima [De anima] (Cambridge, ), . This is also the view advocated in a recent monograph by one of the two co-authors; see P. Gregoric, Aristotle on the Common Sense [Common Sense] (Oxford, ), –. The criticism advanced by the other co-author at the II. Kongress der Gesellschaft für Antike Philosophie in Hamburg in July , by participants of the Berlin Ancient Philosophy Colloquium in November , especially Christof Rapp and Ben Morison, as well as by some reviewers of the monograph, induced him to rethink his position. 



Klaus Corcilius and Pavel Gregoric

tions than for others. This whole passage is rather puzzling on the standard view, not least because it should be equally easy to answer these questions for all capacities, if their conceptual distinctness were sufficient to grant them the status of parts of the soul. The second passage is the following: T For those who divide the parts of the soul, if they divide and separate them according to capacities, they [sc. parts of the soul] become very numerous, viz. the threptikon, aisthētikon, noētikon, bouleutikon, and, further, the orektikon. (DA . , b–)

This passage is no less puzzling on the standard view. If the standard view were correct, there would be no alternative to dividing the soul into parts according to capacities, whereas the passage clearly implies that there is an alternative, and that in this alternative the number of parts of the soul will not be ‘very numerous’ (πάµπολλα), but presumably manageably small. These two passages should suffice to show that Aristotle did make some distinction between capacities and parts of the soul, and that he did not think the relation between the two to be trivial. But what is this relation? Judging from the second passage, it seems true to say that every part of the soul is a capacity of the soul, but not the converse—not all capacities of the soul are also parts of the soul for Aristotle. Furthermore, it seems that both passages establish a close connection between parthood of the soul and some sort of separability, such that being a part of the soul involves being separable in some way. Unfortunately, Aristotle does not tell us explicitly, here or anywhere else, what this distinction is or what sort of separability it involves. Nevertheless, we think that the two quoted passages give us good textual reasons for thinking that Aristotle indeed distinguished between parts and capacities of the soul and that this distinction involves a particular sort of separability. Another reason for taking the distinction between parts and capacities of the soul seriously is of a more philosophical nature. ‘Parts’ and ‘capacities’ seem to be the only pieces of terminology available to Aristotle by means of which he can draw distinctions among different aspects of the soul, and without some distinction among them he would be unable to organize these aspects in a systematic fashion.  The context of the two passages makes it clear that what Aristotle has in mind here is not separability of the soul, or a part of the soul, from the body, but rather separability of one part or capacity of the soul from another; see n.  below.

Separability vs. Difference



The importance of this issue is obvious. DA .  alone mentions nous, the nutritive capacity, the perceptual capacity, the locomotive capacity, imagination, the capacity for desiring, the capacity for feeling pleasure and pain, and the capacity to form opinions as capacities of the soul. Are we supposed to treat all of these capacities on an equal footing? And if so, what about the capacity for remembering, the capacity for recollecting, the capacity for dreaming, and several other capacities mentioned outside of the De anima, notably in the Parva naturalia? Or, indeed, what about the capacity for thumb-twiddling or for letting one’s hair grow, not mentioned by Aristotle? Does Aristotle offer criteria for distinguishing capacities such as the latter ones from other, more fundamental capacities of the soul? And if so, what are these criteria? Without some distinction among psychic capacities, the way the Aristotelian soul is divided, as well as the number of capacities thus reached, seems to be arbitrary. In what follows we shall argue that, in spite of the absence of a text which expressly formulates the distinction between a part and a capacity of the soul, Aristotle did have a clear criterion for deciding which capacity of the soul does and which does not count as a part of the soul, and that the distinction based on this criterion is capable of saving his theory from the charge of arbitrariness. We take a start by discussing an important juncture in the argument of book  of the De anima, where Aristotle turns from the discussion of the ‘most common’ account of the soul to the discussion of the soul’s capacities. At this juncture he takes up the aporia of parts of the soul, originally raised in DA .  and further elaborated in . . In contrast with . , which is largely aporetic, .  sets forth Aristotle’s positive views. Before we turn to DA . , we would like to mention another, relatively recent account of parts of the soul in Aristotle. It can be found in Jennifer Whiting’s paper published in , which deals with what she calls the ‘locomotive part’ of the soul. Because of her focus on locomotive functions, Whiting does not provide a general account of what it is to be a part of the soul, as opposed to being a capacity of the soul, yet such an account is implied in her paper. In our opinion it presents a notable improvement on the standard view, since it (i) explicitly acknowledges a difference between parts and capacities of the soul, (ii) proposes the criteria for distinguishing 

See T above and DA . , a–b, discussed at some length in sect.  below.



Klaus Corcilius and Pavel Gregoric

them, and (iii) does so by using different sorts of separability. However, her interpretation of the relevant sorts of separability and, consequently, her understanding of the criteria for distinguishing parts from capacities are importantly different from ours, as we indicate in a critical discussion of Whiting’s account in the Appendix.

. Parts of the soul in the argument of De anima .  Having provided a very general definition of the soul in DA . , at the beginning of .  Aristotle says that another enquiry is needed, one which will proceed from what is more obvious to what is more intelligible, and which will be genuinely explanatory. From a he makes a fresh start with his enquiry. He begins by listing different ways in which life (τὸ ζῆν) is said, and he does so because life is what the soul is supposed to explain. ‘We take, then, as our startingpoint for discussion, that what has soul differs from what has no soul, in that the former displays life. Now this word has more than one sense, and if any one of the following [viz. activities] is present in a thing we say that it lives’ (DA a–). This is the first systematic step of the enquiry: the soul is established as the explanans of life. In the second step, at a–, Aristotle identifies four different types of life-activity: (i) (ii) (iii) (iv)

thought; perception; local movement and rest; nutrition, decay, and growth.

The presence of any one of these types of activity, Aristotle claims, is sufficient for the ascription of life. Aristotle does not justify this claim, presumably because he thought it obvious enough. Any of the things that people are inclined to call alive—from humble plants to exalted celestial beings—displays at least one of the four listed types of activity (or so Aristotle believed). In the second step, then, Aristotle provides an observation by means of which he establishes  This statement thereby both confirms and widens the scope of the rather tentative statement at the very beginning of the work, where Aristotle says that the soul is ‘like a principle of the living beings’ (DA . , a–).

Separability vs. Difference



differences in the explanandum: to live is to engage in at least one of the four listed types of activity. In the third step the established differences of the explanandum are projected into the explanans; that is, for each of the four identified types of activities, a corresponding capacity of the soul is postulated. This occurs in the first section of T: T(a) At present we must confine ourselves to saying that the soul is the principle of these [activities] and is divided into these, viz. (i′) (ii′) (iii′) (iv′)

the nutritive capacity [to threptikon]; the perceptual capacity [to aisthētikon]; the capacity for thinking [to dianoētikon]; local motion [kinēsis]. (DA b–)

There is only slight variation between the four types of life-activity listed in the second step and the four new items. With the exception of (iv′), local motion, the other three items are substantive adjectives ending in -ikos. Adjectives formed with the suffix -ikos allow for different semantic nuances, but it is generally agreed that in this context they designate that which enables something, in this case the exercise of the aforementioned life-activities. Thus (i′), (ii′), (iii′), and (iv′) are conceived as capacities of the soul. But this, as we have seen, is not yet to say that each of these capacities also figures as a part of the soul. For the passage continues: T(b) But whether each one of these is a soul or a part of the soul, and if a part of the soul, whether in such a way that it is separable in account only or also in place, in some cases it is not difficult to see, whereas in others there is a problem. (DA b–)

The four capacities of the soul, distinguished on the basis of the observation that life comes in four types of activity, are prima facie candidates for being souls or parts of the soul. More precisely, Aristotle raises two distinct questions concerning the four capacities of the soul: (i) are they souls or parts of the soul, and (ii) if they are parts of the soul, are they separable in account only or also in place? Aristotle’s answer to these questions is anything but straightforward, yet we shall argue that it can be extrapolated from the following passage and confirmed elsewhere in the De anima. The passage comes immediately after T(b). T(a) For just as in the case of plants, when divided some are observed 

This exception, we shall suggest later (p. ), is not accidental.



Klaus Corcilius and Pavel Gregoric

to live though separated from one another—showing that in their case the soul of each individual is actually one but potentially many—so we see this happening also with other varieties of soul in the case of insects which have been cut in two; each of the segments has both perception and local movement, and if it has perception, then also imagination and desire; for, where there is perception, there is also pleasure and pain, and where these are, necessarily there is also appetite. (b) We have no evidence as yet about nous or the capacity for theorizing, but it seems to be a different kind of soul, and it alone is capable of being separated as the eternal from the perishable. (c) The other parts of the soul, it is clear from what we have said, are not separable in the way some claim. But that they are different in account is clear—since to be capable of having opinions and to be capable of perceiving are different, if perceiving is different from having opinions—and likewise each of the other aforementioned [capacities of the soul]. (d) Moreover, in some living beings all these [capacities of the soul] are present, in others some of them, and in still others only one (this is what makes a difference among living beings); however, the cause of this must be considered later. Something very similar happens with the senses; some [kinds of animal] have all the senses, others only some of them, and still others only one, the most indispensable, touch. (DA b–a)

In the following two sections of our paper we shall discuss Aristotle’s answer to each of the two questions by interpreting T. But before we do so, we would like to underline the importance of Aristotle’s discussion of these two questions for his treatment of the  All manuscripts have τῶν ζῴων at b and a. Ross in his editio maior,  ad loc., writes: ‘This is a careless statement, for A. undoubtedly ascribes at least the faculty of perception as well as that of nutrition to all animals (a–b)—even to insects (b–). It would be possible to omit τῶν ζῴων (l. ) as a gloss, or to read τῶν ζώντων, in which case the reference would be to plants. But all the MSS. and all the ancient commentators have τῶν ζῴων (though Philoponus points out that it should have been τῶν ζώντων).’ A similar case is found at b, where all manuscripts read τῶν ζῴων, while Ross in both of his editions decides to print Susemihl’s conjecture τῶν ζώντων, with some support from Themistius and Simplicius. Independently of the textual issues, we believe that in this passage Aristotle must have in mind all living beings, including plants, for plants are the class of living beings in which only one of the four capacities is found, namely the nutritive capacity; so Philop. In DA . – Hayduck; Themist. In DA . – Heinze; and Sophon. In DA . – Hayduck. Moreover, Aristotle says in T(d) that an explanation of the distribution of capacities will be supplied later. The last two chapters of the De anima seem to fit this description best, as most commentators agree, and there Aristotle’s discussion does include plants; cf. DA . , a–. Finally, Aristotle opens .  with the following words: ‘Of the aforementioned capacities of the soul all are found in some beings, as we have said, in others some of them, and in some only one’ (a–). T(d) is the most likely target of Aristotle’s back reference, as the commentators correctly observe.

Separability vs. Difference



soul. So far, Aristotle’s division of the soul into the four psychic capacities is based on the observation that life comes in four types of activity. This looks like a promising start, but how can we be sure that these are exhaustive, i.e. that there are no other relevant types of activity? For instance, in Plato’s brief discussion of plants in the Timaeus (  – ), being alive is indeed connected with those four activities, but also with having opinions, reasoning, feeling pleasure and pain, and having appetites. Moreover, how can we be sure that the four types of activity are fundamental, i.e. that none of them can be explained with reference to another? A scientific principle—and the soul, we take it, is the first principle of the science of living beings—would require a stronger foundation; at any rate, what is derived from the observation that life comes in four types of activity should receive independent justification. That is to say, Aristotle must find a way to establish that the four capacities are indeed all there really is to the soul, that they are the fundamental aspects of the soul which cannot be explained with reference to one another or to any other capacity of the soul, whereas all other capacities can be explained with reference to one or several of them. Unless that is adequately established, Aristotle’s programmatic idea that successive accounts of the four capacities constitute a satisfactory treatment of the soul (DA . , a–; . , a–) seems to rest on shaky ground. That Aristotle assumed that he had adequately established that the four capacities are indeed, in some sense, the fundamental aspects of the soul, and that his programme is thus sufficiently justified, is indicated by the fact that the structure of the rest of the De anima is governed by the list of four capacities produced in T(a). We would like to argue that Aristotle established this in the course of his discussion of questions (i) and (ii). The first question concerns the status of the four capacities, and we shall claim that it depends on a sort of separability. The second introduces two possible criteria for determining whether they are parts of the soul, i.e. the  The connection between life and ‘nutrition, decay, and growth’ in this passage of the Timaeus is restricted to attributing life to ‘growing things’ (ϕυτεύουσιν), such as trees, plants, and seeds. However, at   – nourishment, growth, and decay are closely associated with being a ζῷον, and whatever is alive, including plants, Plato is willing to call a ζῷον; cf.   –.  Even the order of the capacities introduced in T(a) is mirrored in the structure of books  and  of the De anima: Aristotle starts his discussion of the nutritive capacity in . , of the perceptual capacity in . , of the thinking capacity in . , and finally of animal locomotion in . .



Klaus Corcilius and Pavel Gregoric

fundamental aspects of the soul. Both of these criteria are spelt out in terms of two types of separability, namely separability in account only and separability also in place. Obviously, the notion of separability will play a central role in what follows.

. Question (i) The first question is whether each of the four aforementioned capacities is a soul or a part of the soul (ψυχὴ ἢ µόριον ψυχῆς). Although this question does not seem to be answered as directly as the second question, the discussion in T contains some telling indications from which Aristotle’s answer can be plausibly inferred. To start with, let us consider what he says about nous, i.e. the capacity for theorizing, in T(b). He says that it can be separated (ἐνδέχεται χωρίζεσθαι). Very generally, to say that something can be separated, or that it is separable (χωριστόν), means that it can have independent existence. This notion is indebted to Plato’s use of the verb χωρίζειν and the adverb χωρίς to express, for instance, that the soul can exist independently of the body, or that Forms exist independently of the particulars that participate in them. Sometimes Aristotle specifies what it is in relation to which something is separable: to say that x is separable from y means that x can exist independently of y. Occasionally he also specifies the respect in which x is separable from y, e.g. ‘in power’, ‘in place’, and ‘in account’, whereby he introduces different types of independence relation. We shall say more about separability as we proceed, but for now it is important to bear in mind that when Aristotle says that x is separable, or separable from y—without specifying the respect in which it is separable—he always means that x can exist independently, or independently of y. By saying that the capacity for theorizing is separable, then, Aristotle means that it can exist independently. Independently of what? Presumably it can exist independently of the body. Aristotle himself says something to that effect at several places, and his remark that this capacity is separable ‘as the eternal from the perishable’  e.g. Phaedo  ,  ; Parm.  ; Arist. Metaph. Μ , b–; cf. D. Morrison, ‘Χωριστός in Aristotle’ [‘Χωριστός’], Harvard Studies in Classical Philology,  (), –, and G. Vlastos, ‘ “Separation” in Plato’, Oxford Studies in Ancient Philosophy,  (), –.  DA . , a–; . , b–; . , a–; . , a–.

Separability vs. Difference



b

( ) supports such a reading. Since the capacity for theorizing is not the form of any body or part of a body, its existence is not tied to the ephemeral existence of any body or part of a body. Unlike all the other capacities of the soul, the capacity for theorizing cannot be defined as the first actuality of a natural instrumental body, and it requires a different account (cf. . , a–). This is indicated by Aristotle’s statement that the capacity for theorizing constitutes a ‘different kind of soul’ (ψυχῆς γένος ἕτερον, . , b). Although separability from body is no doubt in the background of Aristotle’s claim, what seems to be in the foreground is separability from the other capacities of the soul. Let us proceed on the assumption that there is a kind of living being which lives by intellect alone, namely God and perhaps the movers of the celestial spheres. In such beings the capacity for theorizing is indeed found to exist without any other capacity of the soul. Now if that is true, then clearly there is nothing more to the soul of such beings than the capacity for theorizing. Hence, it cannot be a part of the soul of such beings, because it makes sense to call something a part only if it is distinct from a whole. But in this case the capacity for theorizing is not something distinct from a whole; rather it is a whole, namely the whole soul of divine beings. So the capacity for theorizing, at least as far as divine beings are concerned, is a soul, not a part of the soul. This might also be an implication of the statement that the capacity for theorizing constitutes a ‘different kind of soul’. The same applies if one proceeds on the assumption that Aristotle is talking about the theoretical capacity developed by individual human beings, as Philoponus, for instance, would insist. Supposing that one’s capacity for theorizing continues to exist after the demise of the other capacities of the soul upon one’s death, in such a state the capacity for theorizing would clearly exist without  As Barnes, ‘Concept’,  n.  writes: ‘The ends of II  (a–) and of II  (b–a) are superficially parallel: both deal with the chôrismos of psychic parts. But in fact they discuss perfectly distinct topics: II , the separation of psychic parts from body; II  the separation of psychic parts from one another.’ J. Whiting, ‘Locomotive Soul: The Parts of Soul in Aristotle’s Scientific Works’ [‘Locomotive Soul’], Oxford Studies in Ancient Philosophy,  (), – at –, is of the same mind. Morrison’s interpretation of this passage (‘Χωριστός’, ), starting with the claim that ‘χωρίζεσθαι in line  means “to be theoretically distinguished”, or “to be separated in logos”, not “to be separated in space”’, is peculiar.  Cf. DA . , b–; . , a–; Metaph. Λ , b–; NE . , a–b; V. Caston, ‘Aristotle’s Two Intellects: A Modest Proposal’, Phronesis,  (), – at ; M. F. Burnyeat, Aristotle’s Divine Intellect (Milwaukee,  ), –. In DA . –.  Hayduck.



Klaus Corcilius and Pavel Gregoric

the other capacities of the soul with which it used to coexist during one’s life. Hence, the theoretical capacity in such a state would not be called a part of the soul, but a soul. So whichever way we understand Aristotle’s reference to the capacity for theorizing, it is said to be separable, which implies that it can be found to exist without any other capacity of the soul. And if, or when, a capacity of the soul is found to exist without any other, we submit, it should be regarded as a soul, rather than as a part of the soul. Now there is another capacity that Aristotle frequently calls ‘separable’, and that is the nutritive capacity. For instance, at a– he claims that the nutritive capacity ‘can be separated from the other capacities, whereas others cannot be separated from it in mortal living beings, and that is obvious from the case of growing things’. In plants, Aristotle argues, the nutritive capacity is found to exist without any other capacity of the soul, whereas no other kind of mortal living being is found to exist without the nutritive capacity of the soul. All living beings endowed with perception (animals), whether or not they are also in possession of the locomotive and the thinking capacity, have the nutritive capacity of the soul. Of course, the nutritive capacity of the soul is not found to exist in divine beings, and that is the point of restricting the claim in a– to ‘mortal living beings’ (θνητά). At any rate, since the nutritive capacity is found to exist in plants without any other capacity of the soul, it follows that, in plants, it is not a part of the soul, but a soul. Aristotle seems to say so himself: ‘The principle in plants also seems to be a kind of soul [ψυχή τις]; for it is the only one that both animals and plants share, and it is separated from the perceptual principle, whereas nothing has perception without it’ (DA . , b–). Given that Aristotle speaks of the nutritive capacity of the soul as being separable, and at T(b) claims that only (µόνον) the capacity for theorizing can be separated, is he contradicting himself? Not if we read the whole sentence at b–. What Aristotle is saying there is that the capacity for theorizing is the only one separable in the particular way (καθάπερ) the eternal is separable from 

Cf. DA . , b–; . , a–, b–; . , a–; Somn. , a–. So Polansky, De anima, : ‘Surely the nutritive power can itself be a soul since it is all that plants have.’ 

Separability vs. Difference





the perishable. Unlike the capacity for theorizing, the nutritive capacity is the form of a particular kind or part of body, and its existence is tied to the existence of that kind or part of body. Hence, while the capacity for theorizing is found to exist for all eternity without any other capacity in divine beings, the nutritive capacity is found to exist for a limited period of time without any other capacity in plants. So both capacities are separable, and hence capable of existing independently of all the other capacities of the soul; only the capacity for theorizing is such in virtue of being a form that is not embodied, whereas the nutritive capacity is such in virtue of being the form of a kind of body that is not equipped for sustaining any other capacity of the soul, namely the body of a plant. When it comes to the nutritive capacity and the capacity for theorizing, then, the answer to the first question is that, in some cases at least, they are souls rather than parts of the soul; the nutritive capacity is a soul in the case of plants, and the capacity for theorizing is a soul in the case of divine beings. This does not imply, however, that these two capacities therefore cannot also be regarded as parts of the soul. For instance, the nutritive capacity of the soul is found to exist in animals together with the perceptual capacity at the very least, if not also with the locomotive and the thinking capacity. In all such cases there is more to being a soul than just having the capacity to nourish oneself. Here the capacity to nourish oneself is one of the several capacities of the soul, and hence it is reasonable to call it a part of the soul. Perhaps, although this may be more controversial, the same applies to the capacity for theorizing. In those human beings who develop the capacity for theorizing, or while they exercise it, the capacity for theorizing seems to exist together with all the other capacities of the soul, and hence it can be regarded as a part of the soul. Let us briefly add that a capacity which is found to exist without any other capacity in some cases eo ipso counts as a part of the soul in other cases in which it is found to exist together with one or more capacities. Although in these other cases it may not be separable otherwise, surely it remains separable in account, and that is sufficient, as we shall argue in the next section, to regard it as a part of the soul. In any case, the point we wish to make here is that  The same use of καθάπερ, introducing a specification, can be found one line down, at b; cf. n. .  If C is a capacity that is found to exist without any other capacity, C must have



Klaus Corcilius and Pavel Gregoric

being a soul and being a part of the soul are not mutually exclusive categories. A capacity of the soul can be said to be both a part of the soul and a whole soul, depending on whether it is found to exist with or without other capacities, as the case may be in various genera of living beings. So what is Aristotle’s answer to the first question? According to our interpretation, it is the following: of the four capacities listed in T(a), the nutritive capacity and the capacity for thinking (or a particular aspect of the capacity for thinking) are souls. They are souls in the case of living beings with no other capacities than to nourish themselves (plants) or to theorize (divine beings). The same two capacities are also parts of the soul in the cases where they coexist with one or several other capacities of the soul (in all animals and in theoretically minded humans, respectively). Before we turn to the second question, we would like to make some further observations. We have seen that Aristotle characterizes the nutritive capacity and the capacity for theorizing as separable. The nutritive capacity is separable because in plants it is found to exist without any other capacity of the soul. The capacity for theorizing is separable because in divine beings it is likewise found to exist without any other capacity of the soul. By contrast, Aristotle never says that the perceptual and the locomotive capacities of the soul are separable, at least not without specifying the respect in which they are separable. In Somn. , a–, he says, rather cautiously, that plants ‘do not have the perceptual part [of the soul], whether it is separable or inseparable, although it is separable in power and in being’. So Aristotle is happy to say that the perceptual capacity is separable ‘in power’ (δυνάµει) and ‘in being’ (τῷ εἶναι), but he is reluctant to call it just separable, separable simpliciter. Similarly with the locomotive capacity of the soul. When Aristotle comes to deal with it in . , one of the opening questions an account which makes reference to no other capacity; otherwise, i.e. if C had an account such that it must make reference to another capacity D, then admittedly C would not be able to exist without D. The fact that there may also be cases in which C is found to exist together with D makes no difference to C’s account; C remains separable in account even though in such cases it may no longer be separable from D in other ways. In short, C’s separability simpliciter in one case is sufficient to guarantee its separability in account in all cases. We owe the kernel of this point to Andreas Anagnostopoulos.  So Aquinas, In DA § Pirotta (trans. Foster and Humphries, p. ); P. Siwek (ed.), Aristotelis tractatus De anima, nd edn. (Rome, ),  ad b; H. Seidl (ed.), Aristoteles: Über die Seele (Hamburg, ), .

Separability vs. Difference



is ‘whether it is some one part of the soul that is separable either in magnitude [µεγέθει] or in account [τῷ λόγῳ]’ (a–). Again, he considers the possibility that the locomotive capacity is a part of the soul, whether separable in magnitude or in account, but we have no evidence whatsoever that he is ever willing to consider the possibility that it is separable simpliciter. We can infer from this that separability simpliciter, i.e. without specifying the respect in which something is separable, is applicable only to those cases in which something is existentially independent of any other item in the relevant domain. We have seen that the nutritive capacity and the capacity for theorizing can be found to exist each without any other capacity of the soul, and that is why they qualify for being separable simpliciter. However, the perceptual and locomotive capacities are not found to exist without at least one other capacity of the soul. The perceptual capacity is never found to exist without the nutritive capacity, and the locomotive capacity is never found to exist without both the nutritive and the perceptual capacity. Therefore, Aristotle is unwilling to call them separable simpliciter. Now the fact that the perceptual and locomotive capacities are not separable simpliciter, i.e. are not existentially independent of any other capacity, does not imply that they cannot be found to exist without some other capacity of the soul. The perceptual capacity can be found to exist independently of both the locomotive and the thinking capacity (in humble sessile beasts), or only of the thinking capacity (in more developed mobile beasts). Similarly, the locomotive capacity can be found to exist independently of the thinking capacity (in more developed mobile beasts). So, each one of the four capacities is existentially independent of at least one other capacity. This is important because Aristotle makes much of the existential independency relations among the capacities in T(d) and in some other passages, as we shall see in Section .

. Question (ii) The second question for the four capacities runs: if they are parts of the soul, are they separable in account only or also in place (χωριστὸν λόγῳ µόνον ἢ καὶ τόπῳ)? Let us start with observations about two features of the formulation of the second question. First, the ques-



Klaus Corcilius and Pavel Gregoric

tion has a hypothetical form: if the four capacities of the soul listed in T(a) are parts of the soul . . . We would explain this feature with reference both to something that we have argued in the preceding section and to something that we shall argue in the present section. In the preceding section we have argued that being a soul and being a part of the soul are not mutually exclusive categories. So the hypothetical form of the question allows for the possibility that some of the four capacities, at least in some cases, may not be parts of the soul but souls. In this section we shall argue that one of the four listed capacities does not satisfy the criterion for being a part of the soul, so it would be important to formulate the question so as to allow for the possibility that some of the four capacities are neither parts of the soul nor souls, but mere capacities of the soul. Second, Aristotle seems to presuppose that there is an intrinsic connection between being a part of soul on the one hand and being separable on the other. He does not consider the possibility of an inseparable part of the soul, and the way he formulates the question strongly suggests that being a part of the soul implies either being separable in account only or being separable in account and in place, without there being a third option. In other words, question (ii) has the form of an exhaustive disjunction which can serve as a major premiss of a disjunctive syllogism. Let us first say something about the types of separability that Aristotle introduces here. We take it that ‘x is separable from y in place [τόπῳ]’ means that x can have a location independent of the location of y, i.e. x can be found at a place at which y is not found. Another of Aristotle’s ways of expressing the same relation is by saying that x is separate from y ‘in magnitude’ (µεγέθει). On the other hand, when Aristotle claims that x is separable from y ‘in account’, what he means is that the account of x is independent of y, i.e. there is an adequate definition of x which makes no reference to y. The same relation is sometimes expressed by saying that x is separable from y ‘in being’ (τῷ εἶναι): namely, what it is to be x is  The same exhaustive disjunction seems to be found at . , a–: ‘either being separable or not being separable in magnitude but in account’ (εἴτε χωριστοῦ ὄντος εἴτε µὴ χωριστοῦ κατὰ µέγεθος ἀλλὰ κατὰ λόγον; this is the text printed in Ross’s editio maior, which follows MS E; other editors print καὶ before µὴ).  See DA . , a; . , a–; . , b–; cf. R. D. Hicks (ed.), Aristotle: De anima [De anima] (Cambridge, ),  ad a: ‘κατὰ µέγεθος expresses the same meaning as κατὰ τόπον, spatially, locally, as one physical thing and its accidents are separate from another. Either phrase or both can be opposed to λόγῳ or κατὰ λόγον.’

Separability vs. Difference



independent of what it is to be y, so that x can be adequately defined without y. With these tools in hand, we can turn to the text. In T(a) and (c) Aristotle is making a case against the view that parts of the soul are separable in place. He introduces an empirical observation that he has already submitted at the end of DA .  against those who claim that the soul has parts, and presumably Aristotle’s target here is the same: the doctrine that each part of the soul is located in a different part of the body, as propounded by Plato in the Timaeus. The observation starts with the familiar case of plants that can be divided so that each segment continues to live at a different place, as in grafting. This shows that the soul in each plant ‘is actually one but potentially many’: that is, the soul of each plant can be replicated or, as we might say, ‘cloned’. Then Aristotle observes that this phenomenon has a parallel in the animal world too. Some insects can be divided so that each segment carries on living. More to the point, each segment keeps all the capacities of the soul that the insect had prior to being divided. Presumably Aristotle observed that segments of divided insects wiggle when poked, which was sufficient for him to ascribe them perception and locomotion. And having perception and locomotion seemed sufficient for ascribing to them also imagination, pleasure and pain, and appetite. Now, if each segment of the divided insect preserves all capacities of the soul that the insect had prior to being divided, this shows that all we can ever get at two separate places are two souls preserving all their original capacities, never two different capacities of the soul. Hence, the capacities of the soul are not separable from one another in place. If the soul really had parts localized in different parts of the body, it should be possible to separate parts of the soul by di In the Timaeus ( ;  – ) Plato argues that the rational part of the soul is localized in the head, the spirited in the chest, and the appetitive in the abdomen. Although we find the same parts of the soul mentioned in Plato’s other dialogues, their localization is found only in the Timaeus.  This is one of Aristotle’s favourite empirical observations: DA . , a– ; . , b–; Long. , a–; Juv. , a–b, –; Resp. , a– ; IA , a–b; HA . , b–a; PA . , b–; . , b–; Metaph. Ζ , b–. For discussions of Aristotle’s observation, see R. K. Sprague, ‘Aristotle and Divided Insects’, Méthexis,  (), –; D. Lefebvre, ‘L’argument du sectionnement des vivants dans les Parva naturalia: le cas des insectes’, Revue de philosophie ancienne,  (), –; and A. P. Bos, ‘Aristotle on Dissection of Plants and Animals and his Concept of the Instrumental Soul–Body’, Ancient Philosophy,  (), –.



Klaus Corcilius and Pavel Gregoric

viding the body. However, that is not how things work. With this observation, then, the second disjunct in question (ii) is eliminated. One would expect Aristotle to confirm the first disjunct in question (ii) and thus conclude that the four capacities listed in T(a), with the exception of the capacity for theorizing, are separable in account only, but that is not what we get in T(c). Let us look at the whole section again. T(c) The other parts of the soul, it is clear from what we have said, are not separable in the way some claim. But that they are different in account is clear—since to be capable of having opinions and to be capable of perceiving are different, if perceiving is different from having opinions—and likewise each of the other aforementioned [capacities of the soul]. (b–)

On the standard view, this passage contains Aristotle’s solution to the aporia concerning parts of the soul. Having shown that the four capacities, with the exception of the capacity for theorizing, are not separable—either in place or in account only—Aristotle concludes that they are different in account. Talk of parts of the soul thus collapses into talk of capacities, as it turns out that capacities are merely different, not separable, in account from each other. On this view, then, the question of parts of the soul is settled by the end of DA . . However, this cannot be the correct view. If the question of parts of the soul were indeed settled in DA . , why would it be reopened in . ? Moreover, if all capacities other than the capacity for theorizing are inseparable from each other, how should one interpret all those passages in which Aristotle explicitly says that the nutritive capacity is separable (DA . , b–; . , a–, b–; . , a–; Somn. , a–), or that the perceptual capacity is separable (in some respects; cf. Somn. , a–)? More specifically, what is it in Aristotle’s discussion in T(a) and (b) that justifies the alleged conclusion in (c) that the four capacities of the soul are different in account? The argument from dissection of insects in T(a) shows only this much: that the four capacities of the soul, with the exception of the capacity for theorizing indicated in T(b), are not separable from one another in place. This eliminates  With a majority of ancient and modern commentators, we take this to be the gist of the sentence ‘The other parts of the soul, it is clear from what we have said, are not separable in the way some claim’ at b–; that is, the other parts of the soul are not separable in place, as Plato claims in the Timaeus. Gábor Bétegh suggested in discussion that the sentence could be read differently, to the effect that the other parts of the soul are not separable from the body. This would pick up the sort

Separability vs. Difference



only the second disjunct in question (ii), and from that we are not allowed to conclude that the four capacities of the soul are therefore different in account. The only conclusion logic permits us to draw is that the first disjunct is true: namely, if the four capacities are parts of the soul, they are separable in account only. There seem to be two possible sources of the puzzling inference that the commentators are willing to draw from Aristotle’s disjunction, namely that the four capacities of the soul are merely different in account. First, one might think that separability in account is ruled out along with separability in place. The only ground for thinking so is that Aristotle does not specify the type of separability in the opening sentence of T(c). True, he does not say in b– : ‘The other parts of the soul, it is clear from what we have said, are not separable in place, as some claim.’ One might surmise that Aristotle did not specify the type of separability because he wanted to cover both types, separability in place as well as separability in account only. However, nothing that Aristotle says in the preceding text eliminates the possibility that the capacities of the soul are separable in account only. The fact that each segment of the divided insect has the perceptual as well as the locomotive capacity of the soul certainly shows that the perceptual capacity cannot be removed from the locomotive capacity, but it does not show that the perceptual capacity cannot be adequately defined without the locomotive capacity. As a matter of fact, the perceptual capacity is adequately defined without referring to the locomotive or any other capacity in DA . , as we shall see in Section , which means that it is separable in account. of separability mentioned two lines up with reference to nous, and some such view could be plausibly attributed to Plato, for instance in the Phaedo. However, such a reading does not fit the context, since the preceding and the immediately following text talks about separability and difference of parts or capacities of the soul in relation to one another; cf. nn.  and .  e.g. Hicks (ed.), De anima, : ‘While we deny that the other faculties can have separate existence, we at the same time fully maintain that each of them is logically distinct and separable in thought’; and ‘this [i.e. difference in account] was the alternative to spatial or local distinctness set forth in  b ’ (ibid.  ad b); Ross in his editio maior,  ad b–a: ‘He . . . maintains that the other faculties of the soul, though distinguishable, are not separable from each other’; Polansky, De anima, : ‘That the functional parts of the soul differ in account, while not being separate in other ways, is manifest.’  Actually, we would claim that καθάπερ τινές ϕασιν serves as a specification; cf. nn. , , and . However, since this is not an unambiguous reference to Plato’s Timaeus, we shall leave this point aside.



Klaus Corcilius and Pavel Gregoric

Second, the puzzling inference from Aristotle’s disjunction may rest on the confusion of separability in account with difference in account. Let us explain why it is necessary to keep these two notions apart. Aristotle is often found to say that x and y are the same in number but ‘different in being’ (ἕτερον τῷ εἶναι) or ‘different in account’ (ἕτερον τῷ λόγῳ). What he means is that x and y are two distinct items or features which may coincide in the same subject. For instance, one and the same person can be both a doctor and a patient, but what it is to be a doctor is different from what it is to be a patient, i.e. doctor and patient have different accounts. However, their accounts are not separable from one another, since a doctor is a person who treats a patient, and patient is a person treated by a doctor. So two things can be different in account without being separable in account. To put it more generally, two things that are different in account can be (i) inseparable from one another in account (being a doctor and being a patient), (ii) separable from one another in account (being a doctor and being a musician), or indeed (iii) such that one is separable from the other in account but not vice versa, e.g. being a doctor and being a member of a professional medical association; for the account of a doctor does not include any reference to being a member of a professional medical association, whereas the account of a member of a professional medical association must make a reference to being a doctor. Now Aristotle does not do us the favour of explicitly drawing the conclusion from the elimination of the second disjunct in question (ii). Instead of saying something to the effect that, since the four capacities of the soul (apart from the capacity for theorizing) are not separable in place, therefore they are separable in account only, in T(c) he appends the disappointingly weak claim that the aforementioned capacities of the soul are different in account (λόγῳ ἕτερα, b). No doubt this is the key evidence for the standard view. However, it is perfectly possible to interpret the sentence in b (‘But that they are different in account is clear’) in a way that avoids attributing to Aristotle a puzzling piece of reasoning which contradicts other passages in the corpus and renders his division of  It is important to observe that separability, unlike difference, is not a necessarily symmetrical relation. If a is separable from b, it is not necessary that b is also separable from a. For instance, the nutritive capacity is separable from the perceptual, whereas the perceptual capacity is not separable from the nutritive. By contrast, if a is different from b, it is necessary that b is also different from a. Cf. Whiting, ‘Locomotive Soul’, –.

Separability vs. Difference



the soul arbitrary. We propose, therefore, to read the sentence in b not as a conclusion of the preceding argumentation, but as a manœuvre to keep the question open: by stating that parts of the soul are merely different in account, he is stating much less than expected, leaving it open whether the capacities mentioned in T(a) are in fact parts of the soul or not. Of course, one wonders what Aristotle’s motivation is for keeping the question open. Why does he not state the conclusion of his disjunctive syllogism loud and clear here, and thus settle the issue once and for all? It is hard to say anything with absolute certainty on this point, but the following two considerations suggest themselves. First, the conclusion that the four listed capacities of the soul, if they are parts of the soul, are separable in account only would be somewhat idle in the absence of adequate definitions of these four capacities. Only with such definitions at hand can we see whether these capacities are indeed separable from one another in account, or which capacities have accounts that depend on which other capacities. However, we cannot expect to have such definitions before the necessary work has been done. It is no accident, therefore, that the question of parts of the soul, their separability in place or in account only, reappears towards the end of the De anima, once the definitions of the first three capacities listed in T(a) are provided and it remains to find a definition of the fourth one, the locomotive capacity (DA . , a ff.). Second, there is one item in the list of the four prima facie candidates for parts of the soul which does not satisfy the criterion of parthood—that is, separability in account—and yet Aristotle wants to keep it in play. The item in question is the locomotive capacity of the soul. That the locomotive capacity is not separable in account is clear from the fact that, unlike the other three items, it is nowhere defined in the De anima or elsewhere. Also, unlike the other capacities mentioned in T(a), there is no correlated object of locomotion (a supposed kinēton kata topon). And all attempts at defining it in DA . – end up referring to other capacities, such  Observe that the grammatical structure and the word order of the sentence in b (δέ) speak against taking this claim as a conclusion of what precedes.  In the case of the other three capacities mentioned in T(a) there are such correlates, namely τροϕή, αἰσθητόν, and νοητόν. For further arguments against considering the locomotive capacity to be a part of the soul, see K. Corcilius, Streben und Bewegen: Aristoteles’ Theorie der animalischen Ortsbewegung [Streben und Bewegen] (Berlin and New York, ), –, .



Klaus Corcilius and Pavel Gregoric

as the capacity to have desires, imagination, nous, and at a minimum to the perceptual capacity. So it is clear that the locomotive capacity is not separable in account from the other capacities, since all attempts at stating its account make reference to at least one other capacity. However, saying so at this point would effectively rule out the locomotive capacity at a very early stage in Aristotle’s positive account of the soul. And this would significantly reduce the prima facie plausibility of Aristotle’s treatment, since—as he himself notes twice in the De anima—locomotion is arguably one of the two most prominent features of living beings (DA . , b ff., and . , a–; cf. . , a–). Hence, Aristotle avoids unnecessary complications by refusing to conclude from his disjunctive syllogism that the four capacities of the soul, if they are parts, are separable in account only. In any case, and in addition to the points just mentioned, it is sufficient for our concerns that there is nothing in the text that forces us to think that Aristotle is committed to the highly problematic view according to which difference in account is sufficient for being a part of the soul. So we propose to take Aristotle’s disjunction in the formulation of question (ii) seriously and to draw the only possible conclusion from it: () if the four capacities listed in T(a) are parts of the soul, then either they are separable in account only, or also in place; () the four capacities are not separable in place; () therefore, if the four capacities listed in T(a) are parts of the soul, they are separable in account only. What clearly follows from the conclusion is that the criterion for parthood is separability in account: a capacity of the soul counts as a part of the soul if and only if it is separable in account, i.e. if it has an account or definition which makes reference to no other capacity of the soul. But instead of providing a statement of the criterion, Aristotle remarks that the four listed capacities are different in account. We have argued that this is a manœuvre to keep the locomotive capacity on the table and not to decide on the issue prematurely, 

Cf. Corcilius, Streben und Bewegen, –. We have seen that this applies, strictly speaking, only in the case of living beings whose souls feature at least two of the four capacities. In the case of living beings whose souls are exhausted by the capacity for theorizing (divine beings) or by the nutritive capacity (plants), these two capacities are separable simpliciter. 

Separability vs. Difference



since fully answering the question requires definitions of all of the four capacities, which is what the bulk of the rest of the De anima is about. Before we proceed with our interpretation of Aristotle’s answer to question (ii), it remains to comment on his execution of the manœuvre in T(c). Instead of drawing the correct conclusion from his disjunctive syllogism—namely that the four listed capacities, if they are parts of the soul, are separable in account only—Aristotle appends the observation that they are clearly different in account. The example he uses to illustrate this observation is the relation between the capacity to have opinions and the perceptual capacity. What it is to be the capacity to have opinions is different from what it is to be the capacity to perceive, since perceiving is different from having opinions. The same, Aristotle adds, goes for the other aforementioned capacities. An attentive reader will pause at the question why Aristotle introduces the capacity to have opinions (τὸ δοξαστικόν) in addition to the four listed capacities, when he could make the same point just as well with any two of the four capacities in T(a). And apart from such concerns for economy, why the capacity to have opinions—of all the other capacities of the soul? If he must introduce a fifth one, for whatever reason, why does he not pick one of the capacities mentioned a little earlier, in T(a), such as imagination, pleasure and pain, or appetite? We cannot be sure, of course, but perhaps Aristotle wanted to indicate something. Remember that both Plato and Aristotle conceive of the capacity to have opinions as something different from, but dependent on, the perceptual capacity. The two capacities have different accounts, yet the account of the capacity to have opinions must make a reference to the perceptual capacity, or at any rate to the objects of the perceptual capacity, but not vice versa. The same situation, we have argued, holds between the locomotive capacity and the perceptual capacity, and that is exactly why Aristotle switches from separation in account to difference in account. So the example may be regarded as a tacit admission of the manœuvre and a hint as to why it is done. . Distribution of the four capacities of the soul After the manœuvre, Aristotle adds an observation concerning the distribution of the capacities of the soul mentioned in T(a) among living beings:



Klaus Corcilius and Pavel Gregoric

T(d) Moreover, in some living beings all these [capacities of the soul] are present, in others some of them, and in still others only one (this is what makes a difference among living beings); however, the cause of this must be considered later. Something very similar happens with the senses; some [kinds of animal] have all the senses, others only some of them, and still others only one, the most indispensable, touch. (b–a)

Before we explain the purpose of this observation, let us first say something about its content. The observation points at the fact that the distribution of capacities of the soul among living beings is arranged in an ordered series. In souls which contain a plurality of capacities, each ‘higher’ capacity is found to exist together with all the lower ones: in plants only the nutritive capacity is found, in animals the perceptual capacity is found in addition to the nutritive, in more developed mobile animals the locomotive capacity is found in addition to the perceptual and the nutritive, and in human beings the capacity for thinking is found in addition to the locomotive, perceptual, and nutritive. Aristotle does not explicate this series in so many words, but he clearly has it in mind here, since he claims that the senses are distributed in a similar fashion. Namely, in some animals only touch is found, in others a few other senses are found in addition to touch, and in the most developed animals all five senses are found. The ordered series of the capacities and the senses is restated in a more elaborate way at . , b–a. Now why does Aristotle mention this here? The ‘moreover’ (ἔτι δ’) in b makes it clear that he intends to introduce a further point in the sequence of arguments concerning question (ii). It seems to contain another reason for thinking that the capacities of the soul mentioned in T(a) are not separable in place. Except for the nutritive capacity, which is found to exist apart from all the other capacities in plants, none of the other capacities is found to exist without at least one other capacity: the perceptual capacity is never found to exist apart from the nutritive, the locomotive apart from the perceptual and the nutritive, and the capacity for thinking (in the sublunary sphere, at any rate) apart from the locomotive, the perceptual, and the nutritive capacity. Aristotle’s observation in T(d) certainly testifies to the local inseparability of the four capacities of the soul, but it does more than  Apart from the nutritive capacity in the case of plants, the other exception is the capacity for theorizing in the case of divine beings. In these cases, as we have seen, the two capacities should be considered souls, not parts of the soul.

Separability vs. Difference



that—it suggests that the soul is naturally divided into these four capacities, namely in so far as they are distributed among living beings by way of ordered series. This distribution, Aristotle maintains, sorts out the main genera of living beings, which allows Aristotle to assume, as he does in the second half of DA .  (b– a), that providing successive accounts of the four capacities amounts to a satisfactory account of the soul, an account which is explanatory of the souls that actually exist in the world, namely the souls of plants, animals (sessile and mobile), and humans. The Aristotelian natural philosopher’s treatment of the soul is thus sharply contrasted with what seems to have been the prevailing approach at the time, restricted to studying the human soul only (DA . , b–). So the observation in T(d) brings a sort of empirical confirmation of the tentative list of capacities adduced in T(a), since it is the distribution of precisely these four capacities that sorts out living beings in the main genera. As Aristotle himself says in the parenthetical remark at b–a: ‘This is what differentiates living beings.’ So the four prima facie capacities of the soul are operative in differentiating the main genera of living beings, which means that Aristotle’s choice of the four capacities listed in T(a) has strong vindication in nature, and that justifies the whole programme of the De anima.

. The criteria of parthood at work in De anima .  Having provided a systematic account of the first three capacities of the soul from T(a), Aristotle turns to the last one, the locomotive capacity, towards the end of the treatise, in DA . –. This is where the question of parts of the soul and their separability is taken up again. We have argued that this question was not settled earlier (although the right answer was in the air) because Aristotle hesitated to call the locomotive capacity a part of the soul, and yet he did not want to eliminate it from the list of topics that needed a systematic discussion in order to provide an appropriate treatment of the soul. This hesitation concerning the locomotive capacity is justified in DA . –. Each one of the several attempts at defining the locomotive capacity in the course of the argumentation in DA 

Cf. n.  above.



Klaus Corcilius and Pavel Gregoric

. – ends up making reference to some of the psychic capacities mentioned previously in the De anima. Since it is not separable in account, the locomotive capacity cannot be a part of the soul. It must be something other than a part of the soul then, for instance a capacity of the animal which is to be explained with reference to one or several genuine parts of the soul. We cannot go into the details of Aristotle’s treatment of the locomotive capacity here, but we would like to discuss the passage in which the question of parts of the soul is reopened in . . Here is what Aristotle says: T(a) Since the soul of animals is marked by two capacities, by the discriminative, which is the work of thought and perception, and by moving by way of local motion, let this much suffice about perception and thought, and we should now enquire about the moving [capacity], what of the soul it is, (b) whether it is some one part of the soul separable either in magnitude or in account, or the whole soul; and if it is a part, [we should enquire] whether it is a special part in addition to those usually distinguished or mentioned by us, or one of them. (c) The problem at once presents itself, in what sense we should speak of parts of the soul and how many [we should distinguish]. For in a sense there seem to be indefinitely many parts, and not only those that some people mention when they distinguish the rational, the spirited, and the appetitive, or with others the rational and the non-rational; for if we take the dividing-lines by which they separate into these, we shall find parts far more distinctly separated from one another than these, namely those we have just treated: the nutritive, which belongs to plants as well as to all animals, and the perceptual, which cannot easily be classed as either non-rational or rational; furthermore, the imaginative, which is different in being from all the others, while it presents a great problem regarding which of those it is the same or not the same as—if one posits separate parts of the soul; and in addition to these the desiderative, which would seem to be different both in account and in power from all the others. (a–b)

In section (b) Aristotle uses more or less the same formulation as in T(b). With this he clearly addresses the issue he has discussed earlier, in T and T, namely the question of parts of the soul. He  For an account of Aristotle’s theory of animal motion and the locomotive capacity of the soul see Corcilius, Streben und Bewegen.  There are some notable dissimilarities between T(b) and T(b). First, in T(b) the locomotive capacity is said to be either a part of the soul (and as such being separable either in place or in account) or the whole soul; the latter alternative, namely a capacity being accounted for in terms of the whole soul, is not mentioned in T(b). Second, the option that a capacity of the soul (namely the locomotive capacity) is identical with some of the other capacities or parts of the soul does not occur in T(b).

Separability vs. Difference



asks whether the locomotive capacity is separable in magnitude (or place) or separable in account. Again, difference in account—which was introduced instead of, and often mistaken for, a conclusion of the disjunctive syllogism in T(c)—is not even mentioned. In talking about parts of the soul here, Aristotle allows only for their separability, namely either separability in magnitude or separability in account. Given what we have argued so far, this is only as it should be. However, here we are mostly interested in section T(c), in which Aristotle puts forward his critique of the ways of partitioning the soul proposed by other thinkers, presumably Plato and his followers. His critique is obviously methodological in character. Here Aristotle is not interested so much in which parts of the soul earlier thinkers have postulated, but rather how they arrived at these parts, i.e. what criteria they used to divide the soul. This is important because such a methodological critique seems to work only against the backdrop of a different method of dividing the soul, and hence it should be informative regarding Aristotle’s view as to what the right method of dividing the soul is. And although Aristotle again does not tell us straightforwardly what he takes to be the criterion of parthood of the soul, we believe that his critique in DA .  presupposes that the criterion is separability in account. The argument in T(c) has the structure of a reductio. Aristotle, for the sake of the argument, adopts the Platonist criteria for dividing the soul in order to show that these criteria lead to undesirable consequences. He argues that, given the Platonist criteria for dividing the soul, we will end up with an undesirably large number of parts; hence, the Platonist criteria for dividing the soul must be wrong. Well, what is the Platonist method of dividing the soul? According to Aristotle, the criteria employed by the Platonists are differ Aristotle seems to distinguish two groups of earlier thinkers, those who advocate tripartite division of the soul into the rational, the spirited, and the appetitive part, and those who propound bipartite division into the rational and the non-rational part. Whereas the first group certainly includes Plato and his followers, the second group is harder to fix, not least because Aristotle himself accepts the bipartite division in his ethical works; cf. P. A. Vander Waerdt, ‘Aristotle’s Criticism of SoulDivision’, American Journal of Philology,  (), –.  Brad Inwood observed that a similar line of criticism can be found in Aristotle’s Peri ideōn. Given the criteria used for postulating Forms, Aristotle shows that Plato winds up with too many Forms, including Forms of not-beings, indeterminate and indefinite things, relatives, etc. For a detailed analysis of Aristotle’s arguments in Peri ideōn see G. Fine, On Ideas: Aristotle’s Criticism of Plato’s Theory of Forms (Oxford, ).



Klaus Corcilius and Pavel Gregoric

ence in account and in power, both of which he later summarizes as ‘division and separation according to powers’ (ἐὰν κατὰ τὰς δυνάµεις διαιρῶσι καὶ χωρίζωσι, . , b = T). These criteria are clearly not the correct ones for parthood of the soul, or at any rate not the sufficient ones, since Aristotle thinks that their application leads to an unacceptably large number of parts of the soul. It should be clear by now that what he militates against here is the division of the soul into parts according to different capacities of the soul, and that is precisely what Aristotle’s own position is standardly, and mistakenly, taken to be. What Aristotle finds objectionable about ‘division and separation according to powers’, we submit, is the idea that something counts as a part of the soul only because it is a distinct capacity of the soul: if x is different in account and in power, then x is separate (= x is a part of the soul). This is clearly wrong for Aristotle, because difference in account by no means amounts to separability in account, and still less to separability in place, which is what characterizes parts of the soul in Plato’s Timaeus. It follows that, on the standard view of what parts of the soul are in Aristotle, his critique of the Platonist division of the soul in .  is either unintelligible or destructive of Aristotle’s own theory. More importantly, the critique put forward in T confirms our interpretation by showing that Aristotle did not content himself with difference in account as a sufficient criterion for parthood, but that he insisted on some sort of separability. And given that he elsewhere (in T, but also in DA . , b–) rejects the Platonic view according to which parts of the soul are separable in place, it is more than likely that separability in account is the criterion at work in the background of Aristotle’s argumentation in DA . .

. Conclusions In T Aristotle accomplishes two important tasks. First, he gives us good reasons for accepting his division of the soul into the four capacities listed in T(a). Second, he makes sure that we do not conceive of these four capacities as locally separable parts of the soul,  What he has in mind here is, we suggest, not only the famous ‘principle of contraries’ of Rep.  –, but also the metaphysics of dispositions presupposed by the individuation of capacities in terms of their corresponding activities, which we take the formulation ‘difference in power [δυνάµει]’ to refer to; see Rep.  –; cf. Phdr.    ff. See pp. – for a summary of the standard view.

Separability vs. Difference



as they are conceived in Plato’s Timaeus, because such a conception would pose a threat to the unity of the soul. However, none of what Aristotle says against locally separable parts of the soul implies that he abandons the possibility that the four capacities listed in T(a) are definitionally separable parts of the soul. To be sure, Aristotle is reluctant to say explicitly that they are parts of the soul on account of their definitional separability, but his reluctance, as we have explained, is motivated by two methodological concerns. On the one hand, he does not want to abandon the locomotive capacity at an early stage of his treatment of the soul, although its proper treatment will eventually show that it is not really a part of the soul, since it is not definitionally separable: it cannot be defined without making reference to other capacities of the soul, the perceptual capacity at a minimum. On the other hand, concluding expressly that the four capacities are parts of the soul on account of being definitionally separable would immediately call for their definitions, but their definitions can come only at the end of a proper enquiry into each one of them. If we look at the first three capacities of the soul, we shall find that each one of them is defined with reference to their respective objects. The nutritive capacity is what ‘maintains its possessor as such, while food prepares it for activity’ (. , b–). The perceptual capacity is what ‘receives sensible forms without matter’ (. , b–; cf. . , a–). The thinking capacity is more difficult, as Aristotle’s account of this capacity is more sketchy, but it will not be off the mark to say that it is what receives intelligible forms or what grasps essential features (. , a–, b–; . , b–). We can see that each one of these three capacities is separable in account, for each has a definition that makes no reference to any other capacity of the soul. This is precisely what entitles them, in Aristotle’s theory, to be called parts of the soul. This is not to say that having a correlative object, as such, is sufficient for separability in account. For instance, sometimes Aristotle speaks about the object of desire (τὸ ὀρεκτόν), so one might think that the capacity for desire (τὸ ὀρεκτικόν) has an account that makes reference to such an object, without making reference to any other capacity of the soul, and hence one might conclude that the capacity for desire is a part of the soul. However, Aristotle would object to such an account: ‘The object of thought implies that there is thought of it, but the thought is not relative to that of which it is the



Klaus Corcilius and Pavel Gregoric

thought; for we should then have to say the same thing twice. Similarly sight is the sight of something, not of that of which it is the sight (though of course it is true to say this); in fact it is relative to colour or to something else of the sort’ (Metaph. ∆ , a–b). So one would have to spell out what the object of desire is independently of being the object at which desire is directed, and in doing so one would have to say, for instance, that it is something perceived to be good or thought to be good, thus making reference to the capacities for perception and thought. It would turn out, then, that the capacity for desire is not separable in account after all, and hence not a part of the soul. Given that separability in account is Aristotle’s criterion of parthood, if we want to see whether a capacity of the soul is a part of the soul, we need to check whether its definition makes reference to any other capacity of the soul. () If it does, we can conclude that it is not a part of the soul, but a capacity dependent on whatever capacity is mentioned in its definition. For example, imagination is defined as ‘change which comes about as a result of actual perception’ (DA . , a–), so it is inseparable in account from the perceptual capacity of the soul. Memory is defined as the capacity to ‘have an image regarded as a copy of that of which it is an image’ (Mem. , a–), which means that it is inseparable in account from the capacity of imagination, which is itself in turn inseparable in account from the capacity of perception. Similarly, the criterion is not satisfied by the capacities to experience pleasure and pain, to have desires, to recollect, to dream, and indeed to locomote. Although Aristotle does not want to rule out the locomotive capacity at an early stage of his positive account of the soul, as we have argued, he seems to have supplied subtle hints which are supposed to warn the attentive reader (or a returning reader with the benefit of hindsight) that the locomotive capacity is not really on a par with the other three capacities listed in T(a). One hint is found right there in T(a), where Aristotle uses adjectives formed with the suffix -ikos to designate the first three capacities, whereas the fourth capacity is not called κινητικὸν κατὰ τοπόν, the term he uses elsewhere (. , a, b, a), but rather, less technically, κίνησις. The other hint is found at the beginning of . , where Aristotle expands on the conclusion of the preceding chapter, in which it was 

Cf. Whiting, ‘Locomotive Soul’,  n. .

Separability vs. Difference



established that accounts of each one of the four capacities operative in sorting out living beings into main kinds (including the locomotive capacity: see . , a–) constitute an appropriate account of the soul. However, at the start of .  Aristotle simply omits the locomotive capacity: If one has to say what each one of these [capacities] is—e.g. what is the intellective capacity, what is the perceptual, or what is the nutritive capacity— prior to that one has to say what thinking is and what perceiving is, since activities and actions are prior in account. But if this is so, and prior to these one has to consider their correlative objects, first one must determine about them for the same reason, e.g. about nourishment, the sensible object and the intelligible object. (a–)

One might reply that the locomotive capacity is omitted simply because it would not fit the illustration, since it does not have a correlative object in the way the other three capacities do. That might be correct, but if it has no correlative object in the way the other three capacities do, does that not seem to make its definitional separability problematic? () If the definition of a capacity of the soul does not make reference to any other capacity of the soul—as in the case of the nutritive, the perceptual, and the thinking capacity—should one conclude that it is a part of the soul? In principle we would say yes, although we recognize that there may be cases which, on the face of it, suggest otherwise. These cases seem to threaten our view that the criterion of parthood of the soul is separability in account, so let us have a look at them. The first case is that of the special senses. Each special sense is defined with reference to one kind of special sensible, without referring to any other capacity of the soul. Thus vision is defined as the ability to receive colours, hearing is defined as the ability to receive sounds, etc. This suggests that the special senses are separable in account, and hence that they are parts of the soul. If we find this suggestion unpalatable, as we of course do, should we not abandon the view that separability in account is the criterion of parthood of the soul? Not at all. Although Aristotle does treat the special senses as independent capacities in DA . –, which is methodologically a perfectly sound thing to do, in . – he introduces a series of considerations which are supposed to show that the special senses are not really in-



Klaus Corcilius and Pavel Gregoric

dependent capacities, but only distinct aspects of one single thing, namely the perceptual capacity of the soul. That is, questions such as how we perceive the common sensibles, or how we discriminate between objects of different special senses, can only be plausibly answered if we suppose that the perceptual capacity of the soul is really what takes on colours, sounds, and all the other special sensibles. Here is what Aristotle says in De sensu , a–: If, then, the soul perceives sweet with one part and white with another part, then what is made up of these parts is either some one thing, or not. But it must be one; for the perceptual part is one thing. . . . Therefore, there must be some one thing of the soul with which everything is perceived, as has been said before, each kind [of special sensible] through one [of its aspects, i.e. special senses].

This shows that the definition of the perceptual capacity as the ability to take on sensible forms is very general and only provisional, because sensible forms come in five different kinds (for simplicity, we take all tangible qualities as one kind of sensible form), and the activity of receiving different kinds of sensible forms is different in each case, corresponding to seeing, hearing, etc. It is tempting to think that these activities and their corresponding objects belong to different individual senses, but Aristotle warns us that this is a misleading way of thinking. It is the perceptual capacity of the soul that receives all kinds of sensible forms without matter. The five special senses are just names for its ability to receive five different kinds of sensible forms. Hence, our view that separability in account is the criterion of parthood of the soul is not threatened by the case of the special senses. The second case is that of the reproductive capacity (τὸ γεννητικόν), which seems to be defined as the ‘ability to produce another thing such as oneself’ (. , a–). Well, if this is the account of the reproductive capacity, and if it does not make reference to any other capacity of the soul, should we not count it as a part of the soul? Although this would require a more detailed discussion, we would argue that the object of the reproductive capacity is really the same substantial form that the nutritive capacity maintains for  Of course, there is more to this story because the abilities to take on different kinds of special sensibles come in ordered series. Moreover, once a living being has a perceptual capacity with these abilities, and depending on the number and refinement of these abilities, it will also have the ability to grasp features such as common sensibles and perhaps accidental sensibles; cf. Gregoric, Common Sense, –.

Separability vs. Difference



the individual living being by means of taking in and processing food. What the reproductive capacity does is perpetuate this form in another individual (cf. . , b–). Thus we would claim that the account of the reproductive capacity does make reference, if only implicitly, to the nutritive capacity of the soul, so that we should not count it as a part of the soul. Despite some cases which may require more detailed consideration, then, it is plausible to think that if the definition of a capacity of the soul does not make reference to any other capacity of the soul, it is a part of the soul. So the view that separability in account is the criterion of parthood of the soul seems to stand up to scrutiny. To conclude, what are parts of the soul for Aristotle? The resulting view is this. Parts of the soul are the fundamental capacities of the soul whose existence we minimally have to assume in order to be able to provide a satisfactory account of the soul on which the science of living beings will be based. There are other nonfundamental capacities of the soul which living beings may have simply in virtue of possessing the relevant fundamental capacities, or in virtue of possessing the relevant fundamental capacities and satisfying some further conditions, such as having the right sort of bodily organs or enjoying the right sort of environment. The distinction between parts of the soul and mere capacities of the soul, i.e. between the fundamental and non-fundamental capacities of the soul, lies in the fact that definitions of the former make no reference to any other capacity of the soul, whereas definitions of the latter make reference to at least one other capacity of the soul. This distinction not only saves Aristotle from inconsistency and arbitrariness, to which he would fall prey on the standard interpretation of what parts of the soul are in his theory, but allows him to proceed systematically in building a hierarchy of psychic capacities with three fundamental and irreducible capacities at the top. And quite generally, this seems to constitute a very reasonable theoretical position.  For instance, Jerry Fodor in The Modularity of Mind (Cambridge, Mass., ), , is looking for, mutatis mutandis, a similar distinction between what he calls ‘mental capacities’ and ‘psychological faculties’: ‘There are, of necessity, far more mental capacities than there are psychological faculties on even the most inflationary census of the latter. . . . A census of faculties is not, in short, equivalent to an enumeration of the capacities of the mind. What it is instead is a theory of the structure of the causal mechanisms that underlie the mind’s capacities. It is thus perfectly possible for



Klaus Corcilius and Pavel Gregoric APPENDIX

Whiting on Parts of the Soul To start with similarities between Whiting’s and our understanding of what parts of the soul are for Aristotle, Whiting () argues against identifying parts of the soul with capacities of the soul, i.e. against the ‘standard view’. More importantly, she connects parthood of the soul with the notions of difference and separability. However, she does so in a way that is different from ours. In order to make the differences clear, it will be useful to list the types of difference/separability we have encountered so far: (i) difference ‘in account’ (or ‘in being’ or ‘in function’): x has an account or definition which is different from y, and y has an account or definition which is different from x. (ii) separability ‘in account’ (or ‘in being’ or ‘in function’): x is definitionally independent of y, i.e. x has an account or definition which makes no reference to y. (iii) separability ‘in place’ (or ‘in magnitude’): x is locally independent of y, i.e. x can exist at some location without y. (iv) separability simpliciter: x is existentially independent of y, i.e. x can exist without y. We have argued that (iv) is the criterion for deciding the question whether a capacity is a soul or not, whereas (iii) is rejected and (ii) accepted as the criterion for deciding the question whether a capacity is a part of the soul or not. Whiting proposes to distinguish (iv) and (iii) in the following way: For things can be separable in magnitude from one another—or separable in place from one another—without being separable ἁπλῶς: a hand, for example, may be separable both in magnitude and in place from a foot in the sense that each is composed of distinct and non-overlapping bits of matter, even though neither has the kind of self-sufficiency required for separability ἁπλῶς. () Whiting reserves separability simpliciter for the theoretical capacity, and claims separability in place for the other three capacities listed in T(a). That is to say, she argues that parts of the soul are separable in place or all hands to be agreed about what capacities a mind has and still to disagree about what faculties comprise it’ (Fodor’s italics).  In what follows we take the liberty of examining Whiting’s paper only in so far as it contributes towards what we are interested in here, i.e. towards a general account of parts of the soul in Aristotle. None of our criticisms is meant to diminish the scholarly and philosophical merits of this original and illuminating paper.

Separability vs. Difference



magnitude from one another. Indeed, separability in place, as she construes it, seems to be the criterion of parthood of the soul on Whiting’s account: My own view is that we should model Aristotle’s way of distinguishing the various parts of the soul and their fundamental capacities, however many they prove to be, on his way of distinguishing the various parts of an animal’s body and their fundamental capacities, which is a strategy that makes prima facie sense given his hylomorphism. This will allow us to say that one part of soul may house multiple capacities just as the human tongue, for example, houses the capacities both to taste and to utter sounds (PA a–). But more importantly for present purposes, it will allow us to say that just as the various parts of the body cannot exist apart from one another even though they are separable in magnitude or place from one another in the sense that they are constituted by different portions of matter located in different places, so too the various parts of an animals’ soul cannot exist apart from one another even though they are separable in magnitude or place from one another in the sense that they are embodied in what we might call different ‘physiological systems’— i.e. physiological systems involving bodily organs constituted by different portions of matter and/or located in different places. The nutritive and reproductive capacities are embodied in one physiological system (for Aristotle takes what we would call the ‘digestive’ and ‘reproductive’ systems to form a single system), while the capacities of perception, imagination, and desire are embodied in a different physiological system. Each of these physiological systems is centred in one and the same organ (namely the heart), which helps to explain their unity with one another. But each can (at least in some circumstances) function relatively independently of the other. () This passage clearly shows that Whiting understands separability in place differently from us. On our view, x is separable in place from y if x is locally independent of y, i.e. if x can exist at some location without y. On Whiting’s view, by contrast, x is separable in place from y if x is embodied in different portions of matter located at different places. Although this may be in itself an acceptable notion of local separability, we do not think that this is the notion which Aristotle employs. First, we have argued that separability and difference are two quite different relations; notably, separability can be asymmetrical, whereas difference cannot. Hence, separability cannot possibly be reduced to, or interpreted in terms of, difference. Second, we have interpreted separability in terms of independence, and this interpretation seems to be borne out by the Aristotelian texts we have considered. In order to be locally separable, it is not sufficient for parts of the soul or their corresponding physiological systems to occupy different portions of matter, but they have to be capable of existing independently of each other. Thus the



Klaus Corcilius and Pavel Gregoric

hand and the foot may be spatially different from each other, but it is difficult to see how Aristotle could regard them as spatially separable from each other, since both of them are parts of the same living body. As the observations with the divided insects show, what Aristotle seems to have in mind when he talks of separability in place in connection with parts of the soul is local independence rather than mere difference in location in otherwise coherent bodies. Whiting’s construal of Aristotle’s notion of separability in place, however, dissociates separability from independence and associates it with mere difference. Furthermore, the physiological systems in which different parts of the soul are realized happen to be largely overlapping. Apart from the heart, the digestive and the perceptual physiological systems, for example, share the whole network of blood vessels and channels inside the body. So, they are constituted only by partially different portions of matter. In fact, it seems that the portion of matter shared by these two physiological systems is larger than the portions of matter which belong to one without the other. Therefore, even the notion of difference in place, according to Whiting’s construal of separability in place, is a qualified or attenuated one. More generally, we find the suggestion that parts of the soul are separable in place, on any construal of spatial separability, implausible. The idea that the soul is divided into locally separable parts was advocated by Plato in the Timaeus, as Whiting herself agrees ( n. ), and Aristotle seems to find this idea fundamentally objectionable, not least because it undermines the soul’s unity (cf. DA . , b–). Moreover, the observation about divided insects is, as we have argued above, intended to show, against the Platonic division of the soul, that the capacities listed in T(a) as prima facie candidates for parts of the soul are not separable in place or magnitude. However, Whiting seems to argue () that the observation shows that only capacities such as the perceptual and the locomotive, which are explicitly mentioned in the passage on divided insects (. , b–), are not separable in place or magnitude, leaving it open whether the other capacities listed in T(a) are separable in place or magnitude or not. If that is what the observation shows, however, then it is hard to see how Aristotle’s observation about the divided insects in T is supposed to make it clear that ‘the other parts of the soul’—presumably the capacities listed in T(a) other than the capacity for theorizing—‘are not separable in the way some claim’ (b–). Moreover, looking at the shorter of the two passages from Whiting’s paper quoted above, we would like to observe that, although a hand severed from the rest of the body will not be able to exist self-sufficiently, as required by separability ἁπλῶς (at any rate not as a hand properly speaking), the rest of the body, including the foot, will. That is why the hand is not se-

Separability vs. Difference



parable from the body either locally or ἁπλῶς, whereas the rest of the body is separable from the hand both locally and ἁπλῶς. We also have an objection to Whiting’s interpretation of T(a). We have seen that parts of the soul, according to Whiting, are distinguished by separability in place, as she construes it. The role of difference and separability in account, according to Whiting, is to differentiate, within a single part of the soul, more basic from less basic capacities. Her example is what she calls the ‘locomotive part of the soul’. The locomotive part, according to Whiting, houses a number of capacities, namely the perceptive, the imaginative, the desiderative, and the practically intellective capacities of the soul (,  ff.,  ff.). All of these capacities are (i) different in account from one another. However, the perceptive capacity is (ii) separable in account from the other ones, whereas none of the others is separable in account from it: the perceptive capacity is the only capacity which can be defined without reference to the other capacities, whereas none of the other capacities can be defined without reference to the perceptive capacity. The whole complex of capacities constituting one part of the soul is called after this basic capacity within the complex. This enables Whiting to argue that the aisthētikon listed in T(a), to be counted as a part of the soul, is not really the perceptive capacity, but the whole complex including the perceptive capacity as its basic component. It seems that this goes beyond what the text says and requires. Remember that Aristotle arrives at the list of four items in T(a), including the aisthētikon, by specifying four types of activity such that engaging in at least one of them is sufficient to attribute life. There is no reason to suppose that aisthēsis, as one of the activities (a), refers to anything more complicated than sense-perception. Similarly, there is no reason to suppose that the other capacities of the soul in T(a), derived from the corresponding life-activities, refer to some complexes of capacities. Finally, Whiting includes practical nous among the capacities constituting the ‘locomotive part’, which means that it is different but not separable in account from the perceptive capacity. Indeed, Whiting claims that phantasia, desire, and practical nous are ‘extensions or developments’ of the perceptive capacity, albeit in different degrees of complexity. However, Aristotle insists on the fundamental difference between thinking, which has to do with essential features, and perceiving, which has to do with sensible particulars. If this difference is to be of any significance, it seems unlikely that the capacity of thinking, even if ‘only’ practical, could be reduced to a form of perception, however complex. Although practical thinking no doubt requires perception, desire, and complex forms  ‘What distinguishes rational animals from non-rational ones is simply that they have more complex forms of perception and so more complex forms of imagination’ (Whiting, ‘Locomotive Soul’, ).



Klaus Corcilius and Pavel Gregoric

of phantasia as necessary conditions, it is not reducible to perception, at any rate not in the same way as desire and phantasia are. Phantasia and non-rational desire are both explicitly defined with reference to perception (DA . , a ff., and . , a–), whereas nothing of the sort is said or implied about practical nous. It follows that practical nous, after all, must contain at least one element which is separable in account from the perceptive capacity, namely its noetic element, which means that it cannot belong among the capacities of the ‘locomotive part’.

University of Hamburg University of Zagreb

BIBLIOGRAPHY Aquinas, The Commentary on Aristotle’s De anima, trans. K. Foster and S. Humphries (London, ). Barnes, J., ‘Aristotle’s Concept of Mind’ [‘Concept’], Proceedings of the Aristotelian Society,  (–), –. Bos, A. P., ‘Aristotle on Dissection of Plants and Animals and his Concept of the Instrumental Soul–Body’, Ancient Philosophy,  (), –. Burnyeat, M. F., Aristotle’s Divine Intellect (Milwaukee, ). Caston, V., ‘Aristotle’s Two Intellects: A Modest Proposal’, Phronesis,  (), –. Corcilius, K., Streben und Bewegen: Aristoteles’ Theorie der animalischen Ortsbewegung [Streben und Bewegen] (Berlin and New York, ). Fine, G., On Ideas: Aristotle’s Criticism of Plato’s Theory of Forms (Oxford, ). Fodor, J. A., The Modularity of Mind (Cambridge, Mass., ). Gregoric, P., Aristotle on the Common Sense [Common Sense] (Oxford, ). Hicks, R. D. (ed.), Aristotle: De anima [De anima] (Cambridge, ). Lefebvre, D., ‘L’argument du sectionnement des vivants dans les Parva naturalia: le cas des insectes’, Revue de philosophie ancienne,  (), –. Morrison, D., ‘Χωριστός in Aristotle’ [‘Χωριστός’], Harvard Studies in Classical Philology,  (), –. Polansky, R., Aristotle’s De anima [De anima] (Cambridge, ). Ross, W. D. (ed.), Aristotle: De anima, editio maior (Oxford, ). (ed.), Aristotle: Parva naturalia (Oxford, ). Seidl, H. (ed.), Aristoteles: Über die Seele (Hamburg, ). Siwek, P. (ed.), Aristotelis tractatus De anima, nd edn. (Rome, ). Sorabji, R., ‘Body and Soul in Aristotle’, Philosophy,  (), –. Sprague, R. K., ‘Aristotle and Divided Insects’, Méthexis,  (), –.

Separability vs. Difference



Vander Waerdt, P. A., ‘Aristotle’s Criticism of Soul-Division’, American Journal of Philology,  (), –. Vlastos, G., ‘ “Separation” in Plato’, Oxford Studies in Ancient Philosophy,  (), –. Whiting, J. E., ‘Locomotive Soul: The Parts of Soul in Aristotle’s Scientific Works’ [‘Locomotive Soul’], Oxford Studies in Ancient Philosophy,  (), –.

This page intentionally left blank

ESSENCE AND PE R S E P R E D I C A T I O N IN ARISTOTLE ’ S M E T A P H Y S I C S Ζ 4 MICHA I L M . P E R A M A T Z I S

. Introduction I his commentary on Metaphysics Ζ  David Ross concludes that this chapter ‘does not do what it set out to do, viz. to discover whether essence is substance. It only tells us that it is substances alone that in the primary sense have essence. But this way of talking’, Ross suggests, ‘may be found to be a step towards the answering of the original question.’ One of the central aims of the present study is to argue, against Ross, that in Metaphysics Ζ  Aristotle does not ‘set out to discover whether essence is substance’ or to prove that it is. Rather, I shall suggest that his project consists mainly in offering an account of the concept of essence in terms of a certain type of per se predication. This account is far from being final: for Aristotle will qualify and enrich it in the course of Metaphysics Ζ. Yet it is an important starting-point for his later discussions. Aristotle has already claimed in Metaphysics Ζ  that there are (at least) four ways in which substance (οὐσία) is said, those of the essence, the universal, the genus, and the subject (b–). In that same chapter he discusses subjecthood and reaches his initial conclusions about this notion. To be sure, he will take up these conclu© Michail M. Peramatzis  I am indebted to David Charles, Lindsay Judson, and Frank Lewis for helpful remarks on earlier drafts of this paper. The participants in the unofficial Oxford Metaphysics discussion group, Lucas Angioni, David Bronstein, Laura Castelli, Scott O’Connor, and Nathanael Stein, also helped me significantly to reshape my views and improve my arguments. An earlier, condensed version of this paper was presented at the Oxford University Ancient Philosophy Workshop. I am grateful to the conveners (Terry Irwin and David Charles) and participants for their questions and comments. Special thanks are also due to Brad Inwood for his invaluable suggestions.  See W. D. Ross (ed. and comm.), Aristotle’s Metaphysics [Metaphysics],  vols. (Oxford, ), ii.  (Ross’s emphasis).



Michail M. Peramatzis

sions later in Metaphysics Ζ, to modify or improve them. Similarly, in my view, while in Metaphysics Ζ  he develops his conception of essence, he does not intend primarily to explore its status as substance or to demonstrate once and for all that it is the paradigmatic case of substance. Rather, he turns his attention to this notion because it is one of the four ways, enumerated in Metaphysics Ζ , in which substancehood is mostly understood. If so, Metaphysics Ζ  advances the enquiry into those ‘four ways of οὐσία’ which was initiated with the discussion of subjecthood at Ζ , b ff. Several commentators maintain that the account of essence in terms of per se predication comes to an end at b. For at this point, they argue, Aristotle starts investigating the question of what kinds of object have essence, rather than explicating what essence itself is. The question of what kinds of object have essence leads Aristotle to argue that substances alone or (at least) primarily have essence and definition (a–, –; b–). This brings us to Ross’s second claim, made in the passage quoted at the beginning, that the most important result of Metaphysics Ζ  is that substances alone have essence in the primary way. While it is true that this result is central, it is neither the only nor the most crucial conclusion of Metaphysics Ζ . In general, the question of what types of thing are (or are not) essences places constraints on what the notion of essence itself consists in, and conversely. Similarly, Aristotle’s treatment (from b ff.) of the question of what has (or has not) essence is helpful in shaping an intensional account of what it is for an object to be essence-possessor, just as a correct account of this sort would specify the requirements for something to count as an essencepossessor. What is the relation, though, between the question of (a) what it is for an entity to be essence and that of (b) what kinds of object have essence? We could programmatically sketch a reply to this problem by noting that the two questions are strongly interconnected. I shall argue that, in Aristotle’s account, the notion of essence is understood in terms of a particular (per se) type of predicational connection between x and a certain subclass of x’s attributes, even if it is not identified with this type of connection. In the light of this, we shall be able to address the intensional issue of what it is to be essence-possessor in terms of what it is for an object to be part of the appropriate type of per se predication. Analogously, though,

Essence and per se Predication in Metaphysics Ζ 



it will be possible to read (a) as the question of what it is for an entity (normally an attribute or set of attributes) to be part of the relevant type of per se predication. If so, (a) will prove interdependently related to the intensional question of what it is for an object to be essence-possessor. This last question, though, is conceptually dependent on the extensional question (b) of what kinds of object have essence. If so, (b) will indirectly influence our answer to (a). This is not to say, however, that Aristotle eliminates the difference between (a) and (b). For the notion of being essence applies to an attribute (or set of attributes) which is the essence of the relevant object or type of object. By contrast, the notion of essence-possessor, which underlies (b), applies to an object or a type of object that has the corresponding essence. However, my argument will suggest that the notion of predication can preserve this asymmetry: for, quite generally, predication presupposes the distinction between subject and predicate (or even between object and attribute). If so, per se predication can be sensitive to the difference between the object, x, and x’s attributes (including those essential to x). In treating (a) and (b) in terms of per se predication, then, Aristotle can respect the distinction between the object that has essence and the entity that is essence, without placing this distinction at the centre of attention. In Sections .–. I shall argue that Aristotle subscribes to the view just outlined, in which (a) and (b) are unified under the notion of per se predication. Thus, I see the part of his argument advanced at b–a as characterizing more fully that type of per se predication which approximates his conception of essence. In particular, he formulates the relevant mode of predication in terms of the following primacy condition which must be satisfied by essencepossessors: x is primary and so has essence just in case (i) x’s existence presupposes or entails the predication y−z, where y is not different in nature from z; and (ii) x is part of a predication, x−y or y−x, where x is not different in nature from y. In Section . I shall understand this primary mode of predication and the crucial notion of non-difference in nature as contrasting with accidental predication (a–). This suggests that the favoured type of predication is one where the subject is assigned a predicate that signifies its essence.



Michail M. Peramatzis

In Section . I shall develop this interpretation by comparing the primary mode of predication with the third type of per se (per se) discussed in Posterior Analytics .  and . . For in this type of per se connection, too, the items which are predicationally interrelated seem to be non-different in nature from each other. My conclusion from this comparison is that (even after b) Aristotle understands the notion of essence in terms of per se predication. In this understanding, the preferred predicational mode is narrowed down to that of per se. In section  I shall raise several objections to this Aristotelian conception of essence. Further, I shall seek to explain why Aristotle’s account is vulnerable to these objections, and how it could begin to address them.

. Two unsuccessful types of per se predication From the opening lines of Metaphysics Ζ  it is clear that Aristotle intends to set out the notion of essence (τί ἦν εἶναι) in terms of a certain type of per se predication: Since at the beginning we distinguished in how many ways we define substance, and one of these appeared to be the what-it-is-to-be [τὸ τί ἦν εἶναι], we must now investigate this. And first let us make some logical remarks about this [πρῶτον εἴπωµεν ἔνια περὶ αὐτοῦ λογικῶς], i.e. that the what-it-isto-be is what each thing is said to be in itself [ἕκαστον ὃ λέγεται καθ᾿ αὑτό].  I shall not discuss the important question of the place and import of b– for the argument of Metaphysics Ζ  as it is not crucial to my overall position. Nor shall I examine, in the main body of the present study, the exegetical crux of λογικῶς (b), which I render mechanically as ‘logical’. One of the most attractive views would be that the logical enquiry commencing in Metaphysics Ζ  involves premisses which are different from the principles of the relevant scientific discipline (i.e. first philosophy or metaphysics). Equally, it uses considerations which are more general or abstract than the principles peculiar to this science. Myles Burnyeat (A Map of Metaphysics Ζ [Map] (Pittsburgh, ), –) has argued for this type of view. He suggests that the logical investigation does not invoke partisan ontological theses or distinctions such as those between matter and form or potentiality and actuality. Nor does it lay down in advance the principles which account for the change or the being of things. On this view, the logical type of enquiry dominates the overall discussion of Metaphysics Ζ –. To make this view more precise, one would have to explain the difference between the λογικῶς, ἀναλυτικῶς, and ϕυσικῶς types of enquiry. This important issue, though, lies outside the scope of present concerns.  At b I keep the manuscript reading ἕκαστον, without taking it with τί ἦν εἶναι, which would be a grammatically awkward way to express what in Aristotle’s standard jargon is signified by the dative construction τί ἦν εἶναι ἑκάστῳ/τινί. By contrast, I take ἕκαστον with ὃ λέγεται καθ᾿ αὑτό, as the subject of this anaphoric

Essence and per se Predication in Metaphysics Ζ 



For what-it-is-to-be you is not what-it-is-to-be artistic; for you are not in yourself artistic. So what-it-is-to-be you is what you are in yourself. But not everything that a thing is in itself is what-it-is-to-be that thing; for whatit-is-to-be something is not what that thing is in itself in the way in which white applies to surface, because what-it-is-to-be surface is not what-itis-to-be white. Nor again is the what-it-is-to-be what comes from both, what-it-is-to-be white surface, because this is added on [ἀλλὰ µὴν οὐδὲ τὸ ἐξ ἀµϕοῖν, τὸ ἐπιϕανείᾳ λευκῇ, ὅτι πρόσεστιν αὐτό]. (Metaph. Ζ , b–, –, trans. Bostock, slightly modified)

Aristotle’s first move is to claim that x’s essence consists of a certain type of x’s per se attributes: your essence is not just what-it-is-tobe artistic as you are not in yourself (οὐ γὰρ κατὰ σαυτόν) artistic. The idea is that, while it might be true that you are artistic, being artistic does not belong to you in the relevant per se manner in which essences belong to the objects they are essences of. Let us assume, for the sake of the argument, that you are essentially nothing other than a human. If so, you might cease to be artistic but continue to be a human. Moreover, being artistic belongs to several other types of objects (such as musical instruments, theatrical performances, or abstract paintings) over and above humans such as yourself. It does not seem correct, though, to hold that objects like these are the same as you in essence. From the opening section of Metaphysics Ζ , then, Aristotle implies that the sort of per se predication that essencehood consists in must comply with a requirement for coextensiveness between the terms which signify the essence and the (type of) object that has the relevant essence. Because ‘you’ and ‘being artistic’ do not satisfy this requirement, it would be incorrect to propose as a definition the claim ‘you =def clause. To get this result, Alexander transposes ἕκαστον after λέγεται. This renders the text grammatically neat but may not be absolutely necessary. With or without the transposition, the ἕκαστον makes grammatical sense. If so, the emendations of Ross (ἑκάστου) or of Bonitz and Jaeger (ἑκάστῳ) seem unnecessary.  I shall not go into the problematic details of this example. For example, does it presuppose that there are individual essences peculiar to particulars such as you? Do particulars (in Aristotle’s view) have essences of their own at all? Aristotle’s aim is not, it appears, to make any points about whether particular objects (as opposed to kinds) have essences. Nor does he seem to be interested in raising any questions about whether essences are universal or particular (or both). His example is a simple, straightforward case of a predication (‘you are artistic’) which, even if true, is not the type of per se predication which could illuminate the notion of essence. All of Aristotle’s subsequent examples in Metaphysics Ζ  seem to concern essences of types or kinds. For this reason, I shall assume that, in the present context, the locution ‘the essence/definition of [an] x’ refers to the essence or definition of a type x.



Michail M. Peramatzis 

artistic’. Coextensiveness is a necessary condition for essence and definition and so must be part of any account of these notions that has some claim to being successful. There is a further implicit condition introduced by Aristotle’s present example. On the assumption that you are essentially a human, your essence is to be a human. If so, it would be incorrect to claim that your essence is to be artistic even if you are indeed artistic: for you are not essentially artistic and so your essence is not being artistic. To generalize, if the (type of) object is essentially F, its essence is to be F, being F, or what-it-is-to-be F. Let us call this ‘the isomorphism requirement’ for essencehood. This requirement does not require the fallacious claim that, if an object is essentially F, its essence must itself be F. Thus, if an object is essentially a human, it does not follow that its essence is a human. What follows is only that its essence is to be a human or being a human. Hence, isomorphism preserves the distinction between the (type of) object that has the essence and the feature or way of being that is the essence. Moreover, isomorphism does not range over linguistic expressions or concepts which refer to or describe an object or its essence. Rather, it is a condition about real-world entities, objects, and their essences. If so, an object and its essence are isomorphic in the way just suggested independently of the type of linguistic or conceptual items which correspond to them. These two requirements are clearly not equivalent. A putative definiens-term might be coextensive with a definiendum-term without securing isomorphism. Thus, while ‘human’ and ‘being capable of laughing’ are coextensive, being capable of laughing is not the appropriate isomorphic essence, being a human. However, isomorphism obviously entails coextensiveness. For this reason, the isomorphism requirement is violated by definitions which do not respect coextensiveness. In the present example, ‘being artistic’ and ‘you’ are not coextensive and so the feature being artistic is non-isomorphic to you. Hence, it cannot be your essence. Coextensiveness and isomorphism are incontrovertible parts of the concepts of essence and definition. As such they will be crucial in  I do not use the symbol ‘ = def’ to denote the relation of numerical (or strict) identity but that of being defined as.  Coextensiveness between definiendum- and definiens-terms is a condition that is already present, in one form or another, in the Topics (. , a–; . , b– ) and in the Posterior Analytics (. , a–b; . , a–). Before Aristotle, Plato seems to have introduced this condition in the Meno (  –  ).

Essence and per se Predication in Metaphysics Ζ 



our discussion of the subsequent sections of Metaphysics Ζ : for these criteria will function as constraints on the notion of per se predication appropriate to characterize the notion of essence. Before examining Aristotle’s initial remarks about per se predication, however, it is important to clarify the logical grammar of definitional formulations such as ‘you =def being artistic’. One possibility is to take proposed definitions of this sort as statements of identity between referring expressions: ‘the F [object/type] = the G [object/type]’. In this case, the type or kind denoted by ‘the F’ is identical with the type or kind signified by ‘the G’. Similarly, ‘being F = being G’ would be a statement of identity between the ways or modes of being described by the corresponding predicate-expressions. Alternatively, a definition might consist of a definiendum-term which refers to the type being defined (e.g. ‘[the] F [object/type]’ or simply ‘F’) and a definiens-term which describes the way or mode of being (or features) essential to the type in question (e.g. ‘being F’). In this case, though, it would be implausible to claim that there is identity between the corresponding terms as the definiendum-term is a referring expression, while the definiens-term is a predicate. Correspondingly, it seems incorrect to think that these terms refer to the same (numerically one) real-world item: for an object or a type of object cannot be numerically the same as a mode of being or a feature. There are three important points which should be noted in this connection. First, Aristotle’s terminology for essence (τί ἦν εἶναι τινί/ἑκάστῳ; τὸ µουσικῷ εἶναι; τὸ ἐπιϕανείᾳ εἶναι; τὸ λευκῷ εἶναι; etc.) suggests that he understands essences as ways of being or features: ‘what-it-is-to-be x’, as opposed to ‘the x’. For this reason, my formulations will take essences as modes of being described by predicate-expressions. Second, if Aristotelian definitions are of the form ‘object/type F =def mode of being F’, we should not take definitional statements straightforwardly as claims of numerical identity. For objects (or types of objects) are not identical with modes of being (or features). Similarly, though, we should not rule out at the outset the possibi At b– Aristotle adds a third, incontrovertible constraint which is a necessary condition for essence and definition. This is the rule of non-circularity: a definiens-term must not include the definiendum-term. I shall discuss this passage only in the appendix as it presents exegetical difficulties which lie outside the scope of the main argument of Metaphysics Ζ . Plato invokes a similar non-circularity requirement in the Meno:   – ; cf.   – .



Michail M. Peramatzis

lity that certain types of definition or certain (perhaps fully developed) definitional formulations are identity statements. Third, if the relation between x and x’s essence is not straightforwardly numerical identity, there is conceptual space for a notion of definition in which the way of being described by the definiens-term is explanatorily or/and ontologically prior to the (type of) object referred to by the definiendum-expression. Identity is symmetric and so incompatible with priority (which is asymmetric). If so, identical items cannot be prior to one another in the very same sense in which they are identical. By contrast, non-identity allows for asymmetric relations such as that of priority. To leave open the possibility that essences are prior to (but not identical with) the items they are essences of, I shall deploy the following formulations. Locutions such as ‘[the] F [object/kind]’ or simply ‘F ’ will be referring to the (type of) object being defined, while predicate-expressions, such as ‘being F’, will be definiens-terms which describe ways of being essential to the (type of) object defined. .. First unsuccessful case of per se Aristotle notes that not even all types of x’s per se attributes are parts of x’s essence. To understand which types of per se predication are ruled out from the Aristotelian account of essence, it is important to  My argument does not concern cases in which numerical identity is compatible with asymmetric relations of (for example) epistemic or temporal priority.  An alternative conception would take the relation between definiendum and definiens as numerical identity but hold that the definiens is prior as it somehow illuminates the definiendum. There are two ways in which to set out this view. First, the priority relation might be seen as merely epistemic: while the ontological, real-world relation between definiendum and definiens is strict numerical identity, the best route to our understanding or knowing what the definiendum is depends on our grasping the definiens but not conversely. This merely epistemic notion of priority, though, is not strong enough to support the explanatory, causal, or ontological type of priority that we usually ascribe to the defining features of a (type of) object. Alternatively, one might hold that definitional formulae, such as ‘x =def y’, are statements or sentences that express two distinct relations: for they express not only the relation of strict numerical identity between x and y (a thing and its own essence) but also that of explanatory priority of y over x (a thing’s essence over the thing itself). It is not clear, though, how a single formula or statement, such as ‘x =def y’, could carry out this sort of double semantic duty. In clarifying my view of these issues I have benefited immensely from discussions with David Charles. See his ‘Some Remarks on Substance and Essence in Metaphysics Ζ. ’, in K. Ierodiakonou and B. Morison (eds.), The Philosopher as Historian: Essays in Honour of Jonathan Barnes, forthcoming.

Essence and per se Predication in Metaphysics Ζ 



b

assess the implications of the example offered at  –. There are two ways in which to take this example: () Being white belongs per se to surface but: ¬ [‘surface =def being white’]. Or: () Being a surface belongs per se to (being) white but: ¬[‘white =def being a surface’]. Most commentators argue that it is important, at this juncture, to decide which item is supposed to be a per se attribute and which the intended subject or, correspondingly, what the definiens and what the definiendum are in the examples of definitions that Aristotle provides. My proposal differs radically from construals of this type. First, there is no need, in the present context, to make the choice as between () or (). This is not to say that we should interpret Aristotle’s argument as proceeding on the basis of two incompatible readings jumbled together. Rather, his argument can be seen as covering two instantiating cases of one single mode of predication. If so, there is no reason to commit ourselves to either reading exclusively. In this respect, my interpretation is more neutral than  Indeed, the text is problematic at this point: on one reading of b– being white belongs to surface (ἐπιϕανείᾳ λευκόν), while on another being a surface belongs to (being) white (ἐπιϕάνεια λευκῷ).  In favour of () it is argued that it seems plausible to say that the essence of (being) white consists (partly?) in its being a surface rather than holding that the essence of a surface consists in its being white. If so, the subject of the per se predication and the intended definiendum-term is ‘(being) white’, while being a surface is the per se attribute described by the proposed definiens-term (see M. Frede and G. Patzig (ed., trans., and comm.), Aristoteles: Metaphysik Ζ [Ζ],  vols. (Munich, ), vol. ii ad loc.; Frede and Patzig, however, admit that interpretation () is closer to Aristotle’s way of talking about per se attributes). Against this line of interpretation, some commentators hold that () is more congenial to Aristotelian jargon about per se predication: for Aristotle would normally say that a surface is per se white since being white is defined as a colour-quality of some sort which belongs to some surface or other. Further, in this construal, the aim of the present argument of Metaphysics Ζ  is to rule out from the account of essence the second kind of per se predication introduced in Post. An. . , a–b (see Ross, Metaphysics, ; cf. my ‘Aristotle’s Metaphysics Ζ : Criteria for Definition and τί ἦν εἶναι and their Relation to Posterior Analytics . ’ [in Greek], Deukalion, . (), –). Attributes of this per se type are defined in terms of the subjects to which they belong. Thus, for example, being odd is a per se attribute of number as being odd is defined as being a number of a certain sort. If the connection with Metaphysics Ζ  is correct, this implies that we should understand being white as a per se attribute of surface: for being white is defined in terms of its belonging to a surface.



Michail M. Peramatzis

the alternatives. Second, it is preferable to understand Aristotle’s argument as offering a single reason why both () and () should be blocked from a correct conception of essence. This would make the proposed interpretation inclusive as it would encompass both () and () as examples of a type of per se predication which is discarded from the favoured account of essence. Aristotle’s point, then, might be simply that the type of connection obtaining between surface and being white or, for that matter, between (being) white and being a surface cannot capture the type of link holding between a (type of) object and its essence. In () being white belongs in a given per se or even necessary fashion to surface. Equally, in () being a surface belongs in the relevant per se or necessary manner to (being) white. The grounds for both of these per se or necessary claims would be (for example) that (being) white is just defined by reference to its inhering in a surface. It does not, however, follow from the per se necessary claim ‘surface is per se white’ that all there is to the nature of surface is its being white. Nor does it follow from the other per se necessary claim, ‘being white is per se being a surface’, that the essence of (being) white is exhausted by its belonging to a surface. If so, both of the following formulae are incorrect as definitions: ‘Surface =def being white’; ‘(Being) white =def being a surface’. The reason why these cases are unsuccessful is common to both: merely per se connections do not necessarily imply the relevant essential connections between the subject and the attribute or the attribute and the subject. It is plausible to think that the deeper reason for rejecting such definitional formulae is that they do not even pass the coextensiveness test. For, clearly, not all surfaces are white. If this is correct, Aristotle is ruling out one single type of per se connection which (we might think) has two sorts of instantiating  In these formulations I take the relevant per se predications as being equivalent to claims of necessity in some relevant sense of ‘necessity’. For, clearly, there is a necessary connection between the terms of a per se predication such as ‘being white’ and ‘surface’. The reason would be (for example) that ‘belonging to a surface’ is part of the defining account of ‘being white’. Let us call this sort of necessary connection ‘per se necessity’. That Aristotle himself conceives x’s per se attributes as being x’s necessary attributes (in some relevant sense of ‘necessary’) is obvious from Post. An. . , b–, –, –. I shall not discuss these passages. Nor shall I argue further for the connection between being per se and being necessary as these issues are not vital to my present argument.

Essence and per se Predication in Metaphysics Ζ 



case along the lines of () and (): for both are cases in which the subject and the attribute are related in a per se fashion which is not equivalent to any essential connection. For this reason, it would be incorrect to characterize the notion of essence on the basis of this type of per se predication. Clearly, in this interpretation, it is not necessary to choose () or () as the correct rendering of b–. Indeed, it would be a mistake to attempt to do so. For Aristotle’s aim is not to count one particular type of case as a correct instance of being an essence while ruling out another. Rather, it is a certain type of predicational connection (either between surface and being white or between being white and being a surface) which is deemed unsuccessful in shedding any light on the concept of essence. .. Second unsuccessful case of per se: compoundness and nonisomorphism Aristotle’s next move made at b– raises related interpretational questions about what the intended definiendum and definiens are. In this case, however, it is easier to fix the definiens simply on the basis of the text. ‘Nor is the compound, being a white surface, the essence’: ἀλλὰ µὴν οὐδὲ τὸ ἐξ ἀµϕοῖν, τὸ ἐπιϕανείᾳ λευκῇ [sc. τί ἦν εἶναι ἐστί]. This seems to establish that the proposed definiens-term is ‘being a white surface’, an expression which describes the putative compound essence being a white surface. Beyond this point, though, the argument is obscure. There arise three crucial questions. First, what is the intended definiendum-term, ‘surface’ or ‘(being) white’? Second, what motivates Aristotle to assess compound entities, such as being a white surface, as candidate-essences of non-compound objects, such as surface or (being) white? And, third, why is he blocking the claim of these compound entities to being essences of either of their simple components? My answer to the first question about the intended definiendumterm will follow the same neutral and inclusive strategy deployed in my interpretation of the previous example. Given the per se con The phrase τί ἦν εἶναι ἐστί is understood from b: οὐδὲ δὴ τοῦτο [sc. τὸ καθ ᾿ αὑτό] πᾶν [sc. τί ἦν εἶναι ἐστί, understood from b–]. It should be emphasized that the introductory connectives οὐδὲ δή at b and ἀλλὰ µὴν οὐδέ at b suggest the grammatical parallel I am relying on at this juncture.  Commentators disagree about the answer to this question. Some hold that, while it is reasonable to take the essence of being white as consisting in its being a white surface, it is not plausible to think that a surface essentially is a white surface. If



Michail M. Peramatzis

nection between surface and being white or (being) white and being a surface already introduced at b–, the minimal reply to this question is that the compound entity being a white surface is not the successful essence either of surface or of (being) white. It would be incorrect, then, to opt for either of these two possible definiendum-terms: for they are simply two instances of subjectterms involved in one single type of per se connection. This line of argument suggests a related answer to the second question about Aristotle’s motivation for examining compound entities as candidate-essences. At b– Aristotle has undermined the claim of a certain type of per se attribute to essencehood. This implies that his argument presupposes per se connections such as ‘surface–being white’ or ‘(being) white–being a surface’ (where ‘–’ is an indicator of the relevant sort of per se connection). At b–, then, Aristotle points out that the compound entities (such as being a white surface) which correspond to such predications (τὸ ἐξ ἀµϕοῖν) are not successful as essences of the noncompound constituents of these predications. If so, Aristotle’s aim in denying the status of essence to these compounds is to restrict even more the notion of per se predication in order to approximate a better characterization of essence in terms of this notion. Similarly to my treatment of the previous difficulty, this view presupposes simply that Aristotle is working against the background of per se necessary predications. This strategy of clarifying, underpinning, or even inferring ontological theses on the basis of corresponding predications has already been employed in Metaphysics Z.. This Aristotelian move is epitomized in the phrase of a–: ‘this is apparent in such a predication’ (ὅπερ ἐµϕαίνεται ἐν τῇ κατηγορίᾳ τῇ τοιαύτῃ). In that context, Aristotle claims that predications such as ‘x is walking’, ‘x is good’, ‘x is white’, etc. make it clear that attributes such as being walkso, ‘(being) white’ should be the definiendum-term, while being a white surface the candidate compound essence (see Frede and Patzig, Ζ, ad loc.). Against this, it is argued that talk of per se predication so far implies that the subject of the predication and, hence, the intended definiendum-term is ‘surface’. If so, surface is the entity that the compound, being white surface, is not the successful essence of (see Ross, Metaphysics, , on b).  I am not claiming that Aristotle’s argument requires, in a muddled manner, two distinct or incompatible readings at the same time. Rather, it works against the background of one single type of per se predication with two exemplifying types of case.

Essence and per se Predication in Metaphysics Ζ 



ing, being good, being white, etc. are beings because there is an underlying subject, a substance (perhaps a particular substance), a real-world object (or a real-world object of a certain type) which is walking, is good, is white, etc. (a–). It should be noted that, at present, I am not examining the validity and soundness of this reasoning or its conclusions. Nor am I assuming that Aristotle’s method is entirely reliable. My claim is only that a similar structure is at work in the first section of Metaphysics Ζ . Thus, the per se predications ‘surface–being white’ or ‘(being) white–being a surface’ make it clear that there is a compound, being a white surface, which is described by them. A compound of this sort might be thought of as a possible candidate for being the essence of either of its non-compound constituents. Aristotle seems to argue that a view of this sort would be mistaken. This brings us to the third difficulty raised at the outset: why is Aristotle ruling out compounds of this type from his conception of essence? The reason he gives at b looks obscure, inadequate, and artificially ad hoc: ‘because this is added on’ (ὅτι πρόσεστιν αὐτό). I shall proceed on the assumption that πρόσεστιν implies addition of some sort. Aristotle’s claim, then, is that adding the definiendum-term—either ‘surface’ or ‘(being) white’—to the definiens-side yields complex definiens-terms. If so, he is discussing proposed definitions such as the following: ‘Surface =def being a white surface’; ‘(Being) white =def being a white surface’. His position is, it appears, that complexity yields problematic definitions because compound entities are not essences of (essentially) non-compound objects even if they are linked to these objects in some per se or necessary fashion. Conversely, we might add, noncomplex entities are not essences of (essentially) complex objects even if they belong to these objects per se or necessarily.  There are textual and interpretative problems with this sentence. For instance, the αὐτό at b could be referring either to surface or to (being) white. Frede and Patzig (Ζ, ad loc.) suggest that we should read, together with the best manuscripts, αὕτη or αὐτή instead of αὐτό at b. If this is correct, the referent is the feminine ἐπιϕάνεια. Aristotle’s point, in that case, would be that ‘surface’ is added to something in the definiens. There is no need, for present purposes, to commit myself to any particular reading of this uncertain part of the text.  In this respect I agree with Frede and Patzig (Ζ, ad loc.), who take this passage to be referring to cases of addition. Apart from this point, however, my interpretation need not inherit any of the further components of their construal.



Michail M. Peramatzis

The most plausible reason why Aristotle ostracizes such per se connections from his account of essence is that they clash with the isomorphism requirement which is implicitly in place from b ff. For in per se predications of this sort the putative essence does not match the corresponding compound or non-compound essential structure of the relevant object. Aristotle concludes, in effect, that essencehood is not to be specified in terms of these types of per se predication which entail non-isomorphism between an object and its essence.

. Compoundness and per se The application of the isomorphism requirement to the example given at b– yields an important result: compound entities are not essences of non-compound objects, while non-compound entities are not essences of compound objects. This result renders urgent the further examination of the following two questions: Can compound entities be essences of any objects at all and, if so, of what kind of object? Are compound objects even the type of thing that could have essence? To tackle these questions, Aristotle moves on to the discussion of the following section: Now, since there are compounds in accordance with the other categories too (for there is something which underlies each of them, e.g. quality, quantity, time, place, motion), we must, therefore, examine whether there is an account of what-it-is-to-be each of those compounds, and whether these too have a what-it-is-to-be, as e.g. white man [σκεπτέον ἆρ᾿ ἔστι λόγος τοῦ τί ἦν εἶναι ἑκάστῳ αὐτῶν, καὶ ὑπάρχει καὶ τούτοις τὸ τί ἦν εἶναι, οἷον λευκῷ ἀνθρώπῳ]. Suppose, then, ‘cloak’ to be the name for this [ἔστω δὴ ὄνοµα αὐτῷ ἱµάτιον]. What is the what-it-is-to-be cloak [τί ἐστι τὸ ἱµατίῳ εἶναι]? But, it may be said, this is still not one of the things which are said of a thing in itself. Or is it rather the case that ‘not said of a thing in itself’ is  At b I seclude τί ἦν λευκῷ ἀνθρώπῳ together with Ross and Jaeger. Further, I do not take οἷον λευκῷ ἀνθρώπῳ as shorthand for οἷον τί ἦν εἶναι λευκῷ ἀνθρώπῳ. Rather, I think that it is an epexegesis of (or in apposition to) τούτοις at b. If so, the dative αὐτῷ at b picks up the dative λευκῷ ἀνθρώπῳ of b. Hence, the name ἱµάτιον is not reserved for the essential way of being τί ἦν εἶναι λευκῷ ἀνθρώπῳ but for the type λευκὸς ἄνθρωπος.

Essence and per se Predication in Metaphysics Ζ 



spoken of in two ways, one of them being by addition, while the other not [by addition]? For, in the first case, that which is defined is said by being added to something else, as would happen if, for example, in defining whatit-is-to-be white, one offered the account of white man; in the second case, the reverse occurs, as would happen if, for example, ‘cloak’ signified white man, but one defined cloak as white. In fact, a white man is white, but, none the less, what-it-is-to-be [a white man] is not what-it-is-to-be white. (Metaph. Ζ , b–a, trans. Bostock, with changes)

There is a general point which is worth noting at the outset. Commentators emphasize the distinction between the questions of what essence is and what kinds of item have essences. Thus, they think that the first section of Metaphysics Ζ  addresses the first question: ἐστὶ τὸ τί ἦν εἶναι ἕκαστον ὃ λέγεται καθ ᾿ αὑτό (b– ). By contrast, the discussion from b onwards tackles the second: σκεπτέον ἆρ᾿ . . . καὶ ὑπάρχει καὶ τούτοις [sc. τοῖς συνθέτοις] τό τί ἦν εἶναι (b–). There is no denying that the distinction between the two issues is extremely important. It would be a serious mistake to conflate them. It is equally true, though, that there is conceptual interplay between the two, especially in the present context, which discusses compoundness. Schematically, the intended interdependence relations between the different sorts of question seem to run as follows (where ‘~’ denotes the relation of conceptual interdependence): () What kinds of entity have essence~What kinds of entity are essences. () What kinds of entity are essences~What the notion of essence consists in. () What kinds of entity have essence~What the notion of es Cf. M. J. Woods, ‘Substance and Essence in Aristotle’ [‘Substance’], Proceedings of the Aristotelian Society,  (–), – at –. The view that b divides Metaphysics Ζ  into two separate sections has interesting implications for the notion of logical investigation. On the basis of the etymological connection between λόγος (in the sense of ‘account’ or ‘definition’) and λογικῶς, it might be argued that Aristotle’s ‘logical’ aim in Metaphysics Ζ  is to define the notion of essence, a project that stretches up to b. After b, however, Aristotle concludes his ‘logical’ account of essence in terms of a certain type of per se predication and turns to the extensional questions of what things have essences or what things qualify as essences. In my view, these last two questions are not completely independent of each other, even if conceptually distinct. Further, they are addressed in parallel to shed more light on the notion of essence itself. For this reason, Aristotle’s ‘logical’ account of what essence is continues (perhaps indirectly) after b.



Michail M. Peramatzis sence consists in [from () and () by the transitivity of the relation of interdependence].

Let me spell out this abstract schema. While () seems incontrovertible and () follows validly from () and (), it is important to underpin (). The question of whether compounds have essences (posed at b–) is equivalent to whether compounds as such (or in so far as they are compounds) have essences. If so, Aristotle is asking whether compound objects are essentially compound. Given the isomorphism requirement, if the answer to this question is affirmative, compound objects have essences which themselves are compound, thereby mirroring the essential complexity of the objects they are essences of. Hence, because essences could themselves be compound, the notion of compoundness might (but need not) be part of the concept of essence. For instance, if all essence-possessors are essentially complex with corresponding complex essences, the notion of compoundness could be part of essencehood. If, however, only some but not all essence-possessors are essentially compound, compoundness would not, presumably, be part of the concept of essence. If, by contrast, the answer to the initial question is that compounds as such do not have essences, no objects are essentially compound. Given the isomorphism requirement, this entails that no essence is compound: for no essence reflects any compound structure of the object it is the essence of as no object is essentially compound. But if so, the notion of compoundness could not be a part of the concept of essence. That Aristotle himself sees the connection between the question (raised at b–) of whether compounds have essences and that of whether they are essences can be supported by two textual points. First, he begins enquiring into whether compounds have essences and definitions by stipulating the name ‘cloak’ for the compound entity white man (b–). If so, his question is whether cloak has an essence. Immediately, however, he adds a question about what-it-is-to-be a cloak (b: τί ἐστι τὸ ἱµατίῳ εἶναι;). This implies that he turns his attention to putative compound essences of corresponding compound objects. For ‘what-it-is-to-be a cloak’ stands for what-it-is-to-be a white man. This, however, is not a compound object that might (or might not) have an essence but clearly is a compound entity that might (or might not) be an essence.

Essence and per se Predication in Metaphysics Ζ 



Second, the opening question of the subsequent section (raised at a–) runs as follows: ‘but is what-it-is-to-be a cloak an essence at all?’ While the question of b– was whether the compound white man, signified by the term ‘cloak’, has an essence, Aristotle is now asking whether a compound entity such as what-it-is-to-be a cloak is an essence. These two points suggest that he acknowledges the conceptual interdependence between the two questions. It seems plausible to take this type of interdependence to run along the lines of the argument just offered in support of (). Let us examine this second section of Metaphysics Ζ  more closely. The overall line of argument suggests that Aristotle is envisaging an objection to the claim of compounds to having or being essences. The putative reason offered for preventing compounds from having or being essences is that they are not per se items. Similarly, Aristotle’s counter-reply to this objection involves crucially the notion of being non-per se. It seems, then, that compoundness is discussed in order to shape further the notion of per se predication appropriate to constitute a successful account of essence. If compounds are non-per se, they neither have nor are essences. If so, the notion of compoundness, because it entails being non-per se, is not part of the favoured account of essence. By contrast, if it is possible that compounds be per se, compounds might well have or be essences. Similarly, in so far as it implies being per se, compoundness might be part of the notion of essence. If, for example, all essences that belong in the appropriate per se manner to essence-possessors are compound, the notion of compoundness could be part of essencehood. Although this argument involves many problematic details, there is general consensus about its overall line of reasoning. The example of being a white surface offered at b– naturally leads to the question of whether compounds in general, entities which consist of a subject plus an attribute, have definitions and essences. For rea There are three questions about Aristotle’s claims made at b–. () Why is it important to point out that there are also compounds corresponding to predications other than the one just mentioned, surface–white or white–surface? () Is Ross correct in thinking that Aristotle is now referring to compounds outside the category of substance, that is to say, compounds which do not consist of matter plus form (Metaphysics, )? () Why is ποιόν repeated in Aristotle’s list of non-substance categories at b–, although being white, a ποιόν, is already present in the white surface example? A tentative reply to () would be that reasons of generality require that Aristotle should mention compounds other than white surface. For white surface is not a single, peculiar case of a compound described by a predication. Rather,



Michail M. Peramatzis

sons which remain obscure Aristotle stipulates the term ‘cloak’ as a name for the compound entity white man. Using this stipulation, he then asks whether cloak has an essence and definition. To tackle there are several types of predication which describe compound entities in all different combinations of subject plus attribute. It is, then, important to examine whether compounds, quite generally, have or are essences. My view of () is that it seems implausible to introduce talk of matter–form compounds in this context as there is no mention of the concepts of form or matter so far in Metaphysics Ζ . If so, it would be strange to think that the examples discussed at b– are cases of hylomorphic substance compounds. D. Bostock (trans. and comm.), Aristotle: Metaphysics Books Ζ and Η [Ζ&Η] (Oxford, ), , argues that, in the previous example of white surface, surface is not a substance but is, presumably, a geometrical non-substance entity. By contrast, in the present section Aristotle turns to substance-plus-attribute compounds, such as white man. The difficulty with this suggestion is that even in the present section (b ff.) Aristotle is not, strictly speaking, referring to compounds from substance plus non-substance attributes. Rather, he is interested in compounds from subjects plus non-substance attributes. But if so, the previous example of white surface qualifies as a compound of this sort: for it consists of a subject, surface, plus a quality, being white. Consistently with my reply to (), my answer to () is that Aristotle is not interested in hylomorphic compounds in or outside the category of substance. For Metaphysics Ζ  does not invoke the notions of matter, form, or composite hylomorphic substance. Rather, Aristotle discusses predications and per se connections, and examines whether the corresponding compounds have or are essences. As to (), it should be noted that being white (a part of the compound white surface) is a concrete example of a type of quality-attribute, while ποιόν is the most general category under which all such quality-attributes fall. If so, the mention of ποιόν at b is not, strictly speaking, viciously repetitive. A further consideration is that the list of the categories used at b– might be a commonplace way of referring to the categories. If the list is stereotypical in this sense, it is natural that ποιόν is included anyhow, no matter what example has just been provided.  It is important to clarify the logical form of the intended definiendum. I take it that at b– Aristotle is not asking whether the mode of being or feature being a white man has an essence. If so, the definiendum-term for which ‘cloak’ is stipulated is not ‘being a white man’. Rather, the question is whether the type white man has an essence. Hence, ‘cloak’ is used as a name which signifies the same thing as ‘[the] white man’. There are two points of textual support for this view. First, b– reads ἔστω δὴ ὄνοµα αὐτῷ ἱµάτιον, where αὐτῷ picks up λευκῷ ἀνθρώπῳ at b. If so, the term ἱµάτιον stands for λευκὸς ἄνθρωπος. Aristotle would use λευκῷ ἀνθρώπῳ εἶναι at b if the signification of the term ἱµάτιον was intended to be the relevant way of being. Second, at b– the stipulation is εἰ σηµαίνοι τὸ ἱµάτιον λευκὸν ἄθρωπον. If Aristotle intended to refer to the mode of being, he would write εἰ σηµαίνοι τὸ ἱµάτιον τὸ λευκῷ ἀνθρώπῳ εἶναι (or simply εἰ σηµαίνοι τὸ ἱµάτιον λευκῷ ἀνθρώπῳ εἶναι). If so, the definiendum-term is ‘cloak’, a term which by stipulation signifies the compound type also signified by ‘[the] white man’.  The most plausible reason for this stipulation is that it is required by the logical level of the discussion of Metaphysics Ζ . In this logical framework, explanatory or causal considerations play no significant role. For instance, given the stipulation, the most successful definition of the compound white man, signified by ‘cloak’, would be ‘cloak =def being white man’. In stipulating the name ‘cloak’ for the compound white man, the logical approach seeks to postpone the discussion of the issue of explanatory priority of the definiens over the definiendum. For the putative defining formula,

Essence and per se Predication in Metaphysics Ζ 



this question he examines entities such as being a cloak, candidate compound essences (b: τί ἐστι τὸ ἱµατίῳ εἶναι;). In particular, the question seems to be whether compound entities such as being a cloak are per se attributes or not. The idea, then, would be that if such compound entities are per se attributes, they might be essences of the corresponding compound objects. But if they are not per se, they could not be essences of anything: for, consistently with the foregoing argument of Metaphysics Ζ , x’s essence is a certain subclass of x’s per se attributes. If an attribute of x is non-per se, it cannot be x’s essence. Further, if x has no per se attributes at all, x cannot have any essence. The dialectical structure of the argument suggests that Aristotle seeks to refute those thinkers who argue that compounds such as cloak do not have essences or definitions because they do not have any per se attributes. If his opponents’ view is correct, all the attributes of compounds, attributes which are candidates for being their essences, must be non-per se. If so, they could not be essences of anything and so compounds could not have any essences. However—Aristotle appears to be replying—the notion of being non-per se is understood in two ways, neither of which seems to fit the case of compound entities such as being a cloak. These would be the obvious candidate-essences for compound objects such as cloak: (i) Non-per se by addition: while the definiendum is ‘(being) white’, the definiens is the account of white man (b–: οἷον εἰ τὸ λευκῷ εἶναι ὁριζόµενος λέγοι λευκοῦ ἀνθρώπου λόγον): ‘(Being) white =def [account of] white man’ or simply ‘being a white man’. If so, the definer has added a redundant item in the definiens-side, something which is not necessary for defining the definiendum. ‘being a white man’, does not describe an entity which could play any substantive role in explaining why the compound, white man, is as it is. Without the stipulation it would be obvious that the proposed definition is tautologically schematic, circular, and so explanatorily vacuous. By making this stipulation, then, Aristotle seeks to clarify a point which is distinct from explanatory priority. Similarly, the logical approach does not even allude to the view that x’s essence, as a real-world entity, structures or organizes x into the type of thing x is. For, in this approach, Aristotle is concerned only with specific per se modes in which certain attributes belong to x. Per se predication, though, seems completely blind to the causal relation of making x what it is, the relation obtaining between x’s essence and x. Frede and Patzig (Ζ, ) argue that Aristotle uses ‘cloak’ to avoid engaging in merely verbal debates. They do not, however, spell out their claim further.



Michail M. Peramatzis

As I argued in Section ., definitions of this sort ignore the isomorphism requirement. While the definiendum is simple, the putative definiens is complex. This case, then, matches Aristotle’s example given at b–. There, the compound entity being a white surface is not the essence of surface or of (being) white because (Aristotle argues) the thing being defined—either surface or (being) white—is added on to the thing doing the defining. Here, a further reason seems to be provided why this is so: the compound entity being a white man is not the relevant type of per se attribute of (for example) being white, the non-compound entity being defined. Presumably, per se attributes of this sort cannot be the essence of either of their non-compound constituents as they entail non-isomorphism. (ii) Non-per se by subtraction (or non-addition): while the definiendum is ‘cloak’ (the term signifying the same as ‘white man’), the definiens is just ‘being white’ (b–: οἷον εἰ σηµαίνοι τὸ ἱµάτιον λευκὸν ἄνθρωπον, ὁ δὲ ὁρίζοιτο ἱµάτιον ὡς λευκόν): ‘Cloak =def being white’. Thus, the definer has left out (subtracted) from the definiens-side the expression ‘being a man’, an item which is necessary for defining the definiendum. For this reason, this definition, too, falls short of the isomorphism condition. While the object white man (sig In both cases I formulated the types of being non-per se on the basis of Aristotle’s examples alone, without entering the problematic textual details of b– . In the case of addition the phrase τῷ αὐτὸ ἄλλῳ προσκεῖσθαι λέγεται ὃ ὁρίζεται used at b– matches the example of ‘(being) white =def [account of] white man’: what one defines (ὃ ὁρίζεται) is said (λέγεται) by its being added (τῷ αὐτὸ προσκεῖσθαι) to something else (ἄλλῳ). In the case of subtraction or non-addition, though, it is futile to seek to construe or emend Aristotle’s general formulation put forward at b so as to match the example of –: τῷ ἄλλο αὐτῷ [sc. προσκεῖσθαι λέγεται ὃ ὁρίζεται]. Thus, what one defines is said by something other (τῷ ἄλλο), presumably something different from the definiendum, being added (προσκεῖσθαι) to the definiendum itself (αὐτῷ). This, however, yields the same result as the addition case and so cannot be aligned with the example of ‘cloak =def being white’. Dorothea Frede (her view is mentioned by Frede and Patzig, Ζ, ad loc.) proposes to emend b as follows: τῷ ἄλλο αὐτῷ 〈οὐ〉. In this way, the meaning becomes ‘by another one not being added to the definiendum in the definiens’. This emendation renders b consistent with the example of – but has no manuscript support. I preferred to ignore this problem as Aristotle’s general point about being non-per se by subtraction is obvious from his example. Alternatively, it could be argued that, given Aristotle’s excessively elliptical style, the inversion of the position and grammatical case of αὐτό and ἄλλο in the subtraction case (instead of τῷ αὐτὸ ἄλλῳ, the subtraction case has τῷ ἄλλο αὐτῷ) might be equivalent to the phrase ‘the other way about’, i.e. ‘something else is not added to the definiendum in the definiens’.

Essence and per se Predication in Metaphysics Ζ 



nified by ‘cloak’) is compound, the putative essence, being white, is simple. This type of being non-per se covers the examples of b–. In the earlier passage being a surface might be a per se necessary attribute of (being) white but is not the type of per se attribute which is equivalent to the essence of (being) white. Conversely, being white might be thought of as a per se necessary attribute of surface but does not correspond to the sort of per se attribute which is essential to surface. Similarly, in the present context Aristotle concedes that the compound white man (signified by ‘cloak’) is white. One might even go as far as holding that being white is a necessary part of the compound: for (for example) one cannot characterize the compound white man without mentioning its being white. Even so, however, the essence of the compound white man is not just to be white; being white does not exhaust what it is to be a white man (a–: τὸ δὴ λευκὸς ἄνθρωπος ἔστι µὲν λευκόν, οὐ µέντοι τί ἦν εἶναι λευκῷ εἶναι). Clearly, both (i) and (ii) constitute cases of incorrect definitions for the same reason. The proposed definiens does not match the simple or complex essential structure of the relevant object defined. In (i), instead of the account of being white (which is promising as a candidate-definiens of being white) we get the account of white man. In (ii), instead of the account of white man (which would presumably express the essence of cloak) the proposed definiens is ‘being white’. These mismatches disregard the isomorphism requirement. This defect suggests that an attractive candidate-definition of cloak would run as follows: (iii) ‘Cloak =def [account of] white man’ or simply ‘being a white man’.  The text at a– is problematic. The manuscripts read τὸ δὴ λευκὸς ἄνθρωπος ἔστι µὲν λευκόν, οὐ µέντοι τί ἦν εἶναι λευκῷ εἶναι, which involves an apparently redundant εἶναι between τί ἦν and λευκῷ. One view would be to hold that this εἶναι is not redundant: with the understood dative λευκῷ ἀνθρώπῳ the phrase reads οὐ µέντοι τί ἦν εἶναι [sc. λευκῷ ἀνθρώπῳ] λευκῷ εἶναι, which means ‘however, being a white man is not being white’. That is to say, the essence of white man is not the essence of white. Ross (Metaphysics, ) thinks that this rendering requires the article τό before τί ἦν εἶναι. On the other hand, Jaeger follows Bonitz, who secludes the εἶναι between τί ἦν and λευκῷ as he takes it to be an intrusion, presumably due to a copyist’s error. This would yield the meaning ‘white man is white but is not what it is to be white’. This is a correct result which requires only a minor textual emendation. There is no need, for present purposes, to take any definite stand on this textual issue.



Michail M. Peramatzis

In this case the isomorphism condition is fulfilled: for an (essentially) compound object has a compound entity as its essence. This proposal has an additional advantage. In (iii) the putative compound essence being a white man does not fall under either of the two types of being non-per se. It is a further, genuinely distinct case. This, to be sure, does not entail that it is a per se entity or even the type of per se entity which is equivalent to the essence of cloak. It does, however, imply that it might be the type of per se attribute of cloak which reveals its essence. If so, Aristotle’s dialectical interlocutors are not correct in thinking that compound objects such as cloak do not have essences because all their attributes are non-per se. For there are compound entities such as being a white man which belong to these compound objects without being non-per se in any straightforward way. If so, putative compound essences might be per se in a way which would render them genuine essences of the corresponding compound objects. But if this is correct, compound objects might have essences: they might have compound essences such as being a white man, which would belong to them in the appropriate per se manner. The first conclusion to be drawn from this argument is that Aristotle establishes that compounds might be essences as they might belong in the relevant per se manner to certain types of objects. Further, given isomorphism, because these essences are compound, they would most plausibly mirror the essentially complex structure of compound objects. But if so, Aristotle can also infer that compound objects might have essences: their essences would be the compound entities which might belong to them in the appropriate per se way. This line of argument supports my view of the conceptual interconnections between the questions of what types of entity have essences and what types of entity are essences. Second, these considerations imply that the notion of being per se does not automatically rule out compoundness. Thus, in so far as it entails being per se in the relevant way, compoundness might well be a component of Aristotle’s account of essence. To show that the concept of essence does not even partly involve the notion of compoundness, one would have to deploy further, different arguments from the ones put forward by Aristotle’s dialectical interlocutors in the present section of Metaphysics Ζ . Lastly, it seems that the notion of compoundness is a central theme as this section explores its relations to per se predication and

Essence and per se Predication in Metaphysics Ζ 



essencehood. Thus, Aristotle pursues a line of reasoning which is consistent with the overall argument developed so far. The general idea is that essences are a certain type of per se attributes of the objects they are essences of. Taken together with the isomorphism requirement of the previous section, the present section suggests that, if compounds are essentially compound, their essences too are compound. Such compound essences, however, must be per se entities if indeed they are essences of anything. Hence, in attacking the view that, just by being compound, these entities are not per se or essences, the present section widens the available conceptual space. It underpins the possibility that there might be compounds which are per se in the way appropriate for being essences of a certain type of (compound) objects. The next section of Metaphysics Ζ  will take up precisely this possibility and ask whether and how compound entities actually are per se (in the relevant way) and so essences.

. Essence and primacy In the third section of Metaphysics Ζ  Aristotle writes: But is what-it-is-to-be a cloak a what-it-is-to-be at all or not? [Presumably not] for the what-it-is-to-be-something is precisely the type of thing that something is [ὅπερ γάρ τί ἐστι τὸ τί ἦν εἶναι]; but when a thing is said of something other than itself, it is not precisely a certain type of thing [ὅταν δ᾿ ἄλλο κατ ᾿ ἄλλου λέγηται, οὐκ ἔστιν ὅπερ τόδε τι], as, for example, the white man is not precisely a certain type of thing [ὁ λευκὸς ἄνθρωπος οὐκ ἔστιν ὅπερ τόδε τι], if indeed being a certain type of thing belongs only to substances [εἴπερ τὸ τόδε ταῖς οὐσίαις ὑπάρχει µόνον]; hence, the what-it-is-to-be belongs to those things whose account is a definition. And an account is not a definition if a name signifies the same thing as the account . . . but if it is of some primary thing [ὁρισµὸς δ᾿ ἔστὶν οὐκ ἂν ὄνοµα λόγῳ ταὐτὸ σηµαίνῃ . . . ἀλλ᾿ ἐὰν πρώτου τινὸς ᾖ]; and such are the things which are said not by being predicated of anything other than themselves [τοιαῦτα δ᾿ ἐστὶν ὅσα λέγεται µὴ τῷ ἄλλο κατ ᾿ ἄλλου λέγεσθαι]. Hence, the what-it-is-to-be will not belong to things which are not species of a genus but to species of a genus alone (for these seem not to be said by participation, and as affections, or accidentally) [οὐκ ἔσται ἄρα οὐδενὶ τῶν µὴ γένους εἰδῶν ὑπάρχον τὸ τί ἦν εἶναι, ἀλλὰ τούτοις µόνον (ταῦτα γὰρ δοκεῖ οὐ κατὰ µετοχὴν λέγεσθαι καὶ πάθος οὐδ᾿ ὡς συµβεβηκός)]. (a–, trans. Bostock, modified)

This section poses the direct question of whether compound enti-



Michail M. Peramatzis

ties, such as being a cloak (i.e. being a white man), are essences or not. Equivalently: are compound objects, such as white man, the type of thing that could have an essence, an essence which would be a corresponding compound entity, such as being a white man? The answer to these twin questions seems to be negative for three interconnected reasons. Here is a brief outline of these reasons. () An essence is precisely the type (or the certain type) of thing that something is (a: ὅπερ γάρ τί ἐστι τὸ τί ἦν εἶναι; cf. a: ὅπερ τόδε τι; a: τόδε). Being a white man, though, the obvious candidateessence of the compound object white man, is not precisely the certain type of thing that anything is. Correspondingly, the compound object white man is not (essentially) precisely a certain type of thing (a–: ὁ λευκὸς ἄνθρωπος οὐκ ἔστιν ὅπερ τόδε τι). Aristotle explains why this is so in ().  At a I see no reason to add, with Bonitz or Jaeger, τόδε after γάρ. The phrase ὅπερ τι is clearly parallel to ὅπερ τόδε τι at a– and τόδε at a. The context straightforwardly shows that all these expressions signify similar notions. If so, there is no need to impose linguistic uniformity on the text. Thus, for instance, just as there is no reason to emend a– to read εἴπερ τὸ 〈ὅπερ〉 τόδε 〈τι〉 ταῖς οὐσίαις ὑπάρχει µόνον, similarly it seems unnecessary to add τόδε after γάρ at a.  It might be objected that Aristotle’s argument (or my interpretation of his argument) shifts between two distinct and perhaps irreconcilable construals of the phrases ‘just/exactly/precisely what something is’ or ‘just/exactly/precisely the (certain) type of thing that something is’ (which are renderings of the equivalent Greek locutions ὅπερ τι; ὅπερ τόδε τι; τόδε). At a it is said that the essence is precisely what something is (ὅπερ γάρ τί ἐστι τὸ τί ἦν εἶναι; cf. Γ , b–: ἡ ἑκάστου οὐσία ἕν ἐστιν οὐ κατὰ συµβεβηκός, ὁµοίως δὲ καὶ ὅπερ ὄν τι). If so, being precisely what something is looks like a feature of essences. At a–, though, similar characterizations are linked to candidate essence-possessors (the compound white man is not a ὅπερ τόδε τι) or, more successfully, to substances, the only class of objects which have essences (εἴπερ τὸ τόδε ταῖς οὐσίαις ὑπάρχει µόνον). If so, an object, as essencepossessor, is precisely what it is or is precisely the (certain) type of thing that it is. I do not think that Aristotle’s present line of reasoning involves any ambiguities or fallacious inferences. The essence of x, as an attribute (or set of attributes), is precisely what x is, being F, where x ranges over essence-possessing objects. Equivalently, an attribute (or set of attributes) that is x’s essence indicates precisely the (certain) type of thing that x is, to be F (this is a world–world relation between essential attributes and essences). Correspondingly, the predicates which describe such essential attributes or essences signify precisely what x is, to be F (this is a relation between terms and real-world entities). On the other hand, an essence-possessing object, x, is precisely the (certain) type of thing that x is: x is essentially F, where being F is x’s essence. Given the intuitively plausible isomorphism requirement, this result is unsurprising: if x is essentially F, x’s essence must be being F, and if x’s essence is to be F, x is essentially F. That Aristotle himself acknowledges this shift and finds it unproblematic becomes clearer from Posterior Analytics . , a passage which will be discussed in sect. .. At a– a certain type of attribute, being F, indicates precisely what something is (ὅπερ ἐκεῖνο) or precisely the (certain) type of thing that something is (ὅπερ ἐκεῖνό τι) about x, the object to which it belongs. Similarly, an

Essence and per se Predication in Metaphysics Ζ 



() Objects that get defined and so objects that have essences are ‘primaries’ (a; : ὁρισµὸς δ ᾿ ἔστὶν . . . ἐὰν πρώτου τινὸς ᾖ). Primaries are those objects which are not parts of, and whose existence does not presuppose, predications which interconnect items that are different in nature from each other (a–: ὅσα λέγεται µὴ τῷ ἄλλο κατ᾿ ἄλλου λέγεσθαι). By contrast, as () suggests, a compound object such as white man is not (essentially) precisely a certain type of thing, presumably because it does not fulfil this primacy condition. For its existence presupposes a predicational connection between entities which are different in nature from each other (ὅταν δ᾿ ἄλλο κατ᾿ ἄλλου λέγηται, οὐκ ἔστιν ὅπερ τόδε τι): being white is different in nature from being a man. It is for this reason, then, that a compound entity such as being a white man is not precisely the certain type of thing that anything is and so is not the essence of any corresponding compound object. Thus, if compound entities of this sort are the only possible candidate-essences of compound objects, compound objects do not have essences. If so, they are not (essentially) precisely a certain type of thing (a–: ὁ λευκὸς ἄνθρωπος οὐκ ἔστιν ὅπερ τόδε τι). () Only substance-entities are (essentially) precisely a certain essence-signifying predicate, ‘F’, is predicated of a subject-term, ‘x’, and signifies the ὅπερ ἐκεῖνο or ὅπερ ἐκεῖνό τι of the object x. At the same time, at a–, Aristotle holds that the kind (type of object) man is (essentially, I take it) a ὅπερ ἐκεῖνο or ὅπερ ἐκεῖνό τι, viz. a certain type of animal (ὅπερ γὰρ ζῷόν ἐστιν ὁ ἄνθρωπος): its essence is to be this certain type of animal. Several Platonic passages display a similar shift in describing a Form as F-itself (e.g. Phaedo   : αὐτοῦ τοῦ ἴσου; Rep.   ;  –: αὐτοῦ τοῦ καλοῦ;  : αὐτὸ κάλλος) or as what F truly/unqualifiedly is (e.g. Lys.   –: ἐκεῖνο ὅ ἐστιν πρῶτον ϕίλον; Sym.   – : αὐτὸ ὃ ἔστι καλόν; Rep.   : παντελῶς ὄν;  ;   ;   : εἰλικρινῶς ὄν). The Form F, as an object, is (essentially) F without any qualifications, in an absolute way. At the same time, though, it is what F truly or fully is, to be F or F-ness. There is, however, a crucial difference between the two thinkers. While Aristotle emphasizes the distinction between essence-possessing object (e.g. substance) and essence, Plato is not sensitive to it. Indeed, Plato seems to be identifying the Form F, as an object, and the essence, to be F or F-ness. For further references and discussion of these Platonic phrases see C. H. Kahn, ‘Some Philosophical Uses of “to be” in Plato’, Phronesis, . (), – at –. It seems plausible to think that Aristotle’s use of locutions such as ὅπερ τι, ὅπερ τόδε τι, ὅπερ ἐκεῖνο, or ὅπερ ἐκεῖνό τι has its origins in (and perhaps is intentionally contrasted with) Platonic phrases such as αὐτὸ/ἐκεῖνο ὃ παντελῶς/εἰλικρινῶς/τῷ ὄντι ἔστι F. Aristotle’s inclusion of περ in ὅπερ is emphatic just as the Platonic adverbial phrases παντελῶς/εἰλικρινῶς/τῷ ὄντι are. Indeed, there are two places where Plato himself uses the phrase τοῦτο ὅπερ ἐστίν in a way similar to Aristotelian usage, in the sense of ‘precisely what a thing is’. At Rep.   – thirst is said to be τοῦτο ὅπερ ἐστίν by being of some thing (drink), while at Soph.   – whatever is different is τοῦτο ὅπερ ἐστίν relative to something different (from itself). Neither of these occurrences, though, seems to be referring to Forms themselves.



Michail M. Peramatzis a

type of thing ( –: εἴπερ τὸ τόδε ταῖς οὐσίαις ὑπάρχει µόνον). If so, only substances have essences. Conversely, essences, the entities which are precisely the certain type of thing that something is, belong to substances alone. By contrast, because they are not (essentially) precisely a certain type of thing, non-substance compounds such as white man or, quite generally, non-substance entities do not have essences. Correspondingly, none of their possible candidateessences is a genuine essence. .. Understanding the primacy condition Let us examine more closely Aristotle’s present argument starting from (). For it will become obvious that () encapsulates the basic premiss of this argument. Commentators have not fully appreciated the implications of Aristotle’s formulation of the primacy condition for having essence and definition. Thus, for instance, Ross and Bostock take the phrase µὴ τῷ ἄλλο κατ ᾿ ἄλλου λέγεσθαι used at a– as altogether denying predicational structure to objects that have essences. My view stands in stark contrast to this construal: for, clearly, this phrase does not imply that primaries or essence-possessors should be parts of no type of predication at all. To convey this idea of simplicity or complete lack of predicational structure, Aristotle would normally invoke the requirement of ultimate subjecthood. Thus, in Metaphysics Ζ  he claims that a subject is that which other items are said of but is not itself said of anything else at all (b–: τὸ δ᾿ ὑποκείµενόν ἐστι καθ ᾿ οὗ τὰ ἄλλα λέγονται, ἐκεῖνο δὲ αὐτὸ µηκέτι κατ᾿ ἄλλου). By contrast, in my view, the phrase µὴ τῷ ἄλλο κατ᾿ ἄλλου λέγεσθαι prevents primaries from being parts only of a certain type of predication, the one where the predicational relata are different in nature from each other. If so, predicational structure is admissible even in the case of primaries: for they can be parts of, and their existence can imply, certain sorts of predication provided that the items which are predicationally interrelated are not mutually other in nature.  Here the phrase µηκέτι κατ᾿ ἄλλου expresses the idea of complete lack of predicational structure (cf. a–: τὸ µὴ καθ᾿ ὑποκειµένου ἀλλὰ καθ᾿ οὗ τὰ ἄλλα).  My view does not depend on taking the negative particle µή as governing only the ἄλλο κατ᾿ ἄλλου in the phrase µὴ τῷ ἄλλο κατ᾿ ἄλλου λέγεσθαι. Rather, I take the negation as ranging over the whole phrase µὴ τῷ ἄλλο κατ᾿ ἄλλου λέγεσθαι, as Ross and Bostock presumably do. This wide scope of the negation, though, does not entail that primaries do not have predicational structure at all. Rather, they do not have predicational structure of the ἄλλο κατ᾿ ἄλλου type.

Essence and per se Predication in Metaphysics Ζ 



A more precise formulation of my interpretation of the primacy criterion would run as follows: x is primary (in the sense of Metaphysics Ζ ) iff: (i) x’s existence presupposes or entails the predication y−z, where y is not different in nature from z; and (ii) x is part of a predication, x−y or y−x, where x is not different in nature from y. To clarify this understanding of the primacy condition, we can use the example of the species man: ‘Man is a certain type of rational being’ or ‘Man is a certain type of animal’. This example seems to agree with (what I take as) Aristotle’s extensional remarks (made at a–) that species and substances alone are τόδεs, primaries, and so essence-possessors: for being a man, being a rational being of a certain sort, and being an animal of the relevant sort are not different in nature from each other.  This conjunctive primacy criterion suggests that (for example) a species is primary not only in that it is not predicationally related to anything different in nature from itself but also in that its existence does not presuppose an ἄλλο κατ᾿ ἄλλου predication, a predication which does not involve the species at all. If so, a species is primary even if its existence implies a differentia–genus predication, provided that the genus and the differentia are not different in nature from each other. Thus, for instance, in the predication ‘the rational being of a certain type is an animal of a certain type’, entailed by the existence of the species man, being a rational being and being an animal of the relevant types are not mutually other in nature. Similarly, in the hylomorphic conception of species, the form–matter predication would not make the species non-primary if form and matter were not different in nature from each other. If so, the predication ‘a certain (human) type of soul is a certain (human) type of organic body’, implied by the hylomorphic analysis of the species man, is interrelating two items which are not different in nature from each other. These consequences could be seen as welcome given some of Aristotle’s claims in Metaphysics Ζ  and Η . Thus, he argues that the genus is nothing over and above the species or it is a sort of underlying matter for the species (a–). Presumably, the relevant sort of (for example) animality (a genus-feature) which is essential to the species man is the humanly rational type of animality. This suggests that being a rational being and being an animal of the relevant types are not different from each other in nature: for both are of the appropriate human type. Similarly, in Metaphysics Η  the ‘ultimate’ matter and form, the ones defining a species, are thought to be one and the same in a way (b–; cf. a–). It seems reasonable to think that the relevant sort of sameness is (minimally) sameness in nature: being a (human) type of organic body is not other in nature than being a (human) type of soul. I shall not discuss in further detail these important points as they fall outside the scope of the present study. It is sufficient, for my purposes, that my understanding of the primacy condition of Metaphysics Ζ  is consistent with, and could accommodate, these later parts of the Metaphysics.  One might object that this primacy criterion cannot preserve the asymmetry



Michail M. Peramatzis

In this interpretation, the most significant part of the primacy condition is the relation of non-difference in nature between two items which are predicationally connected. There are two sources of evidence for taking the qualification ἄλλο κατ᾿ ἄλλου in this way as opposed to Ross’s or Bostock’s deflationary construal ‘one of another’. First, at Metaph. Ζ , a–, the kind of predication which does not connect items that are ἄλλο κατ᾿ ἄλλου is contrasted with accidental predication. I shall discuss this point in Section .. Second, Posterior Analytics .  and .  discuss a particular type of per se predication, per se, which is virtually equivalent to the primacy condition and is understood as essential predication. In Section . I shall support this point by drawing a comparison between Metaphysics Ζ  and the Analytics. Before adducing these positive points in favour of my interpretation, however, it is important to highlight its advantages over the available alternatives. For instance, Ross understands the phrase µὴ τῷ ἄλλο κατ᾿ ἄλλου λέγεσθαι as describing cases where a thing is not ‘asserted of another’ or a thing’s existence ‘does not imply the assertion of something about something else’. He specifies that ‘a term like “white man”, in which an attribute is assigned to a subject distinct from itself’ contrasts with cases such as the species man, whose existence does not presuppose any predication of this sort. Further, he anticipates an objection to this Aristotelian view: ‘it might be said that a term like “man”, of which Aristotle thinks there is an essence, implies the predication of one term of another (“rational” of “animal”)’. In Ross’s view, ‘Aristotle would reply that these are not ἄλλα to one another since “rational” exists only as an attribute of “animal” and has no separate existence’. Ross’s construal is not entirely clear. He holds that the compound white man is not primary as its existence presupposes the attribute between essence-possessors and their essences. For example, if x = a, y = b, and z = c, and a satisfies this criterion, it follows that a, b, and c are non-different in nature from each other. Because non-difference in nature seems symmetric (or perhaps non-symmetric), it follows that b and c, too, might be able to satisfy the primacy condition. But if so, are b and c, too, essence-possessors, with a being their essence? Although the answer to this question should be negative, the primacy condition cannot, as it stands, fully support it. As I noted in sect. , however, I take the asymmetry between object and its (essential) attributes as being implied by the notion of predication, a notion presupposed by the primacy condition. If so, this condition could, at least indirectly, sustain the relevant asymmetry. 

Ross, Metaphysics, .



Ibid. .



Ibid.

Essence and per se Predication in Metaphysics Ζ 



being white belonging to man. This case is blocked by clause (i) of my primacy condition. However, the phrase ἄλλο κατ᾿ ἄλλου λέγεσθαι need not be understood as requiring only this sort of three-part formulation in terms of compound–subject–attribute. In addition to this threefold case, Aristotle might also be claiming that an attribute such as being white is not a primary entity. For it too is itself involved in predications in which it is different in nature from the entities it is assigned to or those assigned to it. While this case is ruled out by clause (ii) of my primacy condition, it is not envisaged at all by Ross’s construal. Thus, for instance, while being white must belong to some object or other, it is different in nature from the object it belongs to. Similarly, there are predications such as ‘[the] white [thing/type] is F’ in which being F is other in nature than being white. For this reason, then, being white is an ἄλλο κατ᾿ ἄλλου. If so, there is no need to restrict non-primacy just to threefold structures of compound–subject–attribute. Conversely, in the case of primaries, it is not merely lack of such threefold structure that secures primacy but the fact that the primary item itself is not part of, and its existence does not presuppose, predications which interrelate entities that are different in nature from each other. Take, for instance, the predications ‘a rational being of a certain type is a man’ or ‘an animal of a certain type is a man’: in belonging to a certain type of rational being or to a certain type of animal, the species-attribute being a man is not other than them in nature. To offer a different sort of case: in being a component of the predications ‘man is a certain type of rational being’ or ‘man is a certain type of animal’, the species man is not different in nature from being a rational being or an animal of the relevant types. If this is correct, the objection that Ross seeks to rebut does not even arise. Species could be primary entities which have essences even if their existence implies the following threefold structure: species-term =def genus- plus differentia-terms. For this threefold structure would not, by itself, prevent them from being primary or having essences. Rather, species are primary entities in so far as they are not parts of, and their existence does not presuppose, predications which interconnect items that are mutually different in  The formulation ‘the white thing/type is F’ is intended to cover both of the following two possible cases: ‘the/this particular white thing is F ’ and ‘the type white is F’.



Michail M. Peramatzis

nature. If so, there is no need to join Ross in arguing that species are not ἄλλο κατ᾿ ἄλλου because differentia-features do not exist independently of the genus. Nor is it necessary to invoke material from Metaphysics Ζ  or Η  (as Ross does) to support this last thesis. Rather, to set out the notion of primacy, it is sufficient to understand µὴ τῷ ἄλλο κατ᾿ ἄλλου λέγεσθαι in the way just proposed. Indeed, this proposal is consistent with the formulation which Ross himself employs (in a less than fully developed fashion) when he speaks of an attribute being assigned to a subject distinct from itself. In my interpretation I have understood the phrase ‘distinct from itself’ as ‘different in nature from itself’. There are similar problems with Bostock’s construal of the primacy condition. He holds that a primary item is not ‘expressed by predicating one thing of another’. He also complains that ‘no satisfying reason [is given] for the claim that only things that are in a certain sense simple can have essences’. Clearly, however, if no predicational structure is allowed for essence-possessors, it would be difficult to find any successful candidates, items which are entirely ‘simple’, as Bostock calls them. To be sure, one might think that only individual objects could have essences as their existence does not presuppose any predicational structure. For they do not belong to (or: they are not in) anything. Nor are they said of anything. Yet, if an individual object, x, has an essence, x’s essence belongs to x. If so, however, the corresponding predicate describing x’s essence is said of x. While x itself is not an attribute which belongs to anything, x’s existence does presuppose a predicational structure in which x’s essence belongs to x. This suggests that the claim of individual objects to simplicity (or complete lack of predicational structure) is not entirely straightforward or justified. Moreover, if one introduces the hylomorphic understanding of individual objects, they too turn out to be compounds of some sort, in which form somehow belongs to, or is predicated of, matter. But if so, their simplicity is significantly diminished. Even outside this hylomorphic framework, though, it is not plausible to construe the 

Bostock, Ζ&Η, . One might reply that an individual object and its essence are not subject to this quasi-predicational analysis as they are strictly identical. This view assumes at the outset that essences (or forms) are individual, an assumption which is not incontrovertible. More importantly, however, it requires that an object be identical with a feature (or way/mode of being). This seems to be an implausible consequence: for it is difficult to sustain strict identity between categorially diverse items. 

Essence and per se Predication in Metaphysics Ζ 



primacy condition for having essence as a requirement for simplicity or lack of predicational structure. For this type of simplicity rules out species from having any essence. A species such as man is said not only of particulars like Socrates or Callias but also of universals such as a certain type of rational being or a certain type of animal. Moreover, species have attributes that belong to them: for example, the attributes of being a certain type of rational being or animal belong to man. If it is important to sustain the possibility that species have essences, the criterion of primacy should not be understood in terms of simplicity. A further difficulty, one noted by Bostock, is that, if species satisfy the primacy condition because they are simple, they are so presumably in that their definition does not predicate one thing of another. If so, the formula ‘man =def rational animal’ should not imply that there is a type of animal which is rational. Nor should it entail that there is a type of rational being which is an animal. But this seems implausible. In my interpretation, however, this problem does not even arise. For there is no reason to understand primacy as denying predicational structure altogether or as involving this type of simplicity. Thus, for instance, the speciesterm ‘man’ is predicationally related to (for example) ‘animal of a certain sort’ or ‘rational being of a certain type’. For it is the case that man is a certain type of rational being. Similarly, it seems correct to say that being a man belongs to a certain sort of animal. Being a man, however, is not different in nature from being an animal or a rational being of the appropriate types. The third difficulty that Bostock himself mentions is that, in his construal of µὴ τῷ ἄλλο κατ᾿ ἄλλου λέγεσθαι, ‘genera and differentiae must also be simple—and, one would have thought, even more simple. How, then,’ Bostock asks, ‘can it be only species that satisfy the condition [of primacy or µὴ τῷ ἄλλο κατ ᾿ ἄλλου λέγεσθαι]?’ Once again, in my interpretation, this problem is blocked. For there is no reason to understand µὴ τῷ ἄλλο κατ ᾿ ἄλλου λέγεσθαι as requiring only a three-part definitional structure of species–genus–differentia, where genera and differentiae appear to be simpler than species as they are not in turn defined in terms of further genera or differentiae. Aristotle’s additional claim is that the species itself, in so far as it is primary, occurs in a certain type of predication where it is not different in nature from the items it belongs to or those be

Bostock, Ζ&Η, .



Ibid. (Bostock’s emphasis).



Michail M. Peramatzis

longing to it. If so, no question arises about genera or differentiae as such. It is worth emphasizing that, in my examples, the speciesterm (for example) ‘man’ is not predicationally related to the genusterm ‘animal’ or the differentia-term ‘rational’. Rather, the relevant predicational connections obtain between ‘man’ and ‘animal of a certain type’ or ‘rational of a certain type’. The last two, though, are not purely genus- or differentia-terms but are hybrid expressions, notational variants of the species-term itself. A note of caution is in order at this juncture. It is certainly possible that being a man belongs to (the) white or sitting (thing/type): ‘the white (thing/type) is a man’ or ‘the sitting (thing/type) is a man’. Similarly, being white or being sitting might belong to a particular man or some men: ‘this man is white/sitting’ or ‘some men are white/sitting’. In these cases, however, being a man is different in nature from being white or sitting. This suggests that primary items, including species, are not such in any old type of predication in which they might be involved but only in those predications where, and only in so far as, they are not different in nature from the items they belong to or those belonging to them. If this is correct, Aristotle is not offering a criterion which simply picks up certain types of item as primary independently of the kind of predication in which they might occur. Rather, his primacy condition suggests that, whatever the primaries or essence-possessors are, they are primary and have essence only in that they are parts of, and only in so far as their existence presupposes, a particular type of predication. .. Primacy vs. accidental predication (a–) This interpretation of the primacy condition is based on the formulation µὴ τῷ ἄλλο κατ᾿ ἄλλου λέγεσθαι introduced at a– and – . Apart from this formulation, however, Metaphysics Ζ  provides a further important point which suggests that my understanding of primacy is more plausible than the available alternatives. To clarify his remark that, in so far as they are primary, only species have essences, Aristotle writes that ‘they seem to be said not by participation or/and affection, nor accidentally’ (a–: ταῦτα [sc. γένους εἴδη] γὰρ δοκεῖ οὐ κατὰ µετοχὴν λέγεσθαι καὶ πάθος οὐδ᾿ ὡς συµβεβηκός). On the assumption that γένους εἴδη refers to infimae species, this bracketed remark implies that, as primary, species (or

Essence and per se Predication in Metaphysics Ζ 



the corresponding species-terms) are not involved in predications which could be characterized (presumably by a Platonist) as ‘predications by participation’ or ‘by affection’. Take, for example, the predications ‘Socrates is wise’ or ‘A courageous (type of) action is good’. In both cases, Platonists would hold that what is denoted by the subject-terms is different in nature from the attributes assigned to it by the predicate-terms. Indeed, they would argue that perceptible objects (such as Socrates) or certain types (such as a courageous type of action) are as they are only by participating in, or being affected in a certain way by, the corresponding Forms of Wisdom or of Good. On an alternative, perhaps more attractive, formulation, one which reflects some claims made in the Sophist, Platonists might think that Socrates is (truly) said to be wise not in his own right but relative to some feature external to or different from himself (  –: τῶν ὄντων τὰ µὲν αὐτὰ καθ ᾿ αὑτά, τὰ δὲ πρὸς ἄλλα ἀεὶ λέγεσθαι). Similarly, a courageous (type of) action is (truly) said to be good not because of its own nature but because it merely partakes in the Form of Good (or in some item of this sort;   –: οὐ διὰ τὴν αὑτοῦ ϕύσιν, ἀλλὰ διὰ τὸ µετέχειν τῆς ἰδέας τῆς θατέρου). This point could be cast in more familiar, Aristotelian terms too.  Aristotle’s claims about primary and non-primary modes of predication might be seen as a direct development of certain Platonic theses put forward in the Sophist. At   –, in the context of rebutting the view of the so-called Late-Learners, the Eleatic Stranger uses the phrase κοινωνίᾳ παθήµατος to describe the relation between two items which are non-identical. This phrase seems to correspond to Aristotle’s κατὰ πάθος used at a. At   – the Stranger draws the distinction between being said αὐτὰ καθ᾿ αὑτά and being said πρὸς ἄλλα (cf.   –: τὸ µὴ ὂν κατὰ ταὐτὸν ἦν τε καὶ ἔστι µὴ ὄν). This distinction corresponds to that of Metaphysics Ζ  between µὴ ἄλλο κατ᾿ ἄλλου λέγεσθαι and ἄλλο κατ᾿ ἄλλου λέγεσθαι. Nor should one worry about the apparent discrepancy between Plato’s use of the accusative πρὸς ἄλλα and Aristotle’s genitive κατ᾿ ἄλλου: for in Metaphysics Ζ , b–, Aristotle seems to describe the relevant per se or primary mode of predication as follows: ὅσα µὴ κατ᾿ ἄλλο λέγεται, ἀλλὰ καθ᾿ αὑτὰ καὶ πρῶτα. Finally, at   – the Stranger argues that, if a highest kind is also said to be of another highest kind, this is so not because of its own nature but because it participates in the form of the different: οὐ διὰ τὴν αὑτοῦ ϕύσιν, ἀλλὰ διὰ τὸ µετέχειν τῆς ἰδέας τῆς θατέρου (cf.   –: τὸ µὴ ὂν βεβαίως ἐστὶ τὴν αὑτοῦ ϕύσιν ἔχον). A similar language of participation is again used at   – . Claims of this sort correspond to Aristotle’s usage of κατὰ µετοχήν at a. Frede’s interpretation of the Sophist (‘Plato’s Sophist on False Statements’, in R. Kraut (ed.), The Cambridge Companion to Plato (Cambridge, ), – at –) suggests that he takes this Platonic work to be introducing essential or per se predication of some sort along the lines of the Aristotelian primary type of predication set out in Metaphysics Ζ . Frede’s understanding of the first use of ‘. . . is . . .’ in Plato’s Sophist corresponds to the µὴ ἄλλο κατ ᾿ ἄλλου λέγεσθαι type of predication developed by Aristotle at a– and –.



Michail M. Peramatzis

Thus, for instance, in being primary, the species-term ‘man’ is not part of any accidental predication (a: οὐδ᾿ ὡς συµβεβηκός). If so, Aristotle seems to be marking out as primary predications such as ‘Socrates is a man’, ‘man is an animal/rational being of a certain sort’, or ‘an animal/rational being of certain type is a man’. For, in so far as he is a man, Socrates is not other in nature than being a man (although being a man is not just for something to be Socrates). Nor is being an animal/rational being of the relevant sorts different in nature from being a man (or conversely). It should be emphasized that Aristotle’s claim is not that a species-term does not, or cannot, occur in true accidental predications such as ‘the/this white [thing/type] is a man’ or ‘some man/ men are white’. Rather, his point is that, in so far as the species (or anything for that matter) is τόδε, primary, and so essence-possessor, it does not occur in, nor does its existence presuppose, any accidental predications of this sort. While the accidental predications just mentioned might be true, they do not constitute the primary cases of predication Aristotle is seeking for. For, obviously, being white and being a man are different in nature from each other. For this reason, then, being white is not the kind of entity that could be τόδε, primary, or an essence-possessor: for being a man does not belong to it as primary; nor does it itself belong to man as primary. Rather, the two items are ἄλλο κατ ᾿ ἄλλου. This line of argument implies that Aristotle does not simply offer a case (or a type of case) of a primary object or an essence-possessor. His aim is to set out the notion of what it is for an object to have an essence (or to be an essence-possessor) in terms of the primary type of predication. In predications of this type, then, there is no accidental connection between two items. Nor are the two predicational relata different in nature from each other. If so, my interpretation of the primacy condition implies that even in this section of Metaphysics Ζ  Aristotle pursues his initial project of explicating the notion of essence in terms of a certain sort of predication. In Section . I shall argue that this, too, is a type of per se predication and shall specify more accurately the type of per se predication it is. .. Primacy, τόδε, substances, and γένους εἴδη Before taking up the notion of per se predication, however, it is important to establish that neither step () nor () of Aristotle’s pre-

Essence and per se Predication in Metaphysics Ζ 



sent argument is as basic as (), his introduction and clarification of the primacy condition. In () the concept of ‘precisely the certain type of thing that x is’ constitutes a clearer, less technical or less idiosyncratic, way of invoking the notion of x’s essence. As I understood them, the terms ὅπερ τι or ὅπερ τόδε τι seem to be more perspicuous notational variants of the term τί ἦν εἶναι. This is not  Alternatively, as I noted earlier, in using the locutions ὅπερ τι and ὅπερ τόδε τι, Aristotle may be drawing a comparison, and underlining the contrast, between his own concept of essence (usually expressed by terms such as τί ἦν εἶναι τῷ F/ τῷ F εἶναι) and related Platonic notions expressed by locutions such as αὐτὸ/ἐκεῖνο ὃ παντελῶς/εἰλικρινῶς/τῷ ὄντι ἔστι F, which are similar to the Aristotelian terminology of ὅπερ τι and ὅπερ τόδε τι. Moreover, the use of τοῦτο ὅπερ ἐστι by Plato in nontechnical contexts (e.g. Rep.   –, where thirst is τοῦτο ὅπερ ἐστι by being of a drink) suggests that such phrases are less technical and more easily comprehensible ways of referring to the concept of essence than the terms τί ἦν εἶναι τῷ F or τῷ F εἶναι are.  In my interpretation of ὅπερ τι, ὅπερ τόδε τι, and τόδε I have not discussed the possibility that these terms might refer simply to particular or individual objects such as Socrates or Callias, the τόδε τιs or primary substances of the Categories. There are several considerations which count against adopting this interpretational route. (a) Just like the term καθ᾿ ἕκαστον, τόδε τι (or ὅπερ τι, ὅπερ τόδε τι, τόδε) need not mean the same as ‘individual/particular object’. Both καθ ᾿ ἕκαστον and τόδε τι may refer either to individual/particular objects or to particular/determinate types of objects. Thus, for instance, τόδε τι might be picking up (for example) either this man (e.g. Socrates) or this (type of) animal (e.g. man). (b) Whenever individuality or particularity is in view, this notion is fairly consistently expressed by the term τόδε τι. Here, however, we have a variety of equivalent formulae (ὅπερ τι, ὅπερ τόδε τι, τόδε) which are just not the same as τόδε τι. It seems reasonable to think that in Metaphysics Ζ  Aristotle is using different, versatile terminology to refer to a notion which is distinct from, or even more complex than, that of individuality. (c) If the phrases used in this section of Metaphysics Ζ  are understood in terms of individuality, Aristotle’s claim suggests that essences themselves are particular or individual objects (a: ὅπερ γάρ τί ἐστι τὸ τί ἦν εἶναι). This, however, would clash with our standard understanding of essences as (essential) features of, or ways of being for, the objects they are essences of. Alternatively, the individualist might argue that, in Aristotle’s view, only particularized properties or tropes are essences. It would be difficult, though, to cash out this construal in any plausible manner. Is Aristotle introducing individual essences so early and without argument? If so, he would automatically rule out species or (universal) forms from having any essence. For, if the essence is individual, it belongs to one particular object alone. In that case, a single individual essence could not belong to any given one or all of the particular members of the species. Nor could it belong to any given one or all of the instances of the (universal) form. Similarly, it could not belong to the species or the (universal) form themselves understood intensionally. Hence, species or (universal) forms would be left essence-orphan. If so, however, it would be difficult to reconcile this sort of individual, trope-ish essences with Aristotle’s later claim at a– that essences belong to γένους εἴδη alone. This difficulty remains no matter whether we understand εἴδη as species or as (universal) forms. To remove the difficulty, one would have to bring in strong assumptions which are nowhere to be found in Metaphysics Ζ  (or indeed in the wider context of Metaphysics Ζ –). Thus, one would have to assume



Michail M. Peramatzis

to say, however, that step () is not helpful in further illuminating the notion of essence: for it raises a point which naturally leads to the more substantive criterion of primacy set out in (). Similarly, step () is not as central as (). For it offers a merely extensional grip on what objects are τόδε and have essence: it is substances alone which are (essentially) precisely a certain type of thing (a–: εἴπερ τὸ τόδε ταῖς οὐσίαις ὑπάρχει µόνον). If so, essences, the entities which encapsulate precisely the certain type of thing that something is (a: ὅπερ γάρ τί ἐστι τὸ τί ἦν εἶναι), belong to substances alone. From this it would follow that, because they are the only τόδεs and essence-possessors, substances alone are primaries, objects which satisfy the µὴ ἄλλο κατ᾿ ἄλλου λέγεσθαι condition: only substances (or substances of a certain sort) are involved in predications which do not interconnect items that are mutually different in nature. If this is correct, step () also fixes the extension of the notion of primacy in an indirect way: if substances alone are essence-possessors, while essence-possessors satisfy the primacy condition, substances alone are primaries of this sort. To assess this last consequence more fully, it is important to examine it together with (what seems to be) a further extensional remark. Aristotle holds that essences belong to species of genera alone (a–: οὐκ ἔσται ἄρα οὐδενὶ τῶν µὴ γένους εἰδῶν ὑπάρχον τὸ τί ἦν εἶναι, ἀλλὰ τούτοις µόνον). The claim that species alone are essence-possessors is clearly reminiscent of the other ‘alone’ claim made at a–, where substances alone are thought to be τόδεs and essence-possessors. The combination of these two claims entails that the relevant class of substances mentioned at a– must be coextensive with the class ‘species of genera’ referred to at a–. Thus, the relevant type of substance under discussion is not that of the particular, primary substances of the Categories. Nor are these substances the particular hylomorphic compounds that γένους εἴδη refers to forms of genera, forms which are individual. For this to work, one would have to understand a– as suggesting not that essences belong to γένους εἴδη alone but that essencehood (or being an essence) belongs to γένους εἴδη alone. That is to say, only individual εἴδη (forms) are essences. This is required if the individual-form theorist is to reconcile the claim made at a that essence itself is ὅπερ τι (where ὅπερ τι is understood as ‘individual’ by this theorist) with a–, which states that essence belongs only to γένους εἴδη (where γένους εἴδη is understood as ‘individual forms of genera’ by this theorist). Because of all these difficulties, it is preferable to interpret the controversial terms ὅπερ τι, ὅπερ τόδε τι, and τόδε in the way I did, as equivalent to the phrase ‘precisely the certain type of thing that something is’.

Essence and per se Predication in Metaphysics Ζ 



of the Metaphysics. To be sure, Metaphysics Ζ  does not prevent particular substances from being τόδεs or having essences. However, this chapter’s point about τόδεs and essence-possessors does not seem to pertain primarily or directly to substances of this sort. For similar reasons, it seems difficult to sustain an individualist interpretation of the terms τόδε, ὅπερ τι, or ὅπερ τόδε τι, one which holds that these terms refer to particular objects or particularized attributes (tropes). For, in the present context, the relevant type of substances which are τόδε coincide with species, which are not individuals but are types. It is more plausible to conceive of the type of τόδεs and essence-possessing substances under discussion as species within the category of substance, which (together with genera) belong to the class of secondary substances introduced in the Categories. Because they are the only essence-possessors, then, substances or species are the only primaries: they alone are parts of, and their existence presupposes, predications which do not interrelate entities that are mutually different in nature. .. Primacy and per se predication in the Posterior Analytics One might wonder, at this juncture, whether the primary type of predication is a mode of per se predication and, if so, why. This question is not merely exegetical or terminological. If so, it will not suffice to answer it just by appealing to Aristotle’s own formulation offered later, at Metaph. Ζ , b–: ‘those items which are not said relative to something different [from them in nature] but [are  There is a crucial assumption which underlies this argument. I take the phrase γένους εἴδη as signifying species rather than forms of genera. For I find implausible the interpretation of γένους εἴδη in terms of ‘forms of genera’ (cf. Frede and Patzig, Ζ, ad loc.). I see no reason to understand εἴδη as presupposing the matter–form distinction in a context which uses only the notions of predication, per se connections, and compounds consisting of subject plus attribute. Why introduce the notion of form or that of hylomorphic compound in interpreting Metaphysics Ζ  (or indeed Metaphysics Ζ –), a text which does not mention or even allude to these notions? Further, one could argue that Aristotle is signalling that he is using εἶδος in the sense of species, precisely because he also includes the qualification γένους. Thus, while Bostock translates as ‘forms of genus’ (Ζ&Η, ), he notes that here Aristotle uses εἶδος in the sense of species (). For the notion of genus would not be helpful if it were a characterization of form. At this stage of Metaphysics Ζ, one would expect the notion of form to be clarified not in terms of genus but in contrast with the notion of matter or that of hylomorphic compound (cf. Ζ , a–, –). However, the notion of genus is indeed helpful as a qualification of the species: for species are types falling under wider genera. If so, the use of γένους might be seen even as a way of disambiguating εἶδος in favour of the species-sense.



Michail M. Peramatzis

said] in themselves and as primary’ (ὅσα µὴ κατ᾿ ἄλλο λέγεται, ἀλλὰ καθ᾿ αὑτὰ καὶ πρῶτα). Rather, in raising this question, my aim is to point out that, while Metaphysics Ζ  has the resources with which to set out the notion of primary predication in the way I suggested, it is crucial to ask about the deeper theoretical grounds of this notion. My view is that these grounds are to be found in the Posterior Analytics. The first important task is to show that the concept of primacy is Aristotle’s way of describing the appropriate type of per se predication which captures the connection between an object and its essence. This type of per se predication, which is picked up by the primacy condition, is introduced in Posterior Analytics . : ἔτι ὃ µὴ καθ ᾿ ὑποκειµένου λέγεται ἄλλου τινός, οἷον τὸ βαδίζον ἕτερόν τι ὂν βαδίζον ἐστὶ καὶ τὸ λευκόν, ἡ δ᾿ οὐσία, καὶ ὅσα τόδε τι σηµαίνει, οὐχ ἕτερόν τι ὄντα ἐστὶν ὅπερ ἐστίν. τὰ µὲν δὴ µὴ καθ᾿ ὑποκειµένου καθ᾿ αὑτὰ λέγω, τὰ δὲ καθ᾿ ὑποκειµένου συµβεβηκότα. Moreover, certain items are not said of some different underlying subject, as, for example, whereas what is walking, in being something different, is walking (and similarly for what is white), substances, i.e. whatever signifies a this, are not precisely what they are in virtue of being something different. So, items which are not said of an underlying subject [in that way] I call things in themselves, while those which are said of an underlying subject [in that way] I call accidental. (Post. An. . , b–, trans. Barnes, with minor changes)

Here Aristotle’s third type of per se (per se) is characterized as that type of predication where the attribute does not belong to a subject that is different from it (b–: µὴ καθ᾿ ὑποκειµένου λέγεται ἄλλου τινός). Once more, some commentators precipitately infer from this that per se refers to the ultimate subjecthood requirement of  Posterior Analytics .  introduces four different types of per se predication. P is a per se attribute of a subject, S, just in case S is defined by reference to P (but not P alone) and P is one/some among S’s essential attributes (a–). [It is worth emphasizing that P is not the only item mentioned in S’s definition. If this qualification is not in place, the notion of per se would presuppose definitions such as ‘triangle =def lines’ or ‘line =def points’. Further, this notion would entail statements such as ‘all triangles are lines’ or ‘all lines are points’. Jonathan Barnes (trans. and comm.), Aristotle: Posterior Analytics [Analytics] (Oxford, ), –, distinguishes three ways in which to understand Aristotle’s claims at a–. (i) Aristotle conflates the distinction between ontological and definitional dependency (although a triangle is ontologically dependent upon a line, it is not defined as a line or lines). (ii) He does not intend to say that S is defined as P but that P is mentioned in the definition of S. (iii) His examples reflect the geometrical assumptions that a triangle is bounded by lines, while a line is bounded by points. My formulation of per se agrees

Essence and per se Predication in Metaphysics Ζ 



the Categories, where only particular substances are ultimate subjects as they are neither in nor said of anything else. Thus, these commentators conclude that the substances of Posterior Analytics . , the ones which are taken as τόδε τι and per se, are particular substances of this type which are neither predicated of, nor belong to, anything else. This resembles the interpretative strategy of those who take the reference of Metaphysics Ζ  to τόδεs and substances as picking up particulars (a–): for they understand the primacy condition as preventing items with predicational structure from being essence-possessors. Just as in Metaphysics Ζ , the reason for taking per se as equivalent to ultimate subjecthood or for denying predicational structure to per se items is that some commentators overlook the presence and implications of the phrase µὴ ἄλλου λέγεται (b–). Aristotle does not state that per se items are not said of anything at all. He holds simply that they are not said of any subject different from themselves (different in nature, I take it). Clearly, this does not rule out items with predicational structure or items which are not ultimate subjects (such as species) from being per se. To be sure, my understanding of per se does not entail that only items with predicational structure or items which are said of, or belong to, other objects are per se. Nor does it prevent particular substances of the Categories from qualifying as per se. Rather, my point is that Aristotle is also allowing for non-ultimate subjects or items with predicational structure to be per se. Similarly, the claim that particular substances, too, are per se does not render the notion of per se equivalent to ultimate subjecthood. For the reason why particular substances are per se is not that they are not said of anything or do not belong to anything at all. It may well be that even particular substances have predicational structure in that they are subjects to which certain attributes belong. However, in so far as they are per se, their attributes are not other than them in nature. Aristotle’s examples, offered at b–, use the notion of being different or not being different (in nature) to clarify his conception with (ii).] Moreover, P is a per se attribute of S just in case P is defined in terms of S (but not S alone; a–b). Per se (b–) is about necessary connections between items such as being slaughtered and dying. This last type of per se is clearly irrelevant to present concerns.  To make this sort of claim, Aristotle would presumably use a formulation along the lines of Metaph. Ζ , b–: µηκέτι κατ᾿ ἄλλου.



Michail M. Peramatzis b

of per se ( : ἕτερόν τι ὂν; b: οὐχ ἕτερόν τι ὄντα). Entities such as being walking are not per se attributes of the corresponding object, that which is walking or the walking thing/type, in so far as they are parts of predications where the predicational relata are mutually other in nature. Thus, for instance, in the predications ‘the/ this walking [thing/type] is a man’ or ‘some men are walking’, it is clear that being walking and being a man are different in nature from each other. Indeed, it is partly because a thing (or a subclass of a type) is essentially a man that these accidental predications are true. In this sense, what is walking, or the walking thing/type, is something other in nature than being walking: it is (essentially) a man (b–: οἷον τὸ βαδίζον ἕτερόν τι ὂν βαδίζον ἐστί). Further, because of this underlying essential predication, it can be said to be (accidentally) walking. Using the conceptual machinery of Metaphysics Ζ , the walking thing, a putative object, is neither primary nor essence-possessor as its corresponding attribute, being walking, does not belong to it in the per se mode. If so, being walking, the candidate-essence of the walking thing, is not an essence at all. Aristotle notes, by contrast, that, in belonging to some objects, certain entities are per se attributes of these objects in so far as they are not different in nature from them. Suppose, for instance, the predications ‘Socrates is a man’, ‘man is a rational being of a certain sort’, or ‘a rational being of a certain sort is a man’. In all three cases the items which are predicationally interconnected are not different in nature from each other. Further, in all three examples the objects picked up by the subject-terms are precisely the certain type of thing they are without being other in nature than what their predicates signify (b: οὐχ ἕτερόν τι ὄντα ἐστὶν ὅπερ ἐστίν). Hence,  In his concluding remarks about per se (b–) Aristotle does not use the phrase µὴ ἄλλου λέγεται employed at b– or the participial clause οὐχ ἕτερόν τι ὄντα found at b. He says simply that ‘those items which are not said of any subject I call per se, while those which are said of a subject I call accidental’. Thus, one might object to my interpretation by arguing that this general, concluding characterization does not place any weight on the idea that the items which are predicationally interconnected are not mutually ἄλλα or ἕτερα in nature. Rather, it might be said, in his summarizing remarks Aristotle is describing per se items as ultimate subjects where no predicational structure obtains, along the lines of the particular substances of the Categories. This objection presupposes a construal of b– which is less than compelling. Aristotle has already specified that per se items are not said of any subject which is different in nature from themselves (b–) and that they are what they are without being different in nature from what they are predicationally related to (b–). If so, his closing comment at b– need not repeat these claims in full. He might be summarizing by implicitly referring back to these two points. Hence,

Essence and per se Predication in Metaphysics Ζ 



these objects are essence-possessors as they have per se attributes which reveal their substance or essence (b: οὐσία). Correspondingly, the attributes which belong to these objects in the per se manner are their substances or essences. If this is correct, the term οὐσία used at b does not refer to particular substances. Rather, it picks up the substance of certain objects, an entity which is identified with their per se attributes. Unsurprisingly, this is exactly the sense in which the term οὐσία is used a few lines earlier, at a– (ἡ οὐσία αὐτῶν; cf. a, ; b: τί ἐστι). If so, my interpretation of per se as the set of essential attributes of a (type of) object fits better the overall context of Posterior Analytics . . This understanding of per se is equivalent to the kind of predication described by the primacy condition of Metaphysics Ζ . It is also worth noting that, just like the essences of Metaphysics Ζ , which are ὅπερ τι or ὅπερ τόδε τι, the οὐσίαι of Posterior Analytics .  (the substances of things) are τόδε τι. There is a further terminological similarity between the phrases ὅπερ τι or ὅπερ τόδε τι of Metaphysics Ζ  (a–) and the ὅπερ ἐστίν of Posterior Analytics .  (b). In both cases these terms seem equivalent to the phrase ‘precisely the certain type of thing that x is’ or, put more simply, ‘x’s essence’. The conceptual and terminological interconnections with Posterior Analytics .  suggest that the kind of predication intended by the primacy condition of Metaphysics Ζ  is that of per se as set out in the Analytics. If so, Metaphysics Ζ  aims to offer a characterization of the notion of essence in terms of per se predication even if it does not identify the two. This, to be sure, does not answer some of the most important questions that we might have. For instance, why is Aristotle favouring this type of per se predication as an explication of essencehood? Further, even if this type of per se predication provides a fair account of essence, is it the only or the it need not be the case that per se items are not said of any subject at all. This claim would normally be expressed by a formulation such as µηκέτι κατ ᾿ ἄλλου used at Metaph. Ζ , b–. Rather, they are not said of any subject such as the ones implied in the βαδίζον or λευκόν examples just given: τὰ µὲν δὴ µὴ καθ᾿ ὑποκειµένου [sc. τοιούτου; i.e. ἑτέρου τινὸς ὄντος ὅπερ ἐστί] καθ᾿ αὑτὰ λέγω. If this is correct, the scope of µὴ καθ᾿ ὑποκειµένου at b– is constrained by the immediately preceding remark of b–: οὐχ ἕτερόν τι ὄντα ἐστὶν ὅπερ ἐστίν. Moreover, the phrase µὴ καθ ᾿ ὑποκειµένου is tied to the overall context, which is governed by the opening statement of b–: ὃ µὴ καθ᾿ ὑποκειµένου λέγεται ἄλλου τινός. For this reason, I rendered this phrase as ‘items which are not said of an underlying subject [in that way]’, i.e. not predicated of a subject different in nature from them.



Michail M. Peramatzis

dominant part of this account? More importantly, in subscribing to this sort of account of essence, has Aristotle made any substantive conceptual progress? Or is he moving in a circle since he unpacks the concept of essence in terms of a type of predication which crucially involves the notion of x’s not being different from y in nature (a notion which looks dangerously similar to that of non-difference in essence)? I shall address some of these pressing questions in Section . At present, however, it is important to consolidate further the conceptual links between the primacy condition of Metaphysics Ζ  and the notion of per se predication set out in the Analytics. To achieve this, it is helpful to discuss the following passage from Posterior Analytics . : ἔτι τὰ µὲν οὐσίαν σηµαίνοντα ὅπερ ἐκεῖνο ἢ ὅπερ ἐκεῖνό τι σηµαίνει καθ᾿ οὗ κατηγορεῖται· ὅσα δὲ µὴ οὐσίαν σηµαίνει, ἀλλὰ κατ᾿ ἄλλου ὑποκειµένου λέγεται ὃ µὴ ἔστι µήτε ὅπερ ἐκεῖνο µήτε ὅπερ ἐκεῖνό τι, συµβεβηκότα, οἷον κατὰ τοῦ ἀνθρώπου τὸ λευκόν. οὐ γάρ ἐστιν ὁ ἄνθρωπος οὔτε ὅπερ λευκὸν οὔτε ὅπερ λευκόν τι, ἀλλὰ ζῷον ἴσως· ὅπερ γὰρ ζῷόν ἐστιν ὁ ἄνθρωπος. ὅσα δὲ µὴ οὐσίαν σηµαίνει, δεῖ κατά τινος ὑποκειµένου κατηγορεῖσθαι, καὶ µὴ εἶναί τι λευκὸν ὃ οὐχ ἕτερόν τι ὂν λευκόν ἐστιν. Further, those items which signify substance signify, about this thing of which they are predicated, precisely the type of thing that something is or precisely the certain type of thing that something is; those which do not signify substance, but are said of some different underlying subject which is neither precisely this type of thing nor precisely this certain type of thing, are accidental, just like (for example) being white of man. For man is neither precisely what white is nor precisely what a certain type of white is, but presumably animal; for man is precisely an animal. Items which do not signify substance, however, must be predicated of some underlying subject, and there cannot be anything white which is white without being something different. (Post. An. . , a–, trans. Barnes, with changes)  It is significant that this passage is embedded in a context where Aristotle explicitly conducts a ‘logical’ enquiry into why the attributes studied, or proved to belong to certain subjects, by a demonstrative science are not infinitely or indefinitely many (b; a–; b–: λογικῶς). If so, this part of the Analytics might be seen as parallel to the level of discussion of Metaphysics Ζ  (b: λογικῶς). It seems reasonable to suppose that the discussion of Post. An. . , a–b, is developed in a similar ‘logical’ spirit. This suggests that, while the Posterior Analytics is not, quite generally, innocent of ontological or/and explanatory/causal considerations, it does include a number of arguments and discussions which do not require this more ‘charged’ framework. In my view, while Metaphysics Ζ  is in contact with the ‘logical’ parts of the Posterior Analytics, it does not involve the ‘charged’ positions of that work.

Essence and per se Predication in Metaphysics Ζ 



This passage involves several points which underpin the conceptual and terminological affinities already pointed out between Metaphysics Ζ  and Posterior Analytics . . More importantly, though, it argues that per se or primary predication grounds accidental predication. First, it seems clear that the kind of predication discussed in this passage is the per se of Posterior Analytics . , the one corresponding to the primacy condition of Metaphysics Ζ . Aristotle holds that substance-predicates (a: τὰ οὐσίαν σηµαίνοντα) signify precisely the certain type of thing that the subject they are assigned to is. Correspondingly, substance-attributes indicate or reveal precisely the certain type of thing that the object they belong to is (a– : ὅπερ ἐκεῖνο ἢ ὅπερ ἐκεῖνό τι σηµαίνει; cf. ὅπερ τι or ὅπερ τόδε τι at Metaph. Ζ , a–). Thus, for instance, in the predication ‘x is a man’, the predicate-term ‘man’ describes x’s substance or the certain type of thing that x is. By contrast, in accidental predications the predicate-term does not express the substance of anything: for the object referred to by the subject-term and the attribute signified by the predicate-term are different in nature from each other (a: κατ᾿ ἄλλου ὑποκειµένου; a: µὴ εἶναί τι λευκὸν ὃ οὐχ ἕτερόν τι ὂν λευκόν ἐστιν). The locutions κατ᾿ ἄλλου and ἕτερόν τι ὄν clearly imply that the type of predication described in this chapter is equivalent to per se as introduced in Posterior Analytics .  (cf. b–: µὴ καθ᾿ ὑποκειµένου λέγεται ἄλλου τινός; –: τὸ βαδίζον ἕτερόν τι ὂν βαδίζον ἐστί). Second, and more importantly, this passage vindicates my ‘predicational’ interpretation of the notion of per se as developed in Posterior Analytics . . Aristotle’s point is not that, while accidental predicates are said of certain subjects, substance- or per se-terms are not predicated of anything at all because they are ultimate sub In Posterior Analytics .  we find the same formulations as those deployed to specify the notion of per se in Posterior Analytics . , without explicit use of the καθ᾿ αὑτό terminology. Thus, for example, at a– οὐσία-predication is signified by the phrase οὐχ ἕτερόν τι ὄν. In this way, it is contrasted with accidental predication, the one marked out by ἕτερόν τι ὄν (cf. a–, –). Further, a predication is accidental in that an attribute belongs to a subject which is different in nature from it: κατ᾿ ἄλλου ὑποκειµένου (a–, –). These twin formulations, just like the ones invoked in Posterior Analytics . , are virtually equivalent to the µὴ ἄλλο κατ ᾿ ἄλλου λέγεσθαι phraseology with which Metaphysics Ζ  describes the primacy condition. Second, οὐσία-attributes (e.g. species-attributes) seem to reveal, about the subjects they belong to, the ὅπερ ἐκεῖνο or ὅπερ ἐκεῖνό τι of these subjects (a–, , –; cf. a–, ). This is equivalent to, if not identical with, the claim of Metaphysics Ζ  that essences are ὅπερ τι or ὅπερ τόδε τι (a–).



Michail M. Peramatzis

jects. This would be plainly false: for per se- or substance-terms not only signify οὐσία but are also predicated of certain subjects (a: καθ᾿ οὗ κατηγορεῖται). Rather, the idea is that per se- or substancepredicates signify something which is not different in nature from what the subject-term refers to. By contrast, accidental predicates express attributes which are different in nature from their subjects. If so, the notion of per se, just like that of primary predication in Metaphysics Ζ , not only allows for predicational structure but also involves it in a crucial fashion. Posterior Analytics . , though, adds an important point. In accidental predications, the predicate-term must ultimately be said of some subject that is different in nature from the attribute signified by this predicate-term (a–: ὅσα δὲ µὴ οὐσίαν σηµαίνει, δεῖ κατά τινος ὑποκειµένου κατηγορεῖσθαι). For there cannot be any accidental entity which is what it is, without there being some underlying predication which assigns it to a substance-entity different in nature from it (a–: ὅσα δὲ µὴ οὐσίαν σηµαίνει, δεῖ . . . καὶ µὴ εἶναί τι λευκὸν ὃ οὐχ ἕτερόν τι ὂν λευκόν ἐστιν). Let us assume the predications ‘the/this white [thing/type] is a man’ or ‘some men are white’. Aristotle’s argument is not merely that being white and being a man are different in nature from each other. He subscribes to the stronger claim that accidental predications such as ‘x is F’ necessarily presuppose an underlying essential predication ‘x is (essentially) G’ where x also happens to be (or accidentally is) F. Thus, ‘the/this white [thing/type] is a man’ requires that this thing or the type is essentially a man and happens to be white. Similarly, ‘some men are white’ entails that a subclass of the type man, comprising objects which are essentially men, consists of objects which happen to be white. This view does not deny the possibility of (true) predicational relations between substance/species-subjects and accident-predicates, between accident-subjects and substance/species-predicates  Barnes (Analytics, ) thinks that the examples given at a–, –, and –  are exclusively of singular propositions, which are not assumed or proved by any demonstrative science: ‘the/this white [thing/type] is walking’, ‘that big [thing/type] is wood’, ‘the/this wood is big’, ‘the/this man is walking’, ‘the/this white [thing/ type] is wood’, ‘the/this wood is white’, ‘the/this musical [thing/type] is white’. Independently of Aristotle’s view of demonstrative sciences, it is not necessary to take these examples as involving only singular predications whose subject-terms refer to particulars. For, as my renderings suggest, the Greek subject-terms are ambiguous between particulars (such as this particular F) and types (such as the F type or this/ that particular type, G, of F).

Essence and per se Predication in Metaphysics Ζ 



or even between accident-subjects and accident-predicates. It implies, however, that these varieties of accidental predication are necessarily supported by, and partly grounded on, corresponding essential predications, where the entities signified by the subjector predicate-terms are not mutually other in nature. This fits well with the parenthetical remark of Metaphysics Ζ  (a–), where, in so far as they fulfil the primacy condition, the relevant primaries—substances or species—do not belong to anything, or have anything belonging to them, accidentally. For they are not parts of predications where the predicational relata are different in nature from each other (they are not ἄλλο κατ ᾿ ἄλλου). Moreover, even if there are accidental predications in which they occur, these require underlying per se or µὴ ἄλλο κατ ᾿ ἄλλου λέγεσθαι predications where the primacy condition is satisfied.  Aristotle’s position about the grounding or supporting relation between primary or per se and accidental predication does not express any reductionist spirit or intention. Thus, it does not imply that accidental predications are completely parasitic upon essential ones. The claim is not that accidental predications are essentially nothing over and above underlying substance-predications. Rather, for any accidental predication ‘x is F’ there must be an underlying predication which signifies that the subject that happens to be F is essentially G, where F and G are mutually different in nature (a–: ὅσα δὲ µὴ οὐσίαν σηµαίνει, δεῖ κατά τινος ὑποκειµένου κατηγορεῖσθαι, καὶ µὴ εἶναί τι λευκὸν ὃ οὐχ ἕτερόν τι ὂν λευκόν ἐστιν). Accidental predications signify certain accidental features being true of objects, something that cannot be expressed by primary or per se predications by themselves. Nor is Aristotle’s view eliminativist: he does not hold that accidental predications are not true, while primary or per se predications alone are true. Rather, his point is that (true) accidental predications are possible partly because of corresponding primary, per se, or essential ones. Essential predications are only part of the grounds for true accidental ones as there are further features of objects, over and above their essences, which must belong to them for an accidental predication to be true. Yet essential predication might be the most important part of the grounds for true accidental predication. For true accidental predications crucially require a genuine object, or type of object, referred to by their subject-term. This condition, though, obtains only in virtue of their subject-term picking up something which is (essentially) precisely a certain type of thing (signified by a substance/species-predicate) and happens to be what the accidental predicate signifies. Take the examples ‘x is F ’, ‘the/this F [type/thing] is G’, or ‘the/this G [type/thing] is E ’, where ‘F’ and ‘G’ are accidental predicates, while ‘E ’ is essential. Aristotle’s thesis seems to be that in all these cases there is an x which is essentially E and happens to be F or G or both.  There is a possible objection to my interpretation of the primacy condition of Metaphysics Ζ  and the cognate notion of per se discussed in Posterior Analytics . . One might wonder whether primary or per se predication as I understand it collapses into the notion of per se, which Aristotle sets out at Post. An. . , a–, as follows. ‘Something holds of another thing in itself . . . if it is in that thing’s whatit-is—e.g. line of triangle and point of line (for their essence comes from these items, which inhere in the account which says what they are)’: καθ᾿ αὑτὰ δ ᾿ ὅσα ὑπάρχει τε ἐν τῷ τί ἐστιν, οἷον τριγώνῳ γραµµὴ καὶ γραµµῇ στιγµή (ἡ γὰρ οὐσία αὐτῶν ἐκ τούτων



Michail M. Peramatzis . Conclusion: weaknesses of Aristotle’s account

Understanding the primacy condition of Metaphysics Ζ  in the light of the notion of per se as specified in the Analytics suggests that Aristotle is demarcating the concept of having essence in terms of per se predication. If so, the relation between the essencepossessor and its essence is non-difference in nature. Moreover, this relation does not presuppose a further, underlying predication between entities that are mutually different in nature. Even if there is a predication which is implied by the relation between an object ἐστί, καὶ ἐν τῷ λόγῳ τῷ λέγοντι τί ἐστιν ἐνυπάρχει). Here, P is a per se attribute of a subject, S, just in case ‘S = def . . . P . . .’ and P is one/some among S’s essential attributes. The objector would argue that both in per se and in per se there is an essential connection between the object referred to by the subject-term and the attribute signified by the predicate-term: per se attributes are not other in nature than their subjects, while per se attributes are essential to (and so not other in nature than) their subjects. This is not a cautious objection. For, while a per se attribute is not different in nature from its subject, a per se attribute is only one/some among the essential features of its subject. Aristotle does not conceive of per se as being identical with τὸ τί ἐστιν but as being simply one/some item(s) ἐν τῷ τί ἐστιν or a part of τὸ τί ἐστιν (a–). Similarly, he claims that the essence of a subject is made of its per se attributes (a–: ἡ οὐσία αὐτῶν [sc. τῶν ὑποκειµένων] ἐκ τούτων [sc. τῶν καθ᾿ αὑτὰ] ἐστί) and that its definition mentions them (a–: ἐν τῷ λόγῳ τῷ λέγοντι τί ἐστιν [sc. τὸ ὑποκείµενον] ἐνυπάρχει [sc. τὰ καθ᾿ αὑτά]). But if so, x’s essence is not the same as a per se attribute of x: for a per se attribute of x is just one/some part(s) of x’s essence. If this is correct, the notion of per se, as I understand it, sets a more stringent condition than per se does: a per se attribute is not different in nature from its subject and so cannot be just one/some of the essential features of its subject. For just one/some of the essential features of a subject might still be different in nature from this subject. Take Aristotle’s own example at a: being/having a line is a per se attribute of triangle as ‘triangle =def plane closed three-lined [or: three-sided] figure’. Clearly, being/having a line is different in nature from being a triangle. By contrast, a per se attribute of triangle would be (for example) being a plane closed figure of the relevant (three-lined) sort or having three-lines of the appropriate (closed-plane-figure) sort. If so, my interpretation preserves a sharp distinction between per se and per se. It also explains why per se is important in clarifying the notion of essence: for per se attributes of x are identified with x’s essence as they are not different in nature from x. To be sure, my proposal could agree that per se and non-difference in nature can be set out by reference to per se predication if the suggestion is that all attributes which make up the essence of x are collectively taken as x’s per se attributes. For, if so, x’s per se attributes will not be different in nature from x itself. This alternative understanding of non-difference in nature as per se predication plus completeness (or collective inclusion of all per se attributes) is effectively equivalent to my account in terms of per se predication. In this alternative, though, it becomes unclear what the difference between per se and per se is unless one insists that the notion of per se denies predicational structure to per se entities. This last claim, however, seems implausible if, as I argued, Posterior Analytics .  and .  do not rule out this sort of predicational complexity.

Essence and per se Predication in Metaphysics Ζ 



and its essence, this would be a predicational connection between items that are not mutually different in nature. A first, extensional corollary flowing from this position is that substance-species are primary or per se and so essence-possessors precisely in so far as their relation to some certain attribute (or set of attributes) is non-difference in nature. A second extensional consequence is that the notion of primacy, cast in terms of per se, does not prevent compounds, entities with predicational structure, from having essence. Rather, entities which are parts of, and whose existence presupposes, a structure of subject plus attribute could be essence-possessors provided that the two items which are predicationally interconnected are not different in nature from each other. In effect, Aristotle argues that compounds cannot be ruled out as essence-possessors just because they are compounds. For there is a certain type of compounds, those which fulfil the primacy or per se criterion, that have essences. Put simply, only accidental compounds, because they are accidental, do not qualify as essencepossessors. By contrast, essential predicational compounds have essences. If this is correct, however, Aristotle appears to have returned to his starting-point in an explanatory circle. In seeking to specify the concept of x’s essence in terms of a certain type of x’s per se attributes, he has concluded that the primary or per se mode of predication is the most appropriate to capture the notion of essential connection. The primacy condition of Metaphysics Ζ , however, crucially involves the relation of non-difference in nature between x and x’s per se attributes (a–, –). If so, x and x’s essence are not different in nature from each other. The relation of non-difference in nature, though, and the characterizations based on it are simply alternative ways of labelling essential predication, a concept which is permeated by the initial explanandum, the notion of essence. Similarly, in the Analytics per se predication is set out not only by reference to non-difference in nature but, more significantly, in terms of connections where the attribute reveals the subject’s essence or precisely the type of thing that the subject is (Post. An. . , b–, –; . , a–, –). This, however, is nothing but the essential type of connection which obtains between x and x’s essence. This superficially circular explanatory procedure, though, need not render Aristotle’s position vulnerable to charges of vicious cir-



Michail M. Peramatzis

cularity. There is no reason to assume that his project in Metaphysics Ζ  is to offer a strict conceptual analysis of essencehood. Providing a simple explication of this notion would be sufficient for his clarificatory purposes within a quasi-pedagogical or didactic context. In this context, his aim is not to define essencehood on the basis of some prior, axiomatic cornerstone of his metaphysical theory. Rather, he seeks roughly to spell out this notion to fix its meaning in the present chapter and in the rest of Metaphysics Ζ. Why have any further, reductionist ambitions to produce an account in which the notion of essence would be wholly specifiable in terms of some type of per se or merely necessary predication? It seems more plausible to conceive of the relevant type of per se predication as a helpful conceptual tool with which to illuminate, in some measure, the notion of essence. This does not require us to render that conceptual tool independent of, or more basic than, the notion of essence. If this constitutes a promising strategy, there is no difficulty with Aristotle’s results at Metaph. Ζ , a–. It is not surprising that in his attempt to characterize the preferred type of per se predication (the one which captures the relation between x and x’s essence) he has to refer back to the notion of essence, which is implicit in the relation of non-difference in nature. For per se predication could not, by itself, fully explain the conceptual depth of the notion of essence. This sort of result might be assessed in two different ways. The first, more conservative view would be non-reductionist: it would hold that the notion of essence cannot be reduced to that of per se predication or conversely. Rather, the concepts of essence and per se predication are inextricably related as the one cannot be understood independently of the other. If so, it is not surprising that Aristotle’s account specifies the one in terms of the other: for the notions of essence and per se predication are not specifiable without each other. Moreover, both might be equally fundamental in Aristotle’s metaphysical picture. If so, both are to be used in shedding light on derived concepts of Aristotelian metaphysics. Alternatively, and more ambitiously for Aristotle’s essentialist position, he might be arguing that essencehood, by itself, is basic, at least in relation to the notion of per se (or even merely necessary) predication. Essencehood cannot be constructed, as it were, out of less fundamental concepts such as that of per se predication. If so, it is natural that any attempt to unpack the notion of essence

Essence and per se Predication in Metaphysics Ζ 



in terms of per se predication leads back to the relation between x and x’s essence. Equivalently, this is why the notion of essence is an indispensable part of the proper characterization of per se predication itself. This need not entail that the notion of essence or essential connection alone is absolutely fundamental in Aristotle’s metaphysical view. For there might be other notions which cannot be understood just on the basis of essencehood. Or there might be further concepts which are equally basic or even interdependently related to essence. Similarly, there might be extra constituents of Aristotle’s metaphysical theory which are even more central than essencehood in that they ground or explain this important notion. Yet, in this approach, the notion of essence is more fundamental than per se predication. It should be pointed out, at this juncture, that the enquiry of Metaphysics Ζ  is conducted at the ‘logical’ level (b: λογικῶς), which seems to centre on predicational relations: the important considerations, at this level of investigation, concern what is said of what, and in what way. If so, this disjunctive or quasi-indeterminate account of essence, in terms of either the  This view is consistent with and perhaps corroborates Burnyeat’s position about λογικῶς noted earlier. Further, it could make this position more precise, even if it does not attempt to set out in general what a logical investigation consists in. This last project might be impossibly demanding or even doomed to failure, especially if we agree with Burnyeat that the sense of λογικῶς is irreducibly relative or tied to the context in which it is embedded. For, if so, a logical investigation would take on different, even conflicting, forms depending on the context or the type of treatise in which it is conducted. A less ambitious aim is to describe Aristotle’s evidential grounds which support his particular logical results concerning essence and definition (those reached in Metaphysics Ζ ). His grounds consist of premisses about what is predicated of what, and in what way. If so, predicational parameters of this sort constitute the logical apparatus of Metaphysics Ζ . Here is the textual evidence for my view:

b: ὃ λέγεται καθ᾿ αὑτό: F is predicated in the per se mode of x. b–: ἔστι γάρ τι ὑποκείµενον ἑκάστῳ: there is a subject, x, referred to by a subject-term, that underlies each attribute, being F, described by the predicateterm. This claim does not specify the notion of essence but is characteristic of a logical discussion: for a compound is understood simply on the basis of a subjectplus-predicate or even object-plus-attribute structure. There is no mention of (for example) the alternative, hylomorphic account of compounds. b: τῶν καθ᾿ αὑτὸ λεγοµένων (more generally, see b–a): if certain attributes, F, G, etc., of compounds are their essences, they are predicated of them in the relevant per se way. a–: ὅταν δ᾿ ἄλλο κατ᾿ ἄλλου λέγεται: F and G are said of each other but differ in nature. This is the non-primary way of predication. a–: ὅσα λέγεται µὴ τῷ ἄλλο κατ ᾿ ἄλλου λέγεσθαι: F and G are said of each other but do not differ in nature. This is the primary way of predication. [Note  continues overleaf



Michail M. Peramatzis

conservative or the ambitious approaches just outlined, might be a result of conceptual limitations which obtain within the predicational framework of the logical level of discussion. Outside this framework, or in a similar framework which is properly enriched with non-predicational notions or extended beyond exclusively predicational considerations, perhaps it would be clearer which among these two approaches is more precise as a characterization of essencehood. Alternatively, it might be that neither of the two approaches is sufficiently precise: for, as I noted, there might be a further notion (or even more than one), over and above those of essence and per se predication, which is more basic than them. If so, essencehood (as well as per se predication) would be understood on the basis of this more central notion rather than conversely. At any rate, there is no reason to deem Aristotle’s method, argument, or results in Metaphysics Ζ  intolerably circular. There is a further, serious difficulty with Aristotle’s conception of essence as set out in Metaphysics Ζ . This difficulty arises not only because of the notion of non-difference in nature and the related, primary or per se mode of predication. Rather, it also grows out of the point made in the previous paragraph about the ‘logical’ level of investigation of Metaphysics Ζ : for this chapter advances its account of essence against a purely predicational background, in terms of what is said of what, and in what mode. On the basis of the primacy condition alone, Metaphysics Ζ  a–: οὐδ᾿ ὡς συµβεβηκός: F, in so far as it is primary, is not said of x in the accidental (or non-primary) way. All these fundamental considerations of Metaphysics Ζ , which consolidate its λογικῶς conclusions, are about modes of predicational connections between certain kinds of subject- and predicate-terms. Let us call this type of logical framework ‘predicational’. This interpretation could spell out and, in some measure, explain Burnyeat’s general view that logical contexts do not involve important metaphysical distinctions or presuppose the principles proper to the science of first philosophy. For example, the matter–form distinction is alien to this logical, purely or predominantly predicational, account of essence and definition. In this respect, the present account differs radically from (for example) the view described in Metaphysics Ζ –. A possible reason for this difference is that material and formal attributes are virtually indistinguishable in the predicational conceptual framework of Metaphysics Ζ . Take, for example, the definitions ‘white surface =def being white belonging to a surface’ or ‘cloak =def being white belonging to a man’. The predicational approach cannot distinguish between the quasi-material entities referred to by ‘surface’ or ‘man’ and the quasi-formal feature described by ‘being white’. It implies only that the former are underlying subjects to which the latter belong as attributes. This, however, is not the same as the metaphysically important matter–form distinction.

Essence and per se Predication in Metaphysics Ζ 



a

( –, –) introduces the following, restrictive thesis about essence and definition: (RT) Only primaries (substances or species) have essences and get defined. Aristotle, however, seems to be aware of the excessive stringency of, and lack of support for, (RT). For (RT) entails that all nonsubstances are items without any essence or definition at all. Rather, they have only significatory, non-definitional accounts in which there is merely sameness in signification between a name (or an account) and another (more elaborate) account (a–). This, however, categorizes under the same, non-definitional heading the following diverse cases: (i) accounts of accidental attributes such as: ‘being white =def colour-quality of a certain sort belonging to a surface’; (ii) accounts of accidental compounds such as: ‘white man =def being white belonging to a man’; and (iii) contrived accounts such as that in which a stipulated name signifies the same as the epos of the Iliad, the whole account consisting of the twenty-four rhapsodies (a–). It is obvious, though, that (i) and (ii) are better candidates for being definitional and for describing essences than (iii) is. Aristotle holds that this problem can be overcome by taking sameness in signification as merely necessary for having essence and definition. To develop this line, he offers (at a ff.) an alternative, more liberal thesis than (RT): (LT) Substances (as primaries) have essences and definitions in a primary or unqualified way (a, , –; b: πρώτως; ἁπλῶς), while non-substances in a secondary, derivative, or qualified way (a, , , , ; b– : ἑποµένως; πως; οὐχ ἀπλῶς; εἶτα). In (LT) the underlying criterion, over and above sameness in signification, is categorial or predicational unity. An account does not possess this sort of unity simply if it is held together by connectives or if it consists of (well- or ill-formed) continuous strings of terms (b–). Rather, an account is categorially unified if it consists of a predication that assigns one attribute to one object in some one



Michail M. Peramatzis

among the modes of the categories (b–). Clearly, this extra condition of predicational unity prevents cases such as (iii) from having essence or definition. It allows, however, cases such as (i) or (ii) to be definitions in a secondary or derivative way: for these accounts are categorially unified. If so, accidental attributes and accidental compounds have essences in this secondary or derivative way. Further, because the accounts of substances or species satisfy not only this unity condition but also primacy, they are definitions in a primary or unqualified way. Hence, substances or species have essences in this primary or unqualified way. With this distinction between primary and secondary ways in which entities have essence and definition, (LT) refines Aristotle’s views as canvassed in Metaphysics Ζ . These two alternative theses are open to an important objection. The argument of Metaphysics Ζ  seems inconclusive or even indifferent as to which is correct, (RT) or (LT). Nor is this a coincidence. Rather, the criteria for having essence and definition, as set out in the present context, are not helpful in sharply separating (RT) from (LT). If this is correct, it would be difficult to judge which of the two theses is more plausible. Let us first turn our attention to (RT), which favours primacy as the sole criterion by which to decide what items have essence and definition. In this primacy condition, substances (or species) alone are essence-possessors as they are not parts of, nor does their existence presuppose, structures which interrelate entities that are mutually different in nature. However, if this criterion is taken at face value, it cannot prevent non-substances from having essence and definition. Suppose the following (equivalent) significatory, nondefinitional accounts of the accidental attribute being white: (a) ‘Being white =def being white*’ or (a′) ‘Being white =def being a colour of white* type’, where being white* is understood as a certain type of quality belonging to a surface (after all, being white, like any other colourquality, must belong to some surface or other). If this is correct, being white does not partake in, nor does its existence presuppose, any predicational structure involving entities that are mutually distinct in nature. Thus, for instance, being white, being a colour of the appropriate sort, and being white* are non-different in nature from each other. But if so, the accident being white meets the primacy re-

Essence and per se Predication in Metaphysics Ζ 



quirement and so has essence and definition, just as substances do. This is a pressing problem particularly for the primacy requirement and the notion of non-difference in nature. It highlights that these conceptual devices are not adequate to render substances or species the only essence-possessors. Nor can they sustain (RT) in a satisfactory fashion. To be sure, the asterisked entity being white*, described by the definiens-term, partly consists in being a surface, an entity which might be seen as quasi-substantial or primary in the way in which Aristotle takes substances or species to be primary (a–, – ). This, however, poses no problems for (a) or (a′). For we could argue that, because they refer to primaries (e.g. substances or species, such as surface), the accounts of non-primaries are definitions in a derivative or secondary way. By contrast, the accounts of primaries do not include any mention of non-primaries, and so are definitions in a primary way. Clearly, though, this position is virtually equivalent to (LT), despite the fact that it is directly deducible from (RT).  If this line of reasoning is correct, (RT) and (LT) coincide in so far as both allow for substances and non-substances alike to have essence and definition (even if in different ways). This result does not seem surprising if compared with another context where Aristotle is interested primarily in predicational considerations. Thus, at Top. . , b–, he does not even envisage a position as narrow as (RT). Rather, he holds that definition (just like the rest of the predicables) can belong to any and every category, whether substance or non-substance, without even distinguishing between primary and secondary ways in which substance and non-substance items are (respectively) defined (b–). I take this claim to be implying that any entity falling under some one of the categories can be defined and so is an essence-possessor. For example, in speaking about a/the white colour, the predications ‘this is a white’ or ‘this is a colour’ express the what-it-is or the essence of the entity corresponding to the subject-term (b: τί ἐστι λέγει) but signify quality (b–: ποιὸν σηµαίνει). Similarly, the predication ‘this is a foot-long length’ expresses the what-it-is or the essence of the entity corresponding to the subject-term but signifies quantity (b– ). The phrase τί ἐστι λέγει presumably describes the ontologically non-committal mode of predicational expression ‘x is F ’ which involves the predicable of definition. In this expression x’s what-it-is is said to be F. This suggests that, in the nonsubstance cases, x does not range over (types of) objects which are essentially F as this would require x to be a substance or substance-type, with F being x’s essence. Rather, in non-substance cases, x is an attribute or a mode of being: F-ness or being F (or an instance thereof). For only in this way can x essentially be just a white or a colour, a foot-long or a length. In contrast with the ontologically non-committal phrase τί ἐστι λέγει, the locution οὐσίαν/ποιὸν/ποσὸν etc. σηµαίνει seems to reflect the ontologically stronger relation between expressions involving the predicable of definition and types of real-world entity: the expression ‘x is F’ which employs the predicable of definition signifies about x that it is (essentially) a type-F entity, i.e. a type of substance, quality, quantity, etc. Because the predicable of definition can



Michail M. Peramatzis

There are analogous problems, however, with (LT) and the extra requirement of categorial unity that this thesis invokes. These problems arise because of Aristotle’s insistence on framing his view of essence and definition just by reference to predicational considerations, merely in terms of what is predicated of what, and in what manner. If (LT) is correct, non-primaries such as accidental attributes have essence and definition in a secondary way: for their accounts are categorially unified even if they do not fulfil the primacy condition in the manner of (for example) substances. Hence, an account of being white, such as (b) ‘Being white =def a certain type of colour-quality belonging to a surface’, is a definition in the derivative way. But if so, the derivative definitions allowed by (LT) necessarily contain references to primaries or substance-like entities, such as surface. For accounts such as (b) are categorially unified in that they assign one feature to one object. As I argued in Section ., though, all categorially unified predications, including accidental ones, must be (at least partly) grounded on a primary (or per se) predication in which a substance-predicate is said of the relevant subject. For, if this were not so, a predication could not even implicitly designate any genuine (type of) object by means of its subject-term. Nor could it assign any one feature to some one object through its predicate-term. But if so, it could not be categorially unified in the way required by (LT). For this reason, a secondary definition such as (b) could not properly define a constitute expressions that signify entities in any and every category, non-substances are definable just as substances are. Similarly to (LT) as introduced in Metaphysics Ζ , essences and definitions can belong to non-substance entities too, even if their definitional accounts signify not substance but quality, quantity, etc. (cf. the occurrences of σηµαίνει at Ζ , a–; b–). Topics .  differs from Metaphysics Ζ , however, in that it does not separate the primary way in which substances are defined from the secondary way in which non-substances are. This comparison with Topics .  suggests another way in which to understand the ‘logical’, predicational framework of Metaphysics Ζ . Within this framework Aristotle is relatively indifferent to the ontological aspects of (e.g.) the distinction between substance and non-substance entities even if he does not fail to draw this distinction. Nor does he place appropriate emphasis on the relation of ontological or explanatory priority of substance over non-substance entities, a relation that latches onto this distinction. Rather, he centres on modes of predicational expression. Thus, for instance, the argument of Topics .  counts definitions, essence-signifying expressions, as being on a par with the rest of the predicables (b–). Just as the rest of the predicables belong to any and every category, similarly, the definitional, ‘what-it-is’, mode of predication can signify any and every kind of entity.

Essence and per se Predication in Metaphysics Ζ 



non-primary, accidental entity if it did not refer (at least indirectly) to a primary, substance-entity. If this is correct, (LT) may be put under pressure or even collapse into (RT). Only the accounts of substances, which are per se or primary entities, are definitions in their own right as they do not mention any non-primary, accidental entities. By contrast, the accounts of non-primary, accidental entities must refer to some primary object or other if they are to be genuinely definitional. Hence, they are indispensably (even if indirectly) dependent upon the definitions of primaries: if they do not reveal this type of dependence, they are nothing more than significatory, non-definitional accounts. This suggests that without incorporating primaries and so without presupposing (indirectly) the definitions of primaries, the accounts of non-primaries do not, by themselves, qualify as definitions, not even in any secondary fashion. Hence, (LT) may prove either vacuous or equivalent to (RT). The conclusion seems to be that the criteria of primacy and categorial unity, which are dominant in Metaphysics Ζ , are not sufficient for evaluating and comparing (RT) and (LT). Further, the exclusive focus of this chapter on predicational considerations alone cannot ground a plausible view of essence and definition. Inevitably, this prevents Aristotle’s argument from reaching definite conclusions. It may be that this sort of indeterminacy could  A similar pattern occurs in Metaphysics Ζ . At certain points, this chapter apparently favours (RT). At a– it is claimed that only substances have definitions, while non-substances have definitions only by addition (ἐκ προσθέσεως). Presumably this implies that definitions of non-substances must include the names or the definitions of substances. Similarly, accidental compounds, just like coupled items (e.g. odd-number or snub-nose; a: συνδυαζοµένων), cannot have proper definitions in their own right: the definition of (for example) cloak must refer to a substance or subject to which the attribute of being white belongs. As a states, in offering accounts of non-substances we do not notice that these accounts are not sufficiently precise (οὐκ ἀκριβῶς) to qualify as proper definitions. [The requirement of precision may be equivalent to the claim that definitions should not include additional ‘posits’ (προσθέσεις), i.e. references to subjects or substances. The discussion of ἀκρίβεια and πρόσθεσις in Metaphysics Α  (a–) and Posterior Analytics .  (a–) may be relevant in this connection. There, Aristotle holds that the definition of the unit as an indivisible quantity (without any position) includes fewer ‘additional posits’ (προσθέσεις) than that of the point as an indivisible quantity with some position. A point is defined as a unit plus some position, while a unit is defined as an indivisible quantity by itself, without any additions.] Metaphysics Ζ , however, does not rule out the possibility of adopting (LT). Indeed, at b– and a–  Aristotle seems indifferent as between the following two positions: (i) Nothing apart from substances has essence and definition (b: οὐκ ἔστιν



Michail M. Peramatzis

be overcome only by introducing additional conceptual resources, beyond predication, categorial unity and the notions of per se or primacy. Armed with richer concepts than those used in Metaphysics Ζ , Aristotle would be able to set out the notions of essence and definition more precisely and to provide support for either (RT) or (LT), or to propose some alternative, appropriately improved thesis.

APPENDIX

Isomorphism, Non-Circularity, and the Obscure Example (b–) It is useful to examine the Rule of Non-Circularity (RNC), stated at b–, and its relation to the isomorphism requirement implied by the discussion of the immediately preceding lines. Hence, an account which specifies a thing, in which [account] that thing itself is not mentioned, is the account of what-it-is-to-be that thing, so, if what-it-is-to-be white surface is what-it-is-to-be smooth surface, whatit-is-to-be white and what-it-is-to-be smooth is one and the same thing. (Metaph. Ζ , b–, trans. Bostock, slightly modified) (RNC) could be formulated as follows: (RNC) For an account, L, of x (λέγοντι αὐτό) to be an account of x’s essence, L must not mention x itself (ἐν ᾧ µὴ ἐνέσται λόγῳ αὐτό). It seems plausible to think that (RNC) is independent of but somehow derived from the isomorphism requirement (cf. the ἄρα of b). In this οὐδενός; a–: ὡδὶ µὲν οὐδενὸς ἔσται; οὐδενὶ ὑπάρξει πλὴν ταῖς οὐσίαις; a: µόνων τῶν οὐσιῶν). This is equivalent to (RT). (ii) Substances have essences and definitions mostly, primarily, or unqualifiedly (a: µάλιστα καὶ πρώτως καὶ ἁπλῶς), while non-substances only in some other way (presumably secondary, derivative, or qualified; b: ἄλλως; a: ἄλλον τρόπον; a: ὡδὶ δ᾿ ἔσται). This is just (LT).  Two points should be noted. First, (RNC) is simply a necessary condition for an account, L, to be x’s definition. For an account of x might well satisfy (RNC) and yet fail to be x’s definition: for example, a non-circular account of x might not describe the essence but only a necessary feature of x. Second, Aristotle’s formulation of (RNC) involves an innocuous use–mention confusion: the first αὐτό at b seems to refer to a linguistic or conceptual item, something included in a λόγος of x, while the second to a real-world entity, the thing defined. For this last observation see M. Burnyeat et al., Notes on Aristotle’s Metaphysics Ζ (Oxford, ), , point (iii).

Essence and per se Predication in Metaphysics Ζ 



view, (RNC) is by no means the main point of the first section of Metaphysics Ζ . Rather, this section aims to explicate a notion of per se predication appropriate to capture the concept of essence. Its conclusion is that essencehood cannot be set out on the basis of the type of per se predication which involves non-isomorphic mismatches between the object and its candidate-essence. (RNC), then, is an extra condition which stems from but is not the same as the isomorphism requirement. To show that the two requirements are not equivalent, it is sufficient to note that isomorphism does not imply non-circularity. For, given an isomorphic definiens of x or yz, it does not follow that the corresponding definition is non-circular: x =def x. yz =def yz. These are cases of circular definitions with a definiens that mirrors accurately the essential simple or complex structure of the definiendum x or yz respectively: for, while the definiendum- and definiens-terms of these formulae are isomorphic, yet the putative definitions are circular. If this is correct, the isomorphism condition is clearly not equivalent to (RNC). In what way, however, is (RNC) conceived of as following from (as the ἄρα of b suggests) the isomorphism requirement? This question becomes especially pressing once it is clear that isomorphism does not necessarily entail non-circularity. Thus, Aristotle’s move from the former to the latter cannot be deductive in any strict manner. Let us, then, suggest a looser connection along the following lines. Aristotle argues that nonisomorphism has its own peculiar problems: for non-compound objects do not have compound entities as their essences, while compound objects do not have non-compound entities as their essences. Hence, the isomorphism requirement is firmly in place. Apart from these special problems, however, some cases of non-isomorphism also entail circularity. For, given a per se predication, x−y, if the compound entity, x+y, corresponding to this predication is taken (by violation of the isomorphism condition) as a definiens of either x or y, it will yield a circular definition: x =def x+y. y =def x+y. Similarly, if the definiendum is the compound x+y, while the proposed nonisomorphic definiens is either of the simple constituents of the relevant per se predication, x−y, the resulting definition will again be circular: x+y =def x. x+y =def y.  The underlined items are repetitions of the definiendum occurring in the definiens, items that make the definitions circular.



Michail M. Peramatzis

In all these cases of non-isomorphic definitions, then, the proposed formulae are also circular. Because circularity is a defect which should be avoided by any account of essence and definition, Aristotle adds (RNC) at b–. Thus, it is not part of his proposal that in avoiding nonisomorphism one automatically or necessarily bypasses circularity too. His claim is simply that some cases of non-isomorphism entail circularity and, moreover, circularity yields incorrect definitions. If so, circularity must be avoided, just as (RNC) requires. In the light of these considerations, we can safely infer the following two conclusions. First, the argument of this section of Metaphysics Ζ  (b–) involves two separate, even if interconnected, criteria for essencehood: isomorphism and non-circularity. Second, isomorphism is the crucial move, the one which advances the specification of a notion of per se predication that is useful for understanding essencehood. For isomorphism effectively suggests that essencehood should not be set out by reference to the type of per se connection that implies a mismatch between an object and its essence. By contrast, non-circularity is a criterion which could be introduced independently of isomorphism in any context which discusses essence and definition. For, even if necessary, (RNC) is a ‘thin’, formal condition. Moreover, (RNC) does not follow the pattern of argument deployed so far as it does not make any use of the notion of per se predication to explicate Aristotle’s conception of essence. In favouring isomorphism as the central point of Aristotle’s present argument and in separating this notion from (RNC), my interpretation has a further advantage. It can make more intelligible the impenetrable example provided at b–: ‘so, if what-it-is-to-be a white surface is what-it-isto-be a smooth surface, what-it-is-to-be white and what-it-is-to-be smooth are one and the same thing’ (ὥστ᾿ εἰ τὸ ἐπιϕανείᾳ λευκῇ εἶναί ἐστι τὸ ἐπιϕανείᾳ εἶναι λείᾳ, τὸ λευκῷ καὶ λείῳ εἶναι τὸ αὐτὸ καὶ ἕν). The first obscurity is that the terms of the example do not make good sense: why is the essence of white surface held to be that of a smooth surface and the essence of (being) white the same as that of being smooth? To tackle this question, interpreters invoke a Democritean background in which smooth atoms are conceived of as constituting the surface of white objects. But it does not seem necessary to bring in this atomistic framework to understand the example. Further, there are no textual signposts which suggest the presence of this framework in the first section of Metaphysics Ζ . A second puzzle is whether the connective ὥστε introduces a conclusion which simply follows from (RNC), as the surface syntax appears to suggest. Or, by contrast, if this conclusion is inferred from the overall line of argument developed  That this Democritean assumption makes the example more intelligible is a point on which most commentators agree: cf. Ross, Metaphysics, ; Frede and Patzig, Ζ, ad loc.; Bostock, Ζ&Η, .

Essence and per se Predication in Metaphysics Ζ 



so far [but not from (RNC) alone], how does it follow? A third obscurity is whether this conclusion is a negative result, something to be avoided, which obtains if certain constraints are ignored. Or is it an admissible result reached by applying certain rules on a proposed definition which is initially deficient but still corrigible? Many commentators have understood this example as presupposing a sort of cancelling-out or elimination move: [Being a white surface = being a smooth surface] ≡ [Being a white surface = being a smooth surface] ≡ [Being white = being smooth]. This cancelling-out move, however, is potentially misleading. Aristotle himself offers an example at Metaph. Ζ , b–, which highlights the limitations of this strategy (even if it is unclear whether he himself diagnoses the cancelling-out move as the main source of the problem): if the snub nose is identical with the concave nose, then, by the cancellingout move, the snub is identical with the concave. Not only does this example fall short of identity. It does not even fulfil the coextensiveness requirement: for not all concave things are snub (although the converse is the case). It would clearly be an advantage to offer a plausible interpretation of the  It is worth noting the following two points. First, my phrases ‘being a white surface’, ‘being a smooth surface’, etc. describe essential modes of being or essential features. This corresponds to Aristotle’s use of τὸ εἶναι x, where x is a dative, a construction which denotes the essence of, or what-it-is-to-be x. Second, I use ‘ = ’ to denote numerical identity as it is possible, I think, that all these statements express identity. For there is no prima facie difficulty in asserting identity relations between predicate-expressions such as ‘being F’ and ‘being G’. Further, such expressions might well describe one and the same (numerically) way of being or feature. By contrast, it would be problematic if one held that there is identity between a referring expression, such as ‘the F object/type’, and a predicate expression, such as ‘being G’. Similarly, an object/type cannot be numerically the same as a way of being or a feature. Aristotle clearly formulates both relata of his identities in terms of (essential) ways of being described by corresponding predicate expressions.  The identity statements put forward in Metaphysics Ζ  seem to posit identity relations between the relevant types but not between modes of being or features. See b–: σιµὴ ῥίς, κοίλη ῥίς, τὸ σιµόν, τὸ κοῖλον. Aristotle uses no expressions which could signify modes of being, such as τὸ σιµῇ ῥινὶ εἶναι, τὸ κοίλῃ ῥινὶ εἶναι, τὸ σιµῷ εἶναι, τὸ κοίλῳ εἶναι.  It is unclear whether in Metaphysics Ζ  Aristotle is aware of, or interested in, the limitations of the cancelling-out move. In general, he seems to acknowledge that ‘the snub = the concave’ is a false identity statement. At Sophistici elenchi  he offers what he takes to be the correct definition of being snub: ‘being snub =def concavity in the nose’ (a–). In that context, he seems to take into account the ambiguity of the term ‘the snub’ as between ‘the snub thing/type’ and ‘snubness’ or ‘being snub’. Further, his interest lies with features or ways of being (snubness or being snub) as he claims that ‘the snub’ is not a concave nose but a certain type of affection or attribute (πάθος) of a nose. In Metaphysics Ζ –, though, he is not clear on these



Michail M. Peramatzis

example which avoids saddling Aristotle with the unfortunate cancellingout move. In my view, this result can be achieved as follows. First, the example should be read as offering a necessary condition for the initially proposed identity to be corrected into a true identity. Thus, if the identity (a) Being a white surface = being a smooth surface is to be corrected, the definition corresponding to (a) must avoid the formal mistake just pointed out in the immediately preceding lines (b–), that of circularity: ‘Being a white surface =def being a smooth surface’. For, as it stands, the definiens includes the term ‘surface’, which is part of the definiendum too. In other words, the definitional statement corresponding to (a) must be made to conform with (RNC). But if so, ‘surface’ must be deleted from the definiens. There is no mutual elimination of any definiendum-term together with a definiens-term. Rather, the idea is that the term ‘surface’ occurring in the definiendum must not recur in the definiens. Hence, the improved identity statement would run as follows: (b) Being a white surface = being smooth. The move from (a) to (b) is not explicitly made in the text. This might be seen as a drawback of my interpretation. However, this is not a fatal blow. Moreover, invoking this move seems more parsimonious than bringing in Democritean presuppositions and more promising than employing the cancelling-out move. Let us assume, then, that the transition from (a) to (b) is permissible. If we posited (b), though, we would overlook one of Aristotle’s main proposals offered in the first section of Metaphysics Ζ , the isomorphism condition: complex entities are not essences of (essentially) non-complex objects, issues. Thus, for example, the result ‘the snub = the concave’ reached in Metaphysics Ζ  is embedded in an aporia. This might suggest that he takes it as an absurd, incorrect result. Yet it is not obvious whether he would think that this absurdity obtains because of the cancelling-out move alone or because of some sort of incorrectness inherent in the initial formula ‘the snub nose = the concave nose’ (for example, on the basis of SE , a–, one might think that ‘the snub nose = the concave nose’ is ill-formed; the correct formula would be ‘the snub nose = the nose having the nasal-type-of-concavity’). If Aristotle thinks that the problem is to be located in the incorrectness of ‘the snub nose = the concave nose’, he might be conceiving the cancelling-out move as acceptable. If so, the reason why it yields an incorrect result in this particular case (‘the snub = the concave’) would be that it is applied to an ill-formed or false formula (‘the snub nose = the concave nose’). If this is correct, Aristotle would not take cancelling-out as such to be problematic but cancelling-out if/when applied to ill-formed or false formulae. My interpretation does not see Aristotle as being committed to the cancelling-out move: for I seek to disentangle his example, offered at Metaph. Ζ , b–, from this dubious strategy.

Essence and per se Predication in Metaphysics Ζ 



while non-complex entities are not essences of (essentially) complex objects. In (b), however, while the putative essence is non-compound, the object to which this essence is supposed to belong is compound. One way in which to confer isomorphism on (b) is by reintroducing ‘surface’ in its right-hand side. This, however, would lead us back to the initial, circular formula, (a). It is more promising, then, to render (b) isomorphic by simplifying its left-hand side as follows: (c) Being white = being smooth. Now (c) is explicitly stated in the text as a result of Aristotle’s reasoning. An obvious objection to (my understanding of) Aristotle’s transition from (b) to (c) would be to argue that there is no cogent reason to simplify being a white surface into being white. Why not retain being a surface and discard being white, thereby favouring the identity statement ‘being a surface = being smooth’? My reply is that identities of this last sort have already been ruled out by the argument of b–, where it is denied that being a surface is being white or being white is being a surface. If so, it is better to simplify the left-hand side of (b) in the manner of (c). For the alternative option, ‘being a surface = being smooth’, would not capture the relevant essential connection even if it were grounded on some per se necessary connection between being a surface and being smooth. As I argued in Section ., the bedrock reason for this failure would be lack of coextensiveness between ‘being a surface’ and ‘being smooth’: clearly, not all surfaces are smooth. It should be emphasized that (c) is neither favoured as a correct result nor undermined as incorrect. Aristotle does not claim that, because (a) is true and is reformulated by correct application of valid or reliable rules, (c) follows validly and is true. Rather, his reasoning implies only that deleting recurring items and observing isomorphism are necessary moves for correcting identity staments such as (a). It does not follow from this, however, that these two moves will be sufficient for formulating true identity statements. For, if they were applied to an incorrect identity statement, they could not, by themselves, be relied on to yield a true identity, just as correctly applying a valid deductive rule to false premisses could not, by itself, guarantee a true conclusion. The additional exegetical advantage of setting out Aristotle’ example in terms of (a)–(c) is that in this way the example is organically linked not only with the immediately preceding statement of (RNC) but also with the first section of Metaphysics Ζ  as a whole. For, as I understood it, this example is not only treated in the light of the non-circularity requirement but is also  It may well be that (b) fails even on coextensiveness grounds: for not all white surfaces (it might be argued) are smooth. If so, a fortiori (b) violates isomorphism.  This is so even if it is possible to construct valid deductions from false premisses to true conclusions.



Michail M. Peramatzis

aligned with the isomorphism condition. In effect, this last condition grows out of Aristotle’s attempt to cash out the notion of essence in terms of per se predication, a project which dominates the overall argument advanced at b ff. Therefore, the example of b– forms an integral part of this argument.

Christ Church, Oxford

BIBLIOGRAPHY Barnes, J. (trans. and comm.), Aristotle: Posterior Analytics [Analytics] (Oxford, ). Bostock, D. (trans. and comm.), Aristotle: Metaphysics, Books Ζ and Η [Ζ&Η] (Oxford, ). Burnyeat, M., A Map of Metaphysics Ζ [Map] (Pittsburgh, ). et al., Notes on Aristotle’s Metaphysics Ζ (Oxford, ). Charles, D., ‘Some Remarks on Substance and Essence in Metaphysics Ζ. ’, in K. Ierodiakonou and B. Morison (eds.), The Philosopher as Historian: Essays in Honour of Jonathan Barnes, forthcoming. Frede, M., ‘Plato’s Sophist on False Statements’, in R. Kraut (ed.), The Cambridge Companion to Plato (Cambridge, ), –. and Patzig, G. (ed., trans., and comm.), Aristoteles: Metaphysik Ζ [Ζ],  vols. (Munich, ). Kahn, C. H., ‘Some Philosophical Uses of “to be” in Plato’, Phronesis, . (), –. Peramatzis, M. M., ‘Aristotle’s Metaphysics Ζ : Criteria for Definition and τί ἦν εἶναι and their Relation to Posterior Analytics . ’ [in Greek], Deukalion, . (), –. Ross, W. D. (ed. and comm.), Aristotle: Metaphysics [Metaphysics],  vols. (Oxford, ). Woods, M. J., ‘Substance and Essence in Aristotle’ [‘Substance’], Proceedings of the Aristotelian Society,  (–), –.

FORM AND I N H E R I T A N C E I N AR ISTOTL E ’ S E M B R Y O L O G Y JES S I C A G E L B E R

A’s ontology posits kinds or species of objects whose members have a common essence. As traditionally understood, the Metaphysics identifies an organism’s essence, i.e. what it is, with its form. On such a view, it is natural to suppose that members of the © Jessica Gelber  For helpful questions and suggestions I am indebted to audiences at Princeton, UCLA, University College London, UC Berkeley, and members of the TOPOI Excellenzcluster in Berlin. I am grateful to Robert Bolton, Fabrizio Cariani, Alan Code, Kathleen Cook, Ursula Coope, Dorothea Frede, David Ebrey, Allan Gotthelf, Devin Henry, Joe Karbowski, Sean Kelsey, Jim Lennox, Tony Long, John MacFarlane, and Jessica Moss for reading and discussing earlier drafts of this paper with me. Thanks to Brad Inwood and an anonymous referee for their very useful ideas for improvement. For reading several versions and for countless hours of discussion, I wish to give special thanks to Andreas Anagnostopoulos and Michael Caie.  Cf. Ζ , b–: ‘By “form” I mean the essence of each thing’ (see also Ζ , b). Form is identified with the ‘substance of’ a thing in Metaph. Ζ . The ‘substance of’ a thing is what makes it what it is. What makes something a house or a human being, he says, is its form (b–). In living organisms the form is the soul, and Aristotle says in DA .  that soul is the ‘cause or source of the living body’ in three ways: ‘It is the source of movement, it is the end, it is the cause as substance [οὐσία] of living bodies’ (b–). This traditional view, whereby an organism’s substantial form and essence are identified, has been challenged in the last half-century. D. M. Balme, for instance, argues that form and essence are not identical (‘Aristotle’s Biology was Not Essentialist’ [‘Not Essentialist’], Archiv für Geschichte der Philosophie,  (), –; references are to the reprinted version with appendices in A. Gotthelf and J. G. Lennox (eds.), Philosophical Issues in Aristotle’s Biology (Cambridge, ), –). Balme distinguishes Aristotle’s use of εἶδος to refer to the essence, which ‘picks out only those features for which a teleological explanation holds’ (e.g. eyes in humans) and the ‘actualized’ form, which is qualitatively distinct for each individual and so includes all features of an organism (ibid. , ). (Balme also distinguishes these two uses of εἶδος from one that refers to the species, which is the ‘universal generalized over all animals that have the same essence, as they appear in nature’ (ibid. ).) Defending what I am calling a ‘traditional’ understanding of Aristotle’s metaphysics is outside the scope of this paper; I am here arguing only for consistency between that traditional understanding and the details of the embryology that have appeared to be in tension with it.



Jessica Gelber

same species will be the same in form. Form, for living organisms, is also supposed to play a causal role in generation. This causal role for the form of a living organism is taken up in Generation of Animals, where Aristotle explains animal reproduction in terms of the transmission of form from one generation to the next. There is, however, a prima facie tension between the thesis that form can be identified with an organism’s essence and the claim that form plays a causal role in generation. Many recent interpreters have claimed that if form is to play the causal role Aristotle assigns to it in generation, it must include features specific to the particular individual, and not just those common to all members of the species.  This is a natural assumption given claims such as the one at Metaph. Ζ , a–, that Socrates and Callias are the same in form (εἴδει), or the one at DA . , b–, that natural organisms partake in immortality in the only way they can, namely by producing something that is the same in form (εἴδει).  Aristotle makes it clear at . , b–, that this theory is meant to account for the generation of animals the same in kind (συγγένειαν) as their parents (who are the same in kind (συγγενῶν) as one other).  M. L. Gill, for instance, is persuaded that ‘in his treatment of inheritance in GA . , Aristotle builds all material accidents (such as eye color) into the individual essence of the male parent to explain their replication’ (Aristotle on Substance: The Paradox of Unity (Princeton, ),  n. ). C. D. C. Reeve claims that ‘species form seems far too thin to explain the inheritance of specific traits’ (Substantial Knowledge: Aristotle’s Metaphysics [Knowledge] (Indianapolis and Cambridge, ), ). Note that this issue about whether an organism’s form is qualitatively distinct from the form of other members of the same species is a separate issue from the notoriously controversial issue about whether form is a universal—something predicated of many instances—or whether there are numerically distinct forms for every individual. That these issues are distinct has been noted by R. Sharples, ‘Species, Form, and Inheritance: Aristotle and After’ [‘Aristotle and After’], in A. Gotthelf (ed.), Aristotle on Nature and Living Things: Philosophical and Historical Studies Presented to David M. Balme on his Seventieth Birthday (Pittsburgh and Bristol, ), , and J. Cooper, ‘Metaphysics in Aristotle’s Embryology’ [‘Metaphysics’], Proceedings of the Cambridge Philological Society,  (), –; references are to the reprinted version in D. Devereux and P. Pellegrin (eds.), Biologie, logique et métaphysique chez Aristote (Paris, ), –. (Proponents of the particular-forms view might appeal to GA .  b–, where the particular individual (τόδε τι) is said to be the proper moving cause as well as proper end of generation, ‘for what is made is some οὐσία and a particular [καθ᾿ ἕκαστον]’. I take it that in this passage he is using οὐσία in the sense he does at Cat. , a–, for what is particular (ὁ τὶς ἄνθρωπος, ὁ τὶς ἵππος) as opposed to secondary substances that are general. That Aristotle thinks the particular (καθ᾿ ἕκαστον) exerts a stronger influence in generation than the kind makes sense in the light of the ‘particular–particular’ rule from Metaph. Λ , a–: ‘For it is the individual that is the originative principle of the individuals. For while man is the originative principle of man universally, there is no universal man, but Peleus is the originative principle of Achilles, and your father of you, and this particular b of this particular ba, though b in general is the originative principle of ba taken without qualification’ (trans. Ross).) I am here concerned to defend the view that form in embryology is

Form and Inheritance in Aristotle’s Embryology



Call this form that is qualitatively distinct for each individual organism ‘subspecific form’. Socrates’ subspecific form might include, for example, his snub nose. But having a snub nose is not a part of Socrates’ essence—that feature is not common to all members of the species—and so subspecific form cannot be identified with essence. Here I shall defend the view that Aristotle’s account of inherited characteristics does not require subspecific forms. In the first part of the paper I argue that there is no textual evidence for the view that Aristotle is employing subspecific form in his account of family resemblance. As we shall see, his view is that familial resemblance is not an ‘accidental’ result. Interpreters often infer from this that inherited features are non-accidental with respect to the form that the offspring receives and thus conclude that form must be subspecific. However, I argue that this conclusion is unwarranted, and show that the class of accidental features and the class of features that are due to form do not constitute an exhaustive dichotomy. Beyond there being no direct textual evidence for the thesis that subspecific, familial resemblances are due to form, this interpretation also renders Aristotle’s theory of generation internally inconsistent. According to Aristotle’s account of sexual reproduction in Generation of Animals, only males provide the form and only females provide the matter. As I discuss below, this ‘reproductive species-level form. I take this view to be compatible both with the view that form is a universal and with the view that there are particular forms.  What it means to say that these features are not accidental will be one of the issues to be discussed in what follows. In general, accidental causal relations are contrasted with per se or intrinsic causal relations; a doctor stands in a per se causal relation to the healing that he brings about by exercising his medical skill, but if that doctor also happens to be a musician, then the musician will stand in an accidental causal relation to the healing. The locus classicus for Aristotle’s discussion of accidental causes is his discussion of luck and chance in Physics . –.  For a recent example of this inference see D. Henry, ‘Aristotle on the Mechanism of Inheritance’ [‘Mechanism’], Journal for the History of Biology,  (), – at , and id., ‘Understanding Aristotle’s Reproductive Hylomorphism’ [‘Hylomorphism’], Apeiron,  (), – at . See also Sharples, ‘Aristotle and After’,  with n. , and further n.  below.  This is repeated in a number of passages throughout Generation of Animals, e.g. . , a– (where the action of the male’s semen on the female’s menstrual fluid or καταµήνια is compared to rennet coagulating milk); . , b– (where the father’s form is compared to what the doctor conveys to the patient); and . , a– (where, in giving the reason for the separation of males and females, he says that it is better for the primary cause ‘to which belongs the λόγος and εἶδος’ to be separate from the matter).



Jessica Gelber 

hylomorphism’ is an application of a principle of causal explanation that Aristotle establishes in the Physics. There he argues that explanations of change must involve an agent which imparts a form, as well as a patient which undergoes the change of taking on that form, and that these must be distinct. However, inherited traits can come from both parents, and so on the assumption that familial resemblance is due to form, maternal resemblance would also be due to form. But Aristotle cannot consistently hold that (i) the form transmitted in animal reproduction includes all the subspecific, inherited features, (ii) only the male provides form and only the female provides matter, and (iii) females also transmit subspecific features. There are two strategies for resolving this tension that defenders of the subspecific-forms interpretation tend to adopt. Some attempt to deny (ii) by offering additional textual evidence that the mother provides form, while others attempt to argue that the father is ultimately responsible for resemblance to the mother, denying (iii). Neither of these strategies for defending  I borrow this phrase from Henry, ‘Hylomorphism’. This characterization of reproductive hylomorphism leaves a number of interpretative questions open. It may or may not entail, for instance, that form and matter exhaust the parental contributions—that is, it neither says that the male contributes only form nor that the female contributes only matter.  It is arguable that we could avoid inconsistency by changing (i) to (i*): ‘The form transmitted in animal reproduction prescribes inherited, subspecific features.’ For (i*) leaves open the possibility that the offspring’s form will include only those subspecific features it inherited from the father. In this case, the features the offspring inherits from the mother might be conveyed by and due to the matter while only those that are inherited from the father are conveyed by and due to the substantial form. The triad of claims would not be inconsistent, but merely odd for the following two reasons. First, the mother contributes the same sorts of features as the male (e.g. a particular nose shape), so Aristotle would be saying that the same sort of feature will be a feature prescribed by form only if it was inherited from the father, but not when it was inherited from the mother. Further, even if we accept this asymmetry, it seems that after some finite number of generations, what the father would pass on would be only species-specific features anyway. Granted that maternal resemblance occurs fairly often, each successive generation will have fewer and fewer subspecific features prescribed by or included in its form. I shall not consider this alternative here; it is not one that defenders of the subspecific-forms interpretation adopt, as far as I am aware.  e.g. Balme, ‘Not Essentialist’; Henry, ‘Hylomorphism’; A. L. Peck (trans. and comm.), Aristotle: Generation of Animals [GA] (Cambridge, Mass., ); and J. Morsink, Aristotle on the Generation of Animals: A Philosophical Study [Study] (Washington, ).  e.g. Cooper, ‘Metaphysics’, and Reeve, Knowledge. M. Furth, on the other hand, thinks that Aristotle’s theory of reproduction simply breaks down once it is ‘confronted with some fairly apparent facts about heredity’ (Substance, Form and Psyche: An Aristotelian Metaphysics (Cambridge, ),  with n. ).

Form and Inheritance in Aristotle’s Embryology



the subspecific-forms interpretation is successful, and I think it is instructive to see why this is so. I conclude by suggesting a way to understand Aristotle’s account of inherited characteristics that denies that subspecific, inherited features are included in the form, and instead treats the conception of form in Aristotle’s embryology as one common to members of a species and so identifiable with essence.

. The interpretative problem The source of the interpretative obstacle to treating forms in Aristotle’s embryology as species-level forms is his discussion of resemblance to parents and ancestors in GA . . Earlier in Generation of Animals Aristotle introduces the theory of reproduction in which this account of familial resemblance is supposed to be embedded. The theory of reproduction treats the father and mother as ‘principles’ of generation. The father, who is the agent of the change, supplies the ‘principle of form’. The mother supplies the ‘principle of matter’, which is, according to Aristotle’s theory, menstrual fluid (katamēnia). The father conveys the principle of form in a manner analogous to that by which a craftsman conveys the form in craft production. In both cases form is conveyed by way of certain motions or kinēseis. In craft production these are  Cf. . , a–: ‘As we mentioned, the male and female may be safely set down as the principles of generation: the male as having the principle of movement and generation, the female as having the source of matter.’  Cf. . , b–. There is no adequate English rendering of κινήσεις, and so I leave it untranslated here. This is Aristotle’s general term for ‘change’, and it covers not just locomotive movements, but qualitative changes (such as that from cold to hot), quantitative changes (such as growth or diminution), and occasionally (e.g. in Physics . –) substantial change. The κινήσεις in the spermatic fluids are probably much more like the changes in a chemical reaction than locomotive movements, so in this discussion ‘motion’ is misleading, while ‘change’ is awkward. These κινήσεις are introduced because Aristotle needs to explain the embryo’s formation in a way consistent with two principles from his natural science. First, in any change the agent must be in contact with the patient: cf. Phys. . –, GC . –. Second, the agent must be in actuality what the product is potentially: cf. Phys. . –, Metaph. Θ , b–. Socrates is a human being in actuality, and so satisfies the second criterion. But Socrates does not make contact with the matter when those changes occur. His semen makes contact with the matter (the menstrual fluid in the mother), and so semen satisfies the first criterion. However, that semen is not actually a human, and does not actually have human soul. The puzzle, as Aristotle describes it there, is that the agent of the changes in the menstrual fluid that take place when the heart is being formed can be neither ‘external’ (since then there would be no



Jessica Gelber

the motions of the craftsman’s tools, while in the production of most types of animal, these are kinēseis carried in the semen. With the general theory in place, Aristotle then turns his attention at . , a–b to the following phenomena, all of which are due to the ‘same causes’: Some offspring take after their parents and some do not; some take after their father, some after their mother, both with respect to the whole body and with respect to each part, and they take after their parents more than their earlier ancestors, and they take after their ancestors more than after any chance persons. Males take after their father more, females after their mother. Some take after none of the ancestors, although they take after some human being at any rate; others do not take after a human being at all in their appearance, but have gone so far that they resemble a monster. (Trans. Peck, slightly modified)

The causal explanation Aristotle goes on to provide shows how some of the features that vary among members of a species can be systematically traced back, through a mechanism of inheritance, to those same features in their ancestors. Aristotle does not say exactly which features are inherited, but it is clear that at least some of them will be features below the level of the species—features that vary from one individual to the next. As inherited features, they are importantly different from other sorts of feature that could be called ‘accidental’. Inherited features are not, for instance, like the property of living in Athens, which has nothing to do with the reproductive process. Nor will an organism’s subspecific inheritable features include those subspecific variations that are solely due to environmental contingencies, e.g. cold winds that affect the temperature of the parents’ spermatic fluids, or the amount of menstrual fluid that was available. An example of this sort of accidental variation is thickness of hair in humans. The reason why humans have hair at all, Aristotle tells us, is that hair protects us, and so mere possession of hair can be traced contact with the menses) nor ‘internal’ (since there is nothing that is actually a living organism in that first mixture of semen and menses). Aristotle’s solution involves showing how the principle of change is in a way external and in a way internal: since it is the father’s nature that sets up those κινήσεις that carry the principle of generation, it is external, and since that principle is carried through the κινήσεις, it is internal.  Exceptions are animals that do not emit semen, and those in which males and females are not separated. Aristotle notes that in some cases the female inserts some part of herself into the male; in such cases the male transfers the κινήσεις directly. Cf. . , b–.

Form and Inheritance in Aristotle’s Embryology





back to human form—i.e. to what a human being is. However, human hair comes in degrees of thickness, and human form will not prescribe anything more determinate than an acceptable range of hair thickness. Rather than being due to human form, the determinate thickness of one’s hair is traced back to factors such as the type of moisture and degree of heat that happened to be present while the skin was forming or when the pores were opening. Thick hair, for instance, is due to the loose and thick skin having larger passages and being more ‘earthy’ together with the oily fluid that was present, since hair grows when the fluid evaporates (GA . , a–b). Whether a human has loose and thick skin is in turn due to contingent features of both the environment and the materials available during generation. Thus, variations in hair thickness are caused by environmental contingencies and not the  PA . , a– with a–. For instance, humans walk upright and so need more protection for the ‘nobler’ front side, and walking upright is traceable to human form. The need for protection also explains why humans have eyelashes on both upper and lower eyelids.  J. G. Lennox distinguishes between two ways of thinking about essential features or properties, and I am assuming that he is correct in attributing to Aristotle what he calls a ‘non-typological’ model of essentialism (‘Kinds, Forms of Kinds, and the More and the Less in Aristotle’s Biology’, in Gotthelf and Lennox (eds.), Philosophical Issues in Aristotle’s Biology, –; repr. in Lennox, Aristotle’s Philosophy of Biology: Studies in the Origins of Life Science (Cambridge, ), –). According to this model, kinds are constituted by ‘features with range’: an essential feature is to be understood as one ‘with range’ in that what members of a kind share are features that all fall within some acceptable range of ‘more and less’. This model is to be contrasted with that which countenances some set of qualitatively identical ‘basic’ or ‘stock’ features and treats variations of more and less (e.g. thinner or thicker hair in humans) as ‘add-ons’ to, rather than determinations of, the essential feature.  Of course, in organisms for which thick hair is either necessary or better for performing some essential function, the moving causal explanation of the production of that thick hair would also, presumably, be given in terms of the type of moisture and degree of heat that affects the size of the pores and type of skin. But in these cases thick hair would not be an accidental variation, but would be for the sake of the form of the organism, and so would be included in the form. This raises a question about what difference being useful for some function makes to what we might think of as the physical mechanism by which some part is produced. This, I take it, is part of a larger issue about how to understand Aristotle’s claims (e.g. in GA . , a–b) that some phenomenon occurs both because it is better (i.e. for the sake of some function) and because it happens ‘of necessity’ (i.e. is due to factors such as thickness and thinness of the jawbone). This is not the place to address questions about the relation between formal and material natures, though any treatment of these would shed light, it is to be hoped, on the contrast between features, such as thickness of hair in humans, that are solely due to ‘material’ factors and features that are due to form as well.



Jessica Gelber

organism’s form. We can think of this type of subspecific variation as the ‘by-products’ of animal generation. By contrast, in GA .  Aristotle identifies kinēseis in the reproductive fluids from the parents as the per se causes of inherited features. These kinēseis are drawn from (apo) certain potentials (dunameis) that belong to the generator ‘qua generator’, and not accidentally: I speak of each potential in this manner. The generator is not only a male but also such a male, e.g. Coriscus or Socrates, and he is not only Coriscus but also human. And in this sense some things that belong to the generator are closer and some further qua generator and not accidentally, such as being literate or someone’s neighbour . . . For this reason, movements are present from the potentials in the spermatic fluids of all such things. (. , b–, –)

As Aristotle here explains, Coriscus qua generator is a human, a male, and a particular human male, and there are corresponding kinēseis in his semen for forming parts and features that look like his. As we shall see, because of the way in which both semen and menstrual blood are formed, there will also be kinēseis in the mother’s contribution, the matter. Aristotle describes the mechanism by which the offspring comes to resemble one parent rather than the other in terms of these kinēseis prevailing or failing to prevail. The  These are the παθήµατα that Aristotle discusses in GA  which do not ‘contribute to the account of the being [πρὸς τὸν λόγον συντείνει τὸν τῆς οὐσίας]’ of the organism (. , a–b). These, he says, have causes that must be traced back to the ‘matter and source of motion’ and do not contribute to the λόγος τῆς οὐσίας. For the view that even these παθήµατα are due to the form, see Balme, ‘Not Essentialist’.  Aristotle prepares the ground for this account earlier in Generation of Animals, and so it is not merely an ad hoc addition to the embryological theory. In book  the fact that semen is a residue from the nutritive blood that is distributed to the parts of the body is said to be ‘why we should expect children to resemble their parents: because there is a resemblance between that which is distributed to the various parts of the body and that which is left over’ (. , b–, trans. Peck). He repeats this point at . , a–: ‘As semen is a residue, and as it is endowed with the same kinēsis as that in virtue of which the body grows through the distribution of the ultimate nourishment, when the semen has entered the uterus it sets the residue produced by the female and imparts to it the same kinēsis with which it is itself endowed’ (trans. Peck). And then he reminds us of this again at GA . , b–: ‘To resume then: We repeat that semen has been posited to be the ultimate residue of the nourishment. (By “ultimate” I mean that which gets carried to each part of the body—and that too is why the offspring begotten takes after the parent which has begotten it, since it comes to exactly the same thing whether we speak of being drawn from every one of the parts or passing into every one of the parts, though the latter is more correct)’ (trans. Peck).

Form and Inheritance in Aristotle’s Embryology



kinēseis that prevail will be the causes of the new organism’s body parts and organs. If the male’s kinēseis prevail, the offspring will resemble him. If the male’s kinēseis are too weak, the kinēseis from the mother will take over and the offspring will resemble her. This is meant to explain not just morphological resemblances among particular features such as nose shapes, but sexual differentiation as well. And the description of this mechanism explains why a son can look like his mother and a daughter can look like her father. Since there are kinēseis ‘drawn from’ the father in so far as he is not only male but also a particular male, in some cases the kinēseis corresponding to his gender might prevail—yielding a male offspring—while the kinēseis corresponding to his being a particular male do not. In such cases the mother’s kinēseis (presumably corresponding to her being a particular female) take over, and the result will be a son who resembles his mother. Although the details of the mechanism Aristotle describes in GA .  are obscure, it is clear from his discussion of inheritable traits that these features are not simply accidental by-products. These inherited features are not due to contingencies in the available matter or the environment, and they are not the accidental results of some other process or processes. Inherited features are the per se results of certain kinēseis, which are said to be drawn from potentials that the generator has non-accidentally qua generator, unlike being literate. Thus, inherited features are not, as some scholars say, ‘material accidents’—i.e. accidents due to the matter. Scholars  It is not clear that we should take sexual differences to be just differences in reproductive organs, since Aristotle says that whether the embryo is male or female depends on the degree of heat in the heart (since male and female are defined in terms of the ability and inability fully to concoct nutritive blood into semen). Cf. . , a–, and . , b–.  Or, conversely, the particular κινήσεις might prevail while those for being male do not, resulting in a daughter that looks like her father.  C. Witt, ‘Form, Reproduction, and Inherited Characteristics in Aristotle’s Generation of Animals’, Phronesis,  (), – at , claims that ‘one point of almost universal agreement is that the form or essence does not include accidental, material features of the object’. According to R. W. Sharples, ‘Some Thoughts on Aristotelian Form: With Special Reference to Metaphysics Ζ ’, Science in Context,  (), – at , ‘it seems that both for Aristotle and for Alexander there is in principle a distinction between what is essential to every member of a species and what is not, the latter being accidents due to the matter in each individual’. The English expression ‘material accident’ has no strict equivalent in Aristotle’s Greek, and it is unclear what sort or sorts of effect scholars mean to pick out by that expression. It appears to be used to refer to features that are not due to form, but as we shall see below, not all features that are not due to substantial form are



Jessica Gelber

tend to infer from this that those potentials that belong to Coriscus qua generator and non-accidentally must be part of his form and so part of the form he passes on to his offspring. And since some of these non-accidental potentials are clearly not species-specific, the notion of form Aristotle uses to explain inherited characteristics seems on this view to be subspecific.

. Subspecific form and accidental features As we have just seen, in GA .  Aristotle says that the potentials for inherited traits belong to the father non-accidentally, and scholars take this to indicate that those traits are part of the father’s form. I want to begin questioning this move by noting that whether or not something is accidental is relative to the subject or cause at issue. ‘Accidentally’ (kata sumbebēkos) is an adverbial expression Aristotle uses to describe two broad categories of relation—predication relations and causal ones. A predicate can apply to a subject either accidentally or per se, and something can cause or be caused by something either accidentally or per se. Aristotle’s discussions at Posterior Analytics . , . , . , Metaphysics ∆ , and Physics .  give us two general descriptions of accidental predication. First, a predicate applies accidentally to a subject if it can apply or not apply, as being seated might apply or not apply to Socrates. Second, that which is predicated accidentally is not part of the definition (which is an account signifying the essence) of the subject. Again, being seated does not apply to Socrates in virtue of what he is, essentially, and so is not part of the definition of his essence. accidental, nor are they all ‘due to matter’. In my view, the mistaken assumption that there is an exhaustive dichotomy between effects that are due to form and those that are accidents is the motivation for the subspecific-forms interpretation.  Cf. Cooper, ‘Metaphysics’, : ‘By saying that there are actually in any male animal’s sperm movements belonging to it as that individual qua father Aristotle commits himself to at least the relative particularity of that animal’s form.’ See also n.  above.  As an anonymous referee helpfully pointed out, this distinction is not as clear-cut as my treatment here might suggest. For Aristotle recognizes a difference between an attribute or property that belongs to a subject per se in the sense that it is part of the definition and one that belongs per se because it somehow follows from the definition. So, for instance, having internal angles that sum to  degrees is not part of the definition of triangle (and so not part of the form of triangle), but all triangles must have that property, and Aristotle will call this a per se accident. I am

Form and Inheritance in Aristotle’s Embryology



A cause is accidental in virtue of its standing in an accidental predication relation to the per se or non-accidental cause. For instance, a doctor is the non-accidental cause of healing in so far as having the potential to heal is predicated of the doctor non-accidentally. If the doctor, say Aesclepius, should also be a builder, then the builder would also be the cause of the healing, but only accidentally. The builder can only accidentally cause the healing since having the potential to heal applies to the builder only accidentally. The builder is an accidental cause of the healing in virtue of standing in an accidental predication relation to the potential to heal. I shall not argue for a particular view about what it is to be a per se cause here. I am taking it for granted that it is a fact about the cause specified—the builder or the doctor—that makes him the right sort of thing to bring about the specified effect. It is the fact that the builder has the potential to build a house and the doctor has the potential to heal the sick that renders them the non-accidental causes of houses and healed patients, respectively. So what is a non-accidental result with respect to one potential may be accidental with respect to another, even if both potentials reside in the same object. Making this last point allows us to see how the idea that inherited traits are non-accidental results qua generator is consistent with overlooking this point for the purposes of this discussion, since the question I am concerned to address is whether some attribute being non-accidental entails that it is essential, and not whether being accidental leaves open (as the example just given shows) that it is in some sense essential (as having angles summing to  degrees is an essential accident of triangle).  As has been noted, e.g. by Cynthia Freeland (‘Accidental Causes and Real Explanations’, in L. Judson (ed.), Aristotle’s Physics: A Collection of Essays (Oxford, ), –), Aristotle defines accidental predication and accidental causation in terms of one another, and so it is not obvious whether one or the other is primary. However, it is safe to assume here that, for example, the builder accidentally heals in virtue of being accidentally related to the doctor, and not the other way round. Cf. Metaph. ∆ , a–; Ε , b–a.  While it may be most precise to speak of the potential that the builder or doctor has as the per se cause, I shall speak of the builder and doctor as per se causes in virtue of having that potential. Further, for the purpose of illustration, I limit the relata of causal relations to objects under a description, although Aristotle will also treat events (e.g. going to the marketplace) as causal relata.  Sometimes Aristotle will say that it is not the cause that is accidental but rather the effect. If the baker bakes something tasty, and the tasty thing coincides with the healthy thing (that is, the healthy thing stands in an accidental predication relation to the tasty thing), then the healthy thing is an accidental effect of the baker. See e.g. Metaph. Ε , a–.



Jessica Gelber

the idea that substantial form is species-level form. For in general, what results non-accidentally with respect to one specification of some substance such as Coriscus is not non-accidental relative to every other specification of him. In particular, results that are nonaccidental relative to one specification need not be non-accidental relative to the specification that picks out Coriscus as a substance or specifies his substantial form. For instance, a well-tuned lyre might be a non-accidental result relative to Coriscus qua musician, but accidental qua his possessing substantial form; the potential to tune a lyre belongs non-accidentally to Coriscus qua musician, but accidentally qua substance. This should be uncontroversial whether one thinks Coriscus’ form is a species form or a subspecific one; the potential to tune a lyre is surely something that can belong or not belong to Coriscus qua substance, even if we think that his substantial form is a very determinate, subspecific form. Similarly, features of his offspring might be non-accidental relative to Coriscus qua generator but accidental qua having substantial form; the potentials that belong to Coriscus qua generator (and from which the kinēseis derive that are the per se causes of his offspring resembling him) need not be potentials that are included in his substantial form. In this way, the assumption that it is human species form that Coriscus passes on to his offspring is consistent with the fact that potentials for subspecific, inheritable characteristics belong to him nonaccidentally qua generator, and so form need not be subspecific. In the claim that the well-tuned lyre is a per se effect of Coriscus qua musician, the ‘qua musician’ was supposed to signal that the well-tuned lyre results from some potential Coriscus has in virtue of his capacity to engage in a certain sort of activity, viz. playing music. But what is the qualification ‘qua generator’ signalling in the context of a discussion of inheritable traits? After all, we might think that qua generator Coriscus contributes the principle of form, and so any feature that results from Coriscus’ generative activity is one that results from the form he transmits, which form is thus subspecific. I do not, however, think that ‘qua generator’ must be read this way.  That would follow if Aristotle said that the male parent contributes only form. However, he does not say this (but rather that the male alone contributes form), and it is clear that there are also κινήσεις transmitted. It is by way of (διά) these κινήσεις that form is conveyed. Some interpreters speak as though these κινήσεις are somehow constitutive of (perhaps the physical realization of) the father’s form (e.g. Balme, ‘Not Essentialist’, ), but I am going to argue below that a different understanding of the relation between those κινήσεις and the form is suggested by

Form and Inheritance in Aristotle’s Embryology



Consider the following analogy with the activity of teaching a language, e.g. French. The goal of this teaching activity is the student’s acquisition of the ability to use and understand French, which Aristotle would describe as a change from lacking to having some form, French, that the teacher possesses and that his activity aims to transmit. The full specification of the goal of this teaching activity will include a number of the features that will belong to the student at the end of his education, and exclude others. It will include his being able to speak and read French, for instance, but will exclude the beard the student might be wearing. Even if the student’s beard were in some way connected to the French teacher’s activity—for example, if the student decided to grow a beard because his teacher, whom he respects and wants to emulate in every way, wears a beard—it is still not an integral part of the process of learning French, and so is accidental with respect to the French teacher’s activity. The beard is not transmitted by means of the teaching process. The full specification of the goal of the teaching activity will, however, exclude some features that can be transmitted by that teaching process. For instance, the relevant form (the ability to speak and read French) will not prescribe any one particular accent, but will simply specify a range of acceptable accents. Consequently, the teaching process does not aim at the acquisition of any particular accent within that range. Still, supposing that the student will be taught how to speak French by means of mimicry and repetition, the student may come to have the same particular accent as his teacher, for instance a Parisian accent. This feature—the Parisian accent that the teacher passes on through the teaching process— Aristotle’s analogy between those κινήσεις and the movements of a craftsman’s tools that convey the form of the craft.  Sean Kelsey helpfully suggested teaching as a model for making these distinctions.  I am assuming that the change to possession of the form French is analogous to Aristotle’s description in Physics .  of the change from unmusical to musical that the man undergoes when he acquires the form µουσική. Just as there are formally unimportant but causally significant aspects of learning a language (such as the acquisition of a particular accent), there are similar aspects of learning µουσική, such as style of performance.  A defender of the subspecific-forms interpretation might offer a competing analogy according to which it is a Parisian French accent that is aimed at, on the grounds that acquisition of Parisian French form is the goal, not just French form. I discuss this alternative later.



Jessica Gelber

belongs to the teacher qua French teacher in the sense that it is something that he can pass on to the student by the process of teaching French. But since French form does not prescribe a Parisian accent in particular, that accent is not part of the French form that the teacher possesses and that his teaching aims to transmit. It is along these lines that I propose that we think of a potential belonging to the parent qua generator, i.e. with respect to the parent’s reproductive role. The particular Parisian accent belongs to the French teacher qua French teacher (and not accidentally) because that accent can be non-accidentally transmitted through the process by which French is taught. Similarly, potentials that belong qua generator differ from those that belong accidentally to that generator in that they can be non-accidentally passed on in reproduction. And just as the particular French accent can be non-accidental with respect to the teaching process despite being accidental with respect to French form (the transmission of which is the goal of the teaching process), these subspecific potentials can still be non-accidental with respect to the generative process despite being accidental with respect to species form (the transmission of which is the goal of the reproductive process).

. Subspecific form and reproductive hylomorphism So far I have argued that the fact that subspecific, inherited traits are the results of the kinēseis that are drawn from potentials that the father has non-accidentally does not entail that those traits are due to the substantial form that the father provides. Potentials that belong qua generator need not be ones that belong qua substance, i.e. as part of substantial form. Rather, I have suggested that ‘qua generator’ signals that the potential is one that can be transmitted by the generative process. The discussion of inherited traits in GA .  does not, therefore, definitively commit Aristotle to a notion of subspecific form. But neither does that discussion explicitly rule out subspecific form. In fact, the word for form (eidos) does not even occur in that chapter.  This, I take it, is the point of the contrast with accidental features that can be common to children and parents, but which Aristotle says belong accidentally qua generator. If Coriscus is someone’s neighbour, then his offspring who live with him will have the same neighbour. And Coriscus, being literate himself, is likely to have a literate child. But these are not biologically inherited resemblances.

Form and Inheritance in Aristotle’s Embryology



It is open to the defender of the subspecific-forms interpretation to point out that for all Aristotle says in that chapter, it is possible that all of the potentials that belong qua generator are included in the form passed on in reproduction. Some of those potentials would be accidental with respect to species form, but all of them would be included in the organism’s form. That is, form would include all those potentials for which there are corresponding kinēseis in the spermatic fluids. My opponent might think that in my example I should have said that the form that the student receives from the teacher is Parisian French form—that the form acquired in learning the language is subspecific—even though a Parisian accent is accidental with respect to French. However, the offspring will resemble the mother when the male’s kinēseis fail to prevail, and so the kinēseis that she provides are used during the formation of the embryo. And as Aristotle says again and again, what the mother provides in animal reproduction is matter and not form. But if the subspecific, inherited features that these kinēseis produce are part of form, then either she also contributes form (and so despite what Aristotle initially says, the mother must end up making the same sort of contribution as the male does) or she does not really contribute kinēseis for subspecific features. A common reaction among scholars is to take this apparent inconsistency as an indication of the need to qualify or amend our understanding of Aristotle’s reproductive hylomorphism. A  In addition to his many scattered remarks distinguishing the male and female roles in generation, Aristotle’s lengthy discussion in GA . – of the pangenesis view that assigned the same sort of contribution to both parents is incontrovertible evidence that his considered view is that the male and female must make distinct kinds of contributions in reproduction. Right from the beginning of Generation of Animals Aristotle identifies the female and male as the principles of generation: the male is the active principle in that he has the source of change and generation, and the female is the passive principle who supplies the matter (. , a–). In doing so, Aristotle is self-consciously satisfying constraints on adequate causal explanation that he set out in his discussions of natural science in Physics  (. , a–) and Metaphysics Ζ (Ζ , a–; Ζ , a–). In those passages he makes it clear that any analysis of change must identify not only an agent of the change but also a thing changed—the patient.  See n.  above for a brief discussion of a third option that I am not considering here.  Morsink, Study, –, for example, takes this ‘admission’ of a contribution from the mother to be a qualification of reproductive hylomorphism. Balme, ‘Not Essentialist’, – n. , tries to downplay this tension by claiming that Aristotle’s statement that the male provides form and the female provides matter is ‘only true when carefully qualified’. Henry, ‘Hylomorphism’, defends a view according



Jessica Gelber

less common yet still influential strategy involves explaining how maternal resemblance can be attributed to the form that the father provides. Let us consider these two strategies in turn.

. Does the female also contribute form? Is there any evidence that Aristotle thought that females contribute form? There are at least some respects in which the female’s contribution might be thought of as involving form. The matter that she provides—the menstrual blood—is far from inert or featureless. In the preface to his edition and translation of Generation of Animals Peck points out that the matter in the context of biological reproduction has a quite complex form. As Aristotle explains, the menstrual fluid or katamēnia, like the father’s semen, is a residue ‘cooked up’ from the blood that nourishes and constructs the organism’s body. Because it is a residue from this blood, the katamēnia has, in potential, all the parts of the living body that come to be formed out of it (. , a–). In fact, even though that residue will not be potent enough to allow her to reproduce on her own (since females are less hot than males and so the female’s katamēnia is not as well concocted as the male’s semen), Aristotle does seem to think that in some cases females can generate up to a point (. , a–). Some female animals, Aristotle claims, can make ‘wind eggs’, which are a sort of unfertilized egg (. , a–) that is nevertheless alive in some way (. , a–). The discussion of wind eggs arises prominently in GA . At the end of GA .  Aristotle concludes that for reproduction ‘among the animals in which [the males and females are separate], the female needs the male’ (a–). He then begins GA .  by asking why that should be the case: to which reproductive hylomorphism is really the thesis that the male provides sensory soul. Like Henry, Peck argues that the mother’s role is more extensive than Aristotle’s more general comments might lead us to think; in fact, the mother provides not just matter but also nutritive soul. Since a living organism’s soul is its form, this means that the mother provides at least part of the offspring’s form. Peck thinks this is evident in Aristotle’s discussion of a phenomenon he calls ‘wind eggs’. In sect.  I assess the evidence Peck offers for interpreting reproductive hylomorphism in this way.  This strategy, discussed below, is employed in Cooper, ‘Metaphysics’, and  Reeve, Knowledge. Peck, GA, xiii.

Form and Inheritance in Aristotle’s Embryology



And yet someone might be puzzled about what the cause of this is. If indeed the female has the same soul and the matter is the female’s residue, why does she need the male and not generate all on her own? (a–)

In living organisms the form is the soul. Among animals, adult females have the same sort of soul as the male. Since the female provides the matter, she seems to have both the form and the matter herself. So, Aristotle asks, why does she need the male at all in order to reproduce? His answer is given at a–: The reason is that an animal differs from a plant with respect to sensation. It is impossible for a face or hand or flesh or any other part to exist if it does not have sentient soul either in actuality or in potentiality or in some way or just simply. For it will be like a corpse or part of a corpse. If, then, the male is the agent of this sort of soul, wherever the female and male are separate it is impossible for the female to generate an animal all by herself.

Aristotle here suggests that the reason a female cannot generate a new animal all by herself is that she cannot provide sentient soul, the possession of which differentiates animals from plants. For Aristotle, even plants have soul, but there is a hierarchy of types of soul and plants have only the lowest kind—nutritive soul. Nutritive soul is the set of capacities an organism has for performing basic vital activities such as nutrition and maintenance. In addition to nutritive soul, animals have sentient soul. And humans will have not only nutritive soul and sentient but also rational soul. In the passage above, Aristotle says that males are necessary for animal generation because males provide sentient soul; being able to provide sentient soul is ‘what it is to be’ the male (a–). Still, Aristotle acknowledges that there is reason to be puzzled, particularly since females of some bird species produce what he calls ‘wind eggs’. Although these wind eggs are not alive in the same way that fertilized eggs are, they do perish, which seems to indicate that they were alive in some sense. Wind eggs are not completely devoid of life like wooden or stone eggs (a–). Peck takes this to show that the female must also contribute form: Hence, the meaning of the statement that ‘the male supplies the Form’ can only be that the male supplies that part of the Form known as sentient Soul: everything else, including nutritive Soul, can be, and is, supplied by the female. (Peck, GA, xiii) 

Cf. DA . , b–.



Jessica Gelber

Yet Peck’s suggestion that the female provides everything except that part of soul in virtue of which an organism is an animal— sentient or perceptive soul—is questionable. It is doubtful that the mother supplies the nutritive soul in an unqualified way. While Aristotle does say that wind eggs have soul, he adds that it is clear that they have soul only potentially (a). Wind eggs do not actually have nutritive soul. There are living organisms that have nutritive soul in actuality but lack sentient soul, namely plants. But Aristotle does not think that the mother makes a plant that can then become an animal once sentient soul is added. Although it is like a plant in that it has nutritive soul, the wind egg is not, strictly speaking, a plant. Wind eggs are not generated in the way that plants are, and will not develop further (as a plant would). Moreover, Aristotle says that nothing actually living—nothing ensouled (empsuchon)—is made by the mother (. , a). Thus, even if wind eggs are plant-like, an organism has nutritive soul in actuality only when it has the parts or organs that are needed to perform nutritive soul functions, which parts and organs a wind egg does not have. There may still be room to argue that this discussion of wind eggs shows that the female can sometimes contribute nutritive soul on her own. This nutritive soul would presumably not be the same sort of nutritive soul that a plant has, and might even include the potential presence of sentient soul. However, even if it were the case that sometimes females can contribute an animal’s nutritive soul, this still would not show that Aristotle thinks  In the subsequent lines Aristotle makes it explicit that he means they have nutritive soul in potential.  He does liken the embryo’s life to that of a plant at . , b–. However, in this passage Aristotle’s point is that the newly forming animal is at this stage in possession of nutritive soul only potentially, and not actually, since it does not digest its own nourishment. It is thus like a plant in so far as the plant’s ‘digestion’ takes place in the soil in which it is living. It is only in this respect that Aristotle thinks embryos are like plants. The embryo at this stage is certainly unlike a plant that has nutritive soul in actuality.  . , b–: ‘For neither has [the wind egg] come to be as a plant simply [ἁπλῶς], nor as an animal by copulation.’  Cf. . , b ff. This is also why the σπέρµα and κύηµα have nutritive soul only potentially at . , b–: ‘Well then, it is clear that the σπέρµατα and κυήµατα which are not yet separate on the one hand have nutritive soul potentially, but on the other hand do not have it in actuality until, just like the separated κυήµατα, they draw in nourishment and do the work of this sort of soul.’

Form and Inheritance in Aristotle’s Embryology



this is what usually happens or that it happens among animals other than birds; more argument is needed for the claim that her role always extends as far as contributing something that actually has nutritive soul. Moreover, even if it could be shown that the mother’s contribution—the katamēnia—is something that has form, there is a more general worry about inferring from this that she provides part of the offspring’s form. This inference makes use of too crude a picture of what it is to be the matter for some change. As Aristotle conceives of matter, it is not just physical stuff of the sort studied in materials science departments today. Matter is one of the four causal factors Aristotle introduced in the Physics, where it is called the cause as ‘that out of which something comes to be as a constituent’ (. , b–). Bronze, for example, is the matter of the statue that comes to be formed out of it. What counts as the matter for any given change depends upon what that change is a change into. Aristotle calls that which the change is a change into in this sense the ‘form’ for the change. As Aristotle tells us explicitly in the Physics, ‘matter is relative; for there is different matter for a different form’ (. , b–). For instance, clay might be the matter for the bricks, while bricks are the matter for the house, but clay is not, strictly speaking, the matter for the house. The clay takes on brick form and the bricks take on house form, but the change into a house is not a change of which the clay is the matter. So whether something is the matter for some change or not depends upon what the form of the change is. That which plays the role of the matter for a change will itself be something of some sort, a hylomorphic composite of form and matter. Thus, the matter for a change will have some formal cause of its own, as well as features and properties that belong to  Cf. Metaph. Θ , b–a. The point that clay is not, properly speaking, the matter for the house might alternatively be put by saying that bricks are potential houses to a greater degree than is the clay that the brickmaker uses to make the bricks. The matter for a change must have the form potentially, and it is common Aristotelian doctrine that there are grades of actuality and potentiality. Cf. also DA . , a–b. This Aristotelian doctrine is discussed in A. Code, ‘Soul as Efficient Cause in Aristotle’s Embryology’, Philosophical Topics,  (), – at –. As Code points out, both the form and the matter for the developing embryo can exist at varying levels of passive and active potentiality, respectively. What I want to emphasize here is that the matter the mother provides can be at a high level of passive potentiality without thereby playing an active, rather than passive, role.



Jessica Gelber 

it non-accidentally. But that does not imply that it cannot play the role of matter in a particular change. Bricks, for instance, have form, and they have features or properties that can be manifest in the completed house (e.g. their colour or texture) without thereby making a ‘formal’ contribution to the house in the sense of contributing the form of the house. Aristotle may have thought that the brickmaker provides form—brick form—to the clay the bricks are made from. But he does not think that for this reason the brickmaker provides part of the house form; this is what the housebuilder provides. Similarly, the mother’s role is to provide the matter for animal generation, and this matter is quite complex—perhaps even to the point of having nutritive soul in a qualified way. But this does not change the fact that the mother is contributing the matter for the substantial change, and not form. When Aristotle says that the mother is the passive element that provides the matter and the father is the active element that provides the form, he is distinguishing their roles in the substantial change. To say that ‘matter qua matter is passive’ (GC . , b) is to say something about how matter contributes to some change, and not what matter contributes. Aristotle’s discussion of wind eggs does not provide the textual evidence needed for rejecting or amending his repeated claim that the female does not contribute form. First, the textual evidence adduced is questionable, since it is not clear that Aristotle thinks that wind eggs are actually ensouled. Second, even if they were, his views about the relation between matter and form would not preclude the mother from providing solely the matter for animal generation, even if that matter were ‘informed’ to a high degree. If those kinēseis in the female’s katamēnia are the per se causes of those inherited features by which the offspring resembles its mother, then those features cannot be due to the form.  If Aristotle believed in ‘prime matter’, it would only be in most cases that the matter is a hylomorphic composite.  Even if it were the case that wind eggs had nutritive soul unqualifiedly—and I do not think they do—that would not blur the distinction between the female’s contribution and the male’s. In this case, what would be potentially an animal, in need only of contact with the appropriate active potential, would be not her menstrual fluid but that wind egg that she makes. She would still provide only the matter for the substantial change.  Cf. GA . , b–: ‘Of course the female qua female is passive, the male qua male is active.’

Form and Inheritance in Aristotle’s Embryology



. Is the male responsible for maternal resemblance? An alternative strategy for resolving the tension between maternal inheritance and subspecific forms is to deny that the kinēseis responsible for maternal resemblance come from the mother. Rather, we might suppose that on Aristotle’s view, reproductive hylomorphism is supposed to go all the way down to those inherited characteristics: the father is responsible for all the features of its offspring, including those features by which the offspring resembles its mother and maternal ancestors. Consequently, since all those features that are passed on to the offspring are due to the male, there is no tension between reproductive hylomorphism and the interpretation of forms as subspecific. If we are to adopt this view we must explain how the father is responsible for all those features, in particular for those that seem to be traced back to kinēseis in the mother’s menstrual fluid. One influential strategy, which John Cooper has employed, is to appeal to GA . , a–, where Aristotle says that the mother’s kinēseis, as well as those of the ancestors, are present in potential: Some of the movements are present [eneisi] in actuality, and some in potential; in actuality are those of the generator and of the universal, such as human and animal, in potential those of the female and of the ancestors.

Cooper reads this passage as claiming that the kinēseis responsible for maternal resemblance are potentially present in the male semen. This is a controversial reading, and the text leaves open two more plausible options. These kinēseis could be potentially present in the menstrual fluid or in the embryo, rather than the male’s semen. It is not likely that Aristotle thought that there was a ‘physically realized representation of the movements of the females he can copulate successfully with (and their ancestors)’ in the male’s semen. Cooper suggests, more plausibly, that what Aristotle means when he says the maternal kinēseis are in the male’s semen potentially is that the male’s kinēseis can ‘elevate’ to the level of actuality the female’s kinēseis, which are present only potentially beforehand. Although there are physical kinēseis in the mother’s katamēnia, these are,  Cooper, ‘Metaphysics’,  n. , claims that this ἔνεισι refers back to σπέρµασι at . , b, and that there Aristotle must be referring to the male’s spermatic fluid.  Ibid. .  ‘On this conception, the semen would be said to have these movements poten-



Jessica Gelber

on Cooper’s proposed reading, in the semen potentially in that the semen has the power to make those kinēseis actually present in the offspring’s blood. The kinēseis of the female and ancestors are only potentially present until they become actual movements by the father’s agency, and so in this way the father is responsible for those maternal-resembling features in the offspring. Consequently, maternal resemblance creates no problem for maintaining both that forms passed on in animal reproduction are subspecific and that Aristotle endorsed reproductive hylomorphism. The intuition behind Cooper’s interpretation seems to be that since the father is the primary agent of the whole process of generation, all of the results of the process must be due ultimately to him. Cooper is certainly right that the father, and only the father, is the agent who contributes form. However, as I shall go on to show, this does not entail that all the results of the generative process are due to the form that the father provides. In the next section I provide an alternative interpretation of Aristotle’s account of inherited characteristics that is consistent with reproductive hylomorphism and allows those kinēseis that the mother provides to be the per se causes of those features by which the offspring resembles her. The interpretation I offer, moreover, does not require that the form passed on by the father be subspecific.

. Agents, patients, and tools ‘Matter’ and ‘form’ often refer to the passive and active causal factors, respectively, that Aristotle thinks are involved in any tially, just in virtue of the fact that it is capable of making the embryo have them as movements of its form—despite the fact that the semen does not impose them, in the sense of transferring from itself movements already actually or virtually existing in it, so much as simply work to strengthen movements provided by the mother in the catamenia’ (ibid.).  Reeve offers a slightly different strategy for attributing those kinēseis to the father that makes use of Aristotle’s claim that females are deformed males. According to Reeve, the kinēseis in the female’s menses alter and so deform the male’s movements, which are transmitted to the offspring: ‘Generalizing, we can say that whenever a movement deriving from a male form is altered or deformed by the natural tendencies in the female menses, the resulting fetus will itself be deformed (GA . , a–b). But it will be deformed, as opposed to having an undeformed form contributed by its mother, precisely because it is always the father who contributes the actual movements that concoct the menses’ (Reeve, Knowledge, –).

Form and Inheritance in Aristotle’s Embryology



change. In Aristotle’s embryology, when he says that the mother is the passive element that provides the matter and the father is the active element that provides the form, he is distinguishing their roles in the substantial change. This is evident in his association of father and mother with agent (that which acts) and patient (that which is acted upon), respectively, throughout Generation of Animals. The importance of their having these separate roles underpins arguments Aristotle gives both for and against particularities of his own and his predecessors’ views. An example of this is his argument that the female does not emit sperma like the male, and so the offspring is not a mixture from both parents’ spermata. At the end of a lengthy discussion (spanning GA . –) about the nature of the residue emitted by the female, he concludes that she does not emit sperma as the male does. The ‘universal’ (katholou) reason Aristotle gives for this is as follows: For it is necessary that there be a generator and that out of which, and even if these should be one, at least they must differ in eidos and the logos of these must be different. But in those [organisms] having separate dunameis both the body and the nature of what is acting [poiountos] and what is being acted upon [paschontos] must be different. If then the male exists as mover and agent, and the female exists as patient, the female would contribute to the semen of the male not semen but matter. Which very thing also apparently happens. (GA . , a–)

The reason females cannot contribute the same sort of spermatic fluid as the male does, Aristotle here explains, is that they have different roles in the production of the new organism; the male is the agent and the female is the patient. This is primarily how Aristotle is thinking of the respective contributions of the male and the female in animal generation. According to Cooper, since the father is the active factor, any feature that is produced by the reproductive process must be due to him. Cooper asks us to imagine a sculptor trying to shape some stone that is too soft and thus too difficult for that sculptor to manipulate precisely. Consequently, some features of the finished statue  Aristotle uses the same vocabulary of agent and patient (ποιοῦν and πάσχον) as he does in GC . – to describe the roles of mother and father throughout Generation of Animals, e.g. . , a–; . , b–.  Aristotle thinks that the ‘male’ and ‘female’ δυνάµεις are ‘mixed together’ in plants and hard-shelled organisms (ὀστρακόδερµα); that is, there are not male and female organisms, but each individual has both sorts of principle. Cf. GA . , a–, b–; . , a–.



Jessica Gelber

will not be what the sculptor intended. Cooper thinks that those unintended features should still be attributed to the sculptor’s art: Then whatever features of shape, surface texture, etc., the resulting statue has will have been the product of his art: his art will have been the originating source, and the only originating source, of these outcomes (assuming nothing pushes his hand or falls on the statue while he is working on it that affects these features). The stone itself contributes only as matter, not as a source of any of the changes it undergoes while these outcomes are being achieved. (Cooper, ‘Metaphysics’, –)

We might concede to Cooper that there is some sense in which the sculptor’s art is the ultimate source of all the features in the statue. But the sense in which this is so does not entail that all those features are part of the sculptor’s art. Seeing why this is will provide us with the materials for an interpretation of Aristotle’s account of inherited characteristics that employs species form. The sculptor’s art is, in the language of GC . , a ‘first agent’ (which acts without itself being affected). However, the art, when exercised, must be exercised in a particular way, and with particular tools and techniques. These tools and techniques are ‘last agents’ (which make contact with what they act upon, and are therefore also affected). In general, ‘last agents’ can have per se effects that are distinct from the effects of form, the first agent. For example, the doctor will use food or drugs as tools or instruments by which the form of health is conveyed to the patient. Those instruments might consist of a special diet (e.g. of raw foods and cold liquids) that aims at reducing the temperature of the patient’s blood. The determinate reduction of temperature will be the per se result of the diet. But that medical expertise, the knowledge that the doctor  ‘The same account must hold for acting and suffering as for being moved and moving. For ‘mover’ is said in two ways: that in which the principle of motion exists is held to be a mover (for the principle is the first of the causes); and again, the last [mover] towards the thing being moved and the generation is held to be a mover. Similarly with ‘agent’; for we say that both the physician and the wine heal. So, nothing prevents the first mover in a change from being unmoved (and in some cases this is even necessary), while the last [mover] always moves by being itself moved. Further, in action the first [agent] is unaffected, but the last itself suffers. For as many things as do not have the same matter, these act while being unaffected (e.g. the medical skill, for this is affected in no way by the thing being made healthy while producing health), whereas the food is also affected in a way while producing [health]—for it is heated or cooled or affected in some other way at the same time as it is producing [health]. The medical skill is the principle, and the food is the last [agent] and thing in contact [with what is acted upon]’ (a–b).

Form and Inheritance in Aristotle’s Embryology



has in virtue of which he is said to have that technē, can still be and aim at something general, namely health. Similarly, we can concede that the sculptor’s art is the first agent of the results of the sculpting process without inferring that the sculpting art aims, per se, at all the results of the process; in particular, the art need not prescribe those results of which the particular tools and techniques used in bringing about the statue are the per se causes. Tools can have per se effects that are far more determinate than the proper aim or goal of the first agent that uses them. If this is right, the kinēseis from the female (as well as those from the male) can be the per se causes of features in the offspring, even though only the male provides form. For what Aristotle says about the kinēseis is that they are ‘tools’: . . . as the products of art are made by means of the tools of the artist, or to put it more truly by means of their movement, and this is the activity of the art, and the art is the form of what is made in something else, so is it with the power of nutritive soul. As later on in the case of mature animals and plants this soul causes growth from the nutriment, using heat and cold as its tools (for in these is the movement of the soul), and each thing comes into being in accordance with a certain formula, so also from the beginning does it form the product of nature. (GA . , b–, trans. Platt)

Aristotle tells us in this passage that the kinēseis (‘in’ heat and cold) are used by nutritive soul in a manner analogous to the way that tools are used in craft production. Nutritive soul is the capacity an animal has to engage in various vital activities such as digestion, growth, and reproduction. These activities are, on Aristotle’s view, primarily achieved by means of concoction: food is concocted into blood, blood is concocted into parts and organs,  Cf. NE . , b–: ‘But individuals can be best cared for by a doctor or gymnastic instructor or anyone else knowing universally [καθόλου] what is good for everyone or for people of a certain kind (for the sciences both are said to be, and are, concerned with what is common [τοῦ κοινοῦ]); not but what some particular detail may perhaps be well looked after by an unscientific person, if he has studied accurately in the light of experience what happens in each case, just as some people seem to be their own best doctors, though they could give no help to anyone else. None the less, it will perhaps be agreed that if a man does wish to become master of an art or science he must go to the universal, and come to know it as well as possible; for, as we have said, it is with this that the sciences are concerned.’ See also Rhet. . , b–: ‘No art [τέχνη] considers the individual. The medical art [ἰατρική], for instance, [does not consider] what is healthy for Socrates or Callias, but [what is healthy] for this sort or these sorts (for this is in the province of art [ἔντεχνον], but the individual is indefinite and not knowable).’  See also, for example, . , b–; . , a–b.



Jessica Gelber

and excess nutritive blood is concocted into semen and katamēnia. Concoction endows the blood with kinēseis—a series of heatings and coolings, similar to a chemical process, perhaps—which are used in the growth and maintenance of the parents’ bodies. Since the generative fluids are residues from the concoction of blood, the same kinēseis that were present in that blood will be present in their generative fluids and passed on to the offspring, where they will be used in the formation of its body. In this way, the kinēseis function as tools not only in reproduction but also in the performance of other nutritive soul activities. Concoction, like any vital activity, requires the presence of an organ by which the organism can engage in it. This is why, Aristotle says, the heart must be formed first: the heart is the organ by which nutritive soul activities are primarily exercised, and nutritive soul must be present first since there must be a source or principle (archē) of the subsequent arrangement of the animal’s body. Once the rudimentary heart is formed by the initial action of the semen upon the katamēnia, the kinēseis from both parents are available for use as tools in the generation of the new organism. And, as we saw in the discussion of the passage at . , b ff., there are kinēseis not only for more general traits (e.g. animal or human traits), but also for the very determinate, subspecific features by which offspring resemble their families more than other members of the species. And this is as it should be, since Aristotle thinks the development of the embryo proceeds in stages from most general (first it is only an animal) to most particular (the particular shape of nose, perhaps). Since the kinēseis are used as tools to construct the body at each stage, there are kinēseis for features at all levels of generality. Thus the kinēseis the female provides can be the per se causes that ‘fashion and shape’ the embryo; although only the father is the agent of generation, all of the kinēseis, both those from the father and those from the mother, can be tools.  Aristotle tells us long before the discussion in GA .  that this fact about σπέρµα—that it is formed from nutritive blood and so has those same motions in it—is what explains family resemblance. See . , a–; . , b–.  See GA . , b–a.  See . , b–. Clearly there is a question about what it means to be ‘only an animal’ and so also about what those κινήσεις for more general traits would be for. One option might be that an organism is an animal only when it has a heart (or the analogous organ in bloodless organisms), and so a κίνησις for a more general trait might be one that is the per se cause of the heart.  I am taking Aristotle’s language quite literally in treating the κινήσεις as tools

Form and Inheritance in Aristotle’s Embryology



Now what goes for tools generally should also go for these kinēseis used by nutritive soul. And what I have been suggesting goes for tools generally is that they normally have per se effects that are much more specific than the effects at which the ‘first agent’ that uses them aims. Like a technē, the father’s nature or soul is a first agent; like other tools, those kinēseis are last agents. The per se effects or aims of a first agent (the soul or the technē) responsible for some change usually do not include all of the effects of the last agents. In so far as the aim of the first agent can be characterized generally as the imparting of form (whether the soul of the living organism or the form of health), it does not follow that every effect of the last agents is due to the form. Consequently, Cooper’s inference from the fact that the male is the agent to the claim that the male contributes maternal features need not be made. If we were forced to accept the subspecific-forms interpretation, we might be inclined to make some such move to accommodate maternal resemblance, rather than deny reproductive hylomorphism in the way that Peck does. However, as I argued in Section , we are not so forced. Moreover, in this section I have sketched an alternative way to accommodate Aristotle’s account of inherited characteristics within the hylomorphic framework in which the theory of generation is given. distinguishable from the form or soul that uses them. Alternatively, it might be thought that these κινήσεις are, for Aristotle, just what having soul amounts to. (In this spirit Balme speaks of the κινήσεις in the seminal fluids being potentially the offspring’s soul: see n.  above.) A virtue of that alternative picture is that it sidesteps any need to explain how soul and those physical κινήσεις are related, which is a challenge for anyone who takes literally talk about soul as an agent using κινήσεις. On the alternative view, soul just is reducible to those κινήσεις. It is not obvious that this reductive picture is Aristotle’s, however. Soul or nature is a δύναµις, and is a principle or source from which κινήσεις arise (Metaph. Θ , b–). The soul is not a κίνησις. Of course, there is no distinct physical entity that is the soul. But this should not threaten the conceptual distinction between soul and body.  The idea that κινήσεις, which are ‘in’ or dependent on heating and cooling, have their own per se effects is implied by . , b–a, where Aristotle is making the point that heating and cooling alone are not sufficient to make any of an organism’s functional body parts, despite their being sufficient to produce certain πάθη such as hardness or brittleness.  This rough picture requires much smoothing out. In particular, it must be supplemented by a story about what it means for the soul or nature to ‘use’ those κινήσεις as tools. Aristotle does not describe anything in animal reproduction analogous to the craftsman who is holding the hammer or the doctor prescribing diets. This, I take it, is part of a general question about what Aristotle means when he speaks of the soul ‘using’ the body, or how it can be the source of movement in the



Jessica Gelber . Conclusion

I began by noting that there is a putative inconsistency between the idea that form is essence and the idea that form is the moving cause of animal generation. The form identified with essence in Aristotle’s Metaphysics seems to be one shared by members of a species, but many scholars have held that Aristotle’s account of inherited characteristics shows that he is using subspecific forms in embryology. I have argued that nothing Aristotle says in Generation of Animals constitutes definitive evidence that forms in embryology are subspecific. Moreover, we saw that the subspecific-forms interpretation creates problems for the internal consistency of Generation of Animals that are not easily resolved. If the form that the father transmits in sexual reproduction is subspecific, there is a tension between Aristotle’s reproductive hylomorphism and the idea that the mother contributes subspecific features. I have considered two ways that interpreters have attempted to resolve this tension and argued that both involve assumptions that are questionable, given Aristotle’s other commitments. In the last section I outlined an interpretation of the account of inherited traits that avoids the tension between reproductive hylomorphism and maternal resemblance. The male provides form, and the female provides matter, but they both contribute kinēseis. The kinēseis, and not form, are the per se causes of inherited traits. Since they are tools used in the process of generation, kinēseis can have per se effects distinct from the effects of the first agent who imparts form. This body without itself being moved, which is a subject that needs separate treatment. While this does need to be answered somewhere, I do not think that answer is to be found in Generation of Animals. There he speaks of the soul using the body (or the κινήσεις, which are in the body), but does not try to explain what that means. J. G. Lennox discusses Aristotle’s ascriptions of agency to soul, an organism’s formal nature, in ‘Material and Formal Natures in Aristotle’s De partibus animalium’, in id., Aristotle’s Philosophy of Biology: Studies in the Origins of Life Science (Cambridge, ), –, where he notes, correctly, that this should not be taken as merely metaphorical. The relation between such ascriptions of agency to nature and Aristotle’s teleology is discussed in A. Preus, Science and Philosophy in Aristotle’s Biological Works (Hildesheim and New York, ), ch. , pt. , –. I shall simply state here without defence that this is a worry that arises not only for the soul’s relation to the living body that it uses as its tool, but also for any τέχνη; it is the medical art in the mind of the doctor, not the doctor, that is in the strictest sense the moving cause of the healing. Cf. Phys. . , b–. For a discussion of this point see S. Menn, ‘Aristotle’s Definition of Soul and the Programme of the De anima’, Oxford Studies in Ancient Philosophy,  (), –.

Form and Inheritance in Aristotle’s Embryology



interpretation does not require subspecific forms, and so would allow one both to hold on to the doctrine that form is essence, and to assign to species-form a primary causal role in animal generation. University of California, Berkeley

BIBLIOGRAPHY Balme, D. M., ‘Aristotle’s Biology was Not Essentialist’ [‘Not Essentialist’], Archiv für Geschichte der Philosophie,  (), –; repr. with appendices in A. Gotthelf and J. G. Lennox (eds.), Philosophical Issues in Aristotle’s Biology (Cambridge,), –. Code, A., ‘Soul as Efficient Cause in Aristotle’s Embryology’, Philosophical Topics,  (), –. Cooper, J., ‘Metaphysics in Aristotle’s Embryology’ [‘Metaphysics’], Proceedings of the Cambridge Philological Society,  (), –; repr. in D. Devereux and P. Pellegrin (eds.), Biologie, logique et métaphysique chez Aristote (Paris, ), –. Freeland, C., ‘Accidental Causes and Real Explanations’, in L. Judson (ed.), Aristotle’s Physics: A Collection of Essays (Oxford, ), –. Furth, M., Substance, Form and Psyche: An Aristotelian Metaphysics (Cambridge, ). Gill, M. L., Aristotle on Substance: The Paradox of Unity (Princeton, ). Henry, D., ‘Aristotle on the Mechanism of Inheritance’ [‘Mechanism’], Journal for the History of Biology,  (), –. ‘Understanding Aristotle’s Reproductive Hylomorphism’ [‘Hylomorphism’], Apeiron,  (), –. Lennox, J. G., ‘Kinds, Forms of Kinds, and the More and the Less in Aristotle’s Biology’, in A. Gotthelf and J. G. Lennox (eds.), Philosophical Issues in Aristotle’s Biology (Cambridge, ), –; repr. in Lennox, Aristotle’s Philosophy of Biology: Studies in the Origins of Life Science (Cambridge, ), –. ‘Material and Formal Natures in Aristotle’s De partibus animalium’, in id., Aristotle’s Philosophy of Biology: Studies in the Origins of Life Science (Cambridge, ), –. Menn, S., ‘Aristotle’s Definition of Soul and the Programme of the De anima’, Oxford Studies in Ancient Philosophy,  (), –. Morsink, J., Aristotle on the Generation of Animals: A Philosophical Study [Study] (Washington, ). Peck, A. L. (trans. and comm.), Aristotle: Generation of Animals [GA] (Cambridge, Mass., ).



Jessica Gelber

Preus, A., Science and Philosophy in Aristotle’s Biological Works (Hildesheim and New York, ). Reeve, C. D. C., Substantial Knowledge: Aristotle’s Metaphysics [Knowledge] (Indianapolis and Cambridge, ). Sharples, R. J., ‘Some Thoughts on Aristotelian Form: With Special Reference to Metaphysics Ζ ’, Science in Context,  (), –. ‘Species, Form, and Inheritance: Aristotle and After’ [‘Aristotle and After’], in A. Gotthelf (ed.), Aristotle on Nature and Living Things: Philosophical and Historical Studies Presented to David M. Balme on his Seventieth Birthday (Pittsburgh and Bristol, ), –. Witt, C., ‘Form, Reproduction, and Inherited Characteristics in Aristotle’s Generation of Animals’, Phronesis,  (), –.

AR ISTO T E L I A N PH I L I A, MODERN F R I E N D S H I P ? ALEX A N D E R N E H A M A S

I W are we to make of a treatise on moral philosophy so much of which appears to be a discussion of friendship? What does friendship have to do with morality? From the moral writings of modern philosophers such as Immanuel Kant and John Stuart Mill, who have so little to say about it, the answer we receive is, ‘Nothing’. Yet Aristotle devotes fully one fifth of the Nicomachean Ethics to an examination of the nature and varieties of philia, which he considers essential to the good human life. This difference between Aristotle and antiquity as a whole on the one hand and modernity on the other was the starting-point of C. S. Lewis’s probing exploration of friendship in The Four Loves: ‘To the Ancients,’ he wrote, ‘Friendship seemed the happiest and most fully human of all loves; the crown of life and the school of virtue. The modern world, in comparison, ignores it.’ Friendship, he continued, is ‘the least natural of loves; the least instinctive, organic, biological, gregarious and necessary’, it is independent, even ‘defiant’ of nature: for that reason, the ancient and medieval worlds, whose ‘deepest and most permanent thought . . . was ascetic and world-renouncing’, prized this kind of love above all others. But then, Lewis lamented, came ‘Romanticism’ and, with Romanticism, the exaltation of sentiment, emotion, and instinct: and to them, the tranquillity and evenness of friendship seemed inevitably pallid and lifeless. It is difficult to know whether friendship is in fact less important © Alexander Nehamas  For their comments on previous versions of this essay, I am grateful to Brad Inwood, Anna Marmodoro, and Dory Scaltsas.  Although the Greek word applies, as we shall see, to an immense range of human attitudes and relationships, it is almost invariably translated as ‘friendship’.   C. S. Lewis, The Four Loves (Orlando, Fla., ), . Ibid. –.



Alexander Nehamas

to our lives than it was to the lives of ancient Athenians or medieval Paduans—and it is equally difficult to believe it. It is certain, though, that friendship has become less important to philosophy and perhaps to narrative literature as well—which, it seems to me, was Lewis’s real concern. But neither Romanticism nor the exaltation of sentiment can account for its decline, and neither the ancients nor the medievals were as other-worldly as Lewis thought. Tennyson’s In Memoriam (the last poem, Lewis claimed, to take friendship as its subject) is hardly short on Sentiment and the novel’s neglect is due to the problems friendship creates for the genre’s narrative structure. Rather than the exaltation of sentiment, I would say that it was the lionization of a certain kind of reason that made friendship an unlikely subject for modern philosophical thought and turned the attention of philosophy away from spontaneous relationships that belong to private life and towards our obligations to one another in general, our duties and responsibilities to the world at large. The central and liberating aim of modern philosophy has been to make every human being morally salient to every other. Its task has been to overcome, to make irrelevant, the specific differences that had seemed insuperable to earlier ages and had been used to justify the inhuman treatment of others in earlier times. It is a task that remains incomplete, in both practice and theory: ‘Christian’ and ‘Muslim’, ‘Uighur’ and ‘Han’, ‘male’ and ‘female’ are categories still powerful enough to underwrite the most cruel and discriminating behaviour, and the final pages of the moral treatise that will persuade the world to move beyond them are still unwritten— and unlikely ever to appear. But the two main schools of modern  That, of course, depends on what counts as literature: it is true of the novel, though not of the theatre, film, or television. The subject is too complicated to discuss here. I discuss it in a longer and more systematic project on friendship. I make a few comments on the novel in the next note.  Very briefly, the interactions of friends, however much they may mean to them, are usually insignificant and even boring to third parties. To establish a friendship, a novel would have to represent a considerable sequence of such incidents and, in addition, represent them as insignificant and boring to such third parties (that is, to its readers). Such a sequence is difficult to construct and results in a narrative that is bound to be boring. Bouvard and Pécuchet, as Flaubert himself realized (he complained that no one would ever finish it), is just such a novel.  See J. M. Cooper, ‘Aristotle on the Forms of Friendship’, in id., Reason and Emotion: Essays on Ancient Moral Psychology and Ethical Theory [Reason] (Princeton, ), – at .  This is part of the argument of B. Williams, Ethics and the Limits of Philosophy

Aristotelian Philia, Modern Friendship?



moral philosophy—Kantianism, for which the moral value of every action depends on whether it was done for the right reasons, and Consequentialism, which locates the action’s value in its beneficial effects—agree that to act morally in the first place we must act impartially. And to act impartially we must take into account only the characteristics that every human being shares with every other and ignore their more specific or idiosyncratic features. ‘Rational’, according to some moral philosophers, is a feature of the first kind; ‘Uighur’, of the second: Uighurs should not be treated specially simply because they are Uighurs; to treat them morally, we should abstract from their being Uighurs and consider only what we would do for any human being in that same situation. How then could friendship, which depends exactly on our treating some people differently from the way we treat everyone else simply because they are our friends, easily find a place within such a context? Philosophers of both schools have tried to devise ways to include friendship among the moral goods. But, in the end, any mode of thought that requires that we treat those closest to us as we would treat a stranger is bound to make us suspect that the unashamed preference we give to our friends is an avatar of tribalism. The relationship of friendship to morality remains uneasy and its role in moral philosophy is secondary at best. Most modern philosophers have had little or nothing to say about it. But the ancients did. They said much about it, and much of it was good. Aristotle devotes roughly a whole book of the Eudemian Ethics and two of the Nicomachean to philia—far more than to any other (Cambridge, Mass., ). See also R. M. Rorty, ‘Human Rights, Rationality, and Sentimentality’, in S. Shute and S. Hurley (eds.), On Human Rights: Oxford Amnesty Lectures (New York, ), –. It is also clear that the vocabulary of human rights can be used to justify exactly the kinds of treatment that it explicitly seeks to abolish. I am primarily concerned with the official, so to speak, aspirations of modern moral philosophy.  This bald statement requires serious qualification. Moral philosophy contains more variations—e.g. virtue theory and intuitionism—and not every moral theorist agrees that the moral point of view must necessarily be impartial. But it is the very general lines of this outlook that are important to me now. For some of the complexities of the second issue here see T. Jollimore, ‘Impartiality’, in E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy 〈http://plato.stanford.edu/archives/ fall/entries/impartiality/〉 [accessed  Apr. ].  See e.g. T. Jollimore, Friendship and Agent-Relative Morality (New York, ), for a Kantian approach and, for a consequentialist alternative, P. Railton, ‘Alienation, Consequentialism and the Demands of Morality’, Philosophy and Public Affairs,  (), –.



Alexander Nehamas

single subject. Philia was absolutely central to his ethical framework and his views have been absolutely central to all later thought on friendship. But the connection between Aristotelian philia and modern friendship is, to say the least, complex—complex enough, at any rate, to force the question whether they do in fact have anything in common. The range of relationships Aristotle examines in books  and  of the Nicomachean Ethics is extraordinary. He considers philia necessary to everyone, rich and poor, young and old, human and animal. Philia unites families and species, and holds political parties, social and religious organizations, and even whole cities together. It joins people who are equal in age or in social, financial, intellectual, or moral standing as well as others whose relationships are hierarchical—parents and children, men and women, rulers and ruled, humans and gods. It connects the ignorant to the wise and the beautiful to the ugly. Erotic love is one of its kinds. It arises among travellers, soldiers, and members of a tribe. It governs the relations between host and guest and, perhaps most surprisingly, between buyer and seller. Could this immense variety of relationships correspond to friendship, which, despite its own complexities, we tend to distinguish sharply from the ties that bind family members, lovers, and commercial agents to one another today? And what, if anything, holds these seemingly disparate bonds together? Aristotle’s answer to this last question comes in stages. First, he claims that they all involve philēsis—a word as badly translated by ‘love’ or ‘affection’ as philia is by ‘friendship’: ‘love’ is much too strong for the feelings of debtors and creditors, members of social clubs, or citizens of the same state; and ‘affection’ is much too weak for the feelings of parents for their children or lovers for one another. Philēsis is a very broad, generic attitude: it ranges from a merchant’s cool appreciation of profit to the most intense erotic passion, and it is provoked by everything that, according to Aristotle, human beings care for: practical benefit or profit, pleasure, and virtue (NE . , b–). He observes, second, that although philēsis for inanimate objects is possible, philia with them is not. Philia requires reciprocation, yet no inanimate object I find attractive—wine is Aristotle’s own 

These classifications are Aristotle’s, not mine. In what follows, I refer to passages in books  and  of the Nicomachean Ethics only by Bekker page number. 

Aristotelian Philia, Modern Friendship?



example—can ever care for me in return. It is also impossible for me to wish inanimate objects, as philia also requires, good things for their sake and not, or not only, for my own—if I wish that my wine be good, it is only so that I can enjoy it more. Aristotle calls this other-regarding attitude ‘good will’ (eunoia) and points out that its objects and the objects of philēsis are the same. I can bear you good will, and wish you good things for your sake, either because of the benefits I derive from our relationship, or because of the pleasure our interaction gives me, or finally because I am drawn to your virtues—which, in the human case, include courage, justice, temperance, magnificence, wisdom, and the other features that are necessary to make a life a good and happy one. If, then, I care for you and wish you good things for your sake, not only for mine, and if, moreover, your attitude towards me is the same (and if, Aristotle adds, we are both aware that this is so), we are bound to each other by philia and can be said to be each other’s philos (b– a). The idea that our feelings for some people may sometimes make their well-being as important to us as our very own, that for them to do well is also for us to do well, that their successes and failures are also ours, is absolutely essential to friendship as we understand it today. Since it seems essential to philia as well, it has made it tempting to think that philia is the Greek equivalent of friendship and it explains why the two have been so often identified. Of the three kinds of philia, however, Aristotle thinks that only the bond that draws people to each other’s virtue is ‘perfect’ or ‘complete’, the paradigm of friendship both for him and for the tradition that follows him. The pursuit of pleasure or benefit leads to lesser or inferior relationships, worthy of being considered philia only to the extent that they ‘resemble’ and share some of their features with the perfect sort (b–). But friendship between two people could not be based on their admiration of each other’s virtue unless both were virtuous in the first place. Aristotle did not need to be the great anatomist he was in order to have no illusions about his contemporaries: ‘Such friendships are rare,’ he conceded, ‘for there are few such people’ (b–). That seems not to have disturbed him at all, and on this issue, as on just about everything else, the philosophical tradition has agreed with him. To Cicero, it was per Φίλος, plural ϕίλοι, is usually translated as ‘friend’ (when it is used as a substantive) or ‘dear’ (when it functions as an adjective).



Alexander Nehamas

fectly natural to say that ‘all the affection in the world is shared by no more than a handful of individuals’. St Aelred of Rievaulx, whose twelfth-century De spiritali amicitia is the greatest work that Christianity devoted to friendship, followed suit: ‘Friendship, which, like charity, was first preserved among all by all, remained among the few good [after the Fall].’ La Rochefoucauld raised the stakes: ‘However rare true love may be, it is less so than friendship.’ But he, like everyone else, was trumped by Montaigne’s uncharacteristically extravagant claims for his friendship with la Boétie, ‘so entire and so perfect that certainly you will hardly read of the like, and among men of today you see no trace of it in practice’. That the bond between the virtuous is so rare would not be a problem if it were only one kind of friendship: the other two, less demanding and more appropriate for the rest of us, would then fill the gap. We must ask, then, whether pleasure and benefit produce genuine friendship—or, at least, whether they produce the kind of relationship that contemporary philosophy considers a moral or ethical good. For the fact is that our own use of ‘friendship’ and its cognates is, if not as broad as Aristotle’s, very broad indeed, and what was true in the eighteenth century is still true today: If we observe the common discourse of mankind, we shall find a friend to be one we frequently visit, who is our boon companion, or joins with us in our pleasures and diversions, or encourages us in our business, or unites in the same scheme, or votes the same way at an election, or is our patron, or dependant, who we hope will help us in rising to preferment or increasing our interest.

‘A friend can get me tickets to the concert’ is true even if we are never in touch except when I need tickets or she needs help with her Greek.  Cic. Amic. . , trans. F. Copley (Ann Arbor, ), repr. in M. Pakaluk (ed.), Other Selves: Philosophers on Friendship [Selves] (Indianapolis, ), .  Aelred of Rievaulx, Spiritual Friendship, . , trans. M. E. Laker (Kalamazoo, Mich., ), .  François, duc de La Rochefoucauld, Maxims, , trans. S. D. Warner and S. Douard (South Bend, Ind., ),   Michel de Montaigne, ‘Of Friendship’, in The Complete Works of Montaigne, ed. and trans. D. M. Frame (Stanford, Calif., ), – at .  [Abraham Tucker], The Light of Nature Pursued by Edward Search, Esq., iii/ (), ; quoted by K. Thomas, The Ends of Life: Roads to Fulfilment in Early Modern England (Oxford, ), .

Aristotelian Philia, Modern Friendship?



II Although relationships based on benefit or pleasure are not ideal, they would still be friendships as long as they, too, involved mutual affection and, more important, wishing good things for each other’s own sake: that seems the least we should ask of them. And that, as we saw, is in fact what Aristotle says about all three species of philia at the very beginning of his investigation. The problem, though, is that no sooner does Aristotle say so than he qualifies his account—perhaps he even retracts it. In the very next lines of the Ethics he writes that ‘those who care for each other wish each other good things in so far as they love each other’ (a–). So, some people wish good things for each other to the extent that they are virtuous, others to the extent that they give each other pleasure, and still others to the extent that they bring each other benefits. But when their virtue is not involved, Aristotle tells us, people do not care for each other ‘for themselves’ (kath’ hautous) but only in so far as they derive benefits or pleasure from each other. If, for example, I care for you because you can introduce me to the right people who can help me with my plans, Aristotle believes that I do not care for you because of who you are, because you are a certain kind of person (poios tis), but simply because you are beneficial to me—not for you yourself but only in so far as you are useful. Such relationships, he continues, are merely ‘incidental’ (kata sumbebēkos, a–). When we are drawn to each other’s virtues, by contrast, our bond is perfect because, for Aristotle, character is destiny: it is our character, which is manifested in our virtues, such as they are, that makes us who we truly are; it is our character, and only our character, that determines our essential nature: the rest is accident, features we can acquire or lose without changing in ourselves. And so only those who are drawn to each other’s virtues care for what the other really is, and only they are capable of wishing each other well for the other’s own sake (b–). There seems, then, to be a tension—if not an outright contradiction—between what Aristotle first says about philia and what he adds immediately afterwards. What are we to make of that? On the  The Greek expression is ταύτῃ ᾗ ϕιλοῦσιν, a construction common in Aristotle’s thought, to which we shall return.



Alexander Nehamas

assumption that he is not simply confused, three efforts to resolve that tension have been made so far. The first takes Aristotle to be speaking loosely when he first introduces his view. When he says that all three kinds of philia involve wishing our friends’ good for their own sake, he is offering not a strict but a ‘schematic’ definition of philia in general. A schematic definition is a rough approximation, a general and perhaps vague specification of its subject, with the details to be filled in later. So, roughly speaking, every kind of philia includes good will, but good will is of two kinds: one kind depends on who another person really is; the second, on what that person can do for me. In that way, the perfect philia of the virtuous, which is genuinely other-regarding, remains paradigmatic. Aristotle does seem to think that pleasure can be loved as an end (b–) but insists that philia based upon it is incidental and ends as soon as the other person ceases to provide the pleasure sought. That suggests that although its object may be pursued as an end, the philia itself is merely a means to the pleasure it affords. Since the philia that is based on benefit is clearly of that kind as well, both turn out to be instrumental relationships, relationships we have with others because we believe that they are capable of—instrumental to—satisfying our own interests and purposes. People who engage in such relationships are means to each other’s ends and not ends in their own right, as they are when their virtues are the ground of their attraction. That is the view of Michael Pakaluk: Aristotle’s conclusion is that only friendship that involves reciprocated love based on the virtues of another is a friendship in the proper sense of the term. He is half-tempted to say that no other relationship should even be called friendship, but, as a concession to ordinary usage, and by way of uncovering what is right about that usage, he concedes that any other relation may be counted as a friendship, to a greater or lesser degree, to the extent that it resembles this ideal.

There is a lot to be said, as we shall see, for the idea that the good will of the virtuous is seriously different from the good will of  See M. Pakaluk (trans. and comm.), Aristotle: Nicomachean Ethics Books VIII and IX (Oxford, ), –. Despite not agreeing with him on various issues, I believe that Pakaluk’s work on friendship in general and on Aristotle in particular is crucial to anyone who cares about the subject.  M. Pakaluk, Aristotle’s Nicomachean Ethics: An Introduction (Cambridge, ), –.

Aristotelian Philia, Modern Friendship?



the beneficial and the pleasant. Still, it is not easy to believe that Aristotle would introduce his discussion of philia in such loose terms, especially without warning that he was doing so. When that is the case, he often warns his readers that he is giving them only a rough approximation, to be explained in more detail later. More important, far from being ‘half-tempted’ to withhold even the words for the former from the latter two, he shows absolutely no hesitation in classifying all three as instances of philia. And although we might hesitate to call both Damon and Phintias on the one hand and the person who can get us concert tickets on the other ‘friends’ in the same sense, Aristotle feels perfectly free to call them all philoi. Pakaluk’s view, I believe, assumes that philia is identical with friendship and therefore attributes to Aristotle difficulties created by the idea that friendship can be an instrumental relationship—difficulties, though, to which Aristotle seems oblivious. But since what we are trying to determine is precisely how philia and friendship are related, that is just what we cannot assume. A second way of addressing the tension in Aristotle’s account begins by assuming that the lesser kinds of philia are instrumental and assimilates the relationship of the virtuous to them. It notes that Aristotle writes that even the virtuous ultimately care for what is good for themselves (b–) and that when he later repeats his claim that we wish the greatest goods for those we care for—and for their sake—he immediately adds, ‘But perhaps not exactly all: for everyone wishes the greatest goods for oneself above all’ (a– ). And it concludes that even the philia of the virtuous is selfcentred after all. But, together again with the assumption that philia is friendship, this view implies that Aristotle, and the Greeks more generally, took every friendship to be an essentially self-interested and instrumental relationship. Philia, according to this approach, signifies our bonds ‘with men and women whose good we want because they serve our need, or interest, or pleasure, and for no other reason’. Greek friendship, therefore, was not in the main ‘a subjective bond of affection and warmth, but the entirely objec Similar expressions in the Nicomachean Ethics itself: περιγεγράϕθω, ὑποτυπῶσαι, contrasted with ἀναγράψαι (. , a–); τύπῳ καὶ οὐκ ἀκριβῶς (. , a–). See also DA . , b–: τύπῳ λέγειν as opposed to διασαϕητέον ὕστερον.  G. Vlastos, ‘The Individual as Object of Love in Plato’, in id., Platonic Studies, nd edn. (Princeton, ), – at .



Alexander Nehamas

tive bond of reciprocal obligation’. As such, it was radically different from, and almost certainly inferior to, friendship as it was reconfigured with the emergence of the market and the separation of personal from impersonal relations during the eighteenth century, when, in the words of the sociologist Allan Silver, it came to be ‘grounded in the unique and irreplaceable qualities of partners, defined and valued independently of their place in public systems of kinship, power, utility, and esteem, and of any publicly defined status’. Finally, those who believe that this way of thinking makes the Greeks far too cold, calculating, and impersonal propose a directly opposing interpretation of Aristotle’s remarks on good will. Where the previous approach tended to assimilate the philia of the virtuous to the instrumental nature of its lesser kinds, this third way claims that, despite appearances, the lesser kinds too involve wishing others well for their own sake. If so, every kind of philia would exhibit the most important characteristic of friendship, and we could find not only an anticipation but perhaps also the substance of our own ideas in the Greeks. This is the influential view of John Cooper, who finds good will in its full sense—wishing others well for their own sake—and so the idea of treating others as ends in themselves in every one of the three kinds of Aristotelian philia.  M. Heath, The Poetics of Greek Tragedy (Stanford, Calif., ), –; cited by David Konstan in his valuable overview, Friendship in the Classical World (Cambridge, ), .  A. Silver, ‘Friendship in Commercial Society: Eighteenth-Century Social Theory and Modern Sociology’, American Journal of Sociology,  (), – at –. The view is widely accepted by classicists (see e.g. Konstan, Friendship in the Classical World, ) and philosophers (see L. A. Blum, Friendship, Altruism, and Morality (London, )).  Cooper, ‘Aristotle on the Forms of Friendship’, –. Konstan, Friendship in the Classical World, –, agrees that benefit, pleasure, and virtue produce genuine friendship. However, he claims that these three are only one subclass of ϕιλία in general and the only relationships that count as friendships—the only relationships, he argues, whose participants Aristotle describes as ϕίλοι. By contrast, Konstan continues, although Aristotle considers the relationship of parents to children as an instance of ϕιλία, he does not think of it as a friendship since parents can love their children without needing or expecting reciprocation. That ϕίλος is reserved for a special class of ϕιλία seems most unlikely to me. Aristotle may not use the term when he discusses the family, but he also writes that ‘parents love their children as themselves since what comes from them is so to speak another self, other, that is, by being separate’ (a–). And that is one of the surest and most famous marks of the ϕίλος: ‘The ϕίλος is another self’ (a–; b–). In addition, all ϕιλίαι for Aristotle depend on one of the three grounds he distinguishes at the beginning of his discussion, including the relationship between mothers and children. For example,

Aristotelian Philia, Modern Friendship?



Without entering into the complexities of Aristotle’s text here, I think we can agree that even in a relationship based on benefit or pleasure I do not have to wish good things for you only if they actively promote my own interests. Cooper is right that Aristotle never ‘means that in a pleasure-friendship the one person wants the other to prosper in order that his own (the well-wisher’s) pleasure may be continued or increased’. I can certainly be more generous and enjoy your good fortune without caring whether it adds to mine—provided, though, that it does not undermine my own ends. I may be glad that the grocer down the street is doing well in his business even if I get nothing more out of it—but I will stop short of wishing him to do so well that his store turns into a supermarket whose employees no longer recognize me or whose products are no longer as good as they used to be. Such relationships are, broadly speaking, instrumental though not for that reason exploitative—not at all like Fagin’s shameless manipulation of Oliver Twist into a life of thieving, not even like the casual, impersonal (and still not exploitative) interactions we have every day with various waiters, shopkeepers, or ticket-takers. An instrumental relationship can encompass incidents and attitudes that are not themselves instrumental, just as the closest friends may use one another without for that reason undermining their friendship. That is not something Aristotle missed: he makes it clear that the virtuous are both useful and pleasant—not just in themselves but to each other as well (b–). Still, that is not to say that in such cases ‘one likes and wishes well to someone conceived of as pleasant or advantageous to oneself, and the good one wishes him to have, for his own sake, is therefore restricted to what he can acquire without, thereby or in consequence, ceasing to be pleasant or advantageous’. True, when I wish you good things provided they do not interfere with my interests, I do not do so for my own sake; but if I will stop when they do, I am not wishing them for your own sake either. That requires more of at a– he mentions that ‘sharing the sorrows and joys of another’ is often considered a mark of ϕιλία that is particularly appropriate to the attitude of a mother for her child. And a passage in the Eudemian Ethics (. , a–) states clearly that ‘sharing the sorrows and joys of another’ (Aristotle uses the very same words in both passages) is the characteristic mark (ὅρος) of the ϕιλία of pleasure.  

Cooper, ‘Aristotle on the Forms of Friendship’, . Ibid.  (emphasis added).



Alexander Nehamas

me: nothing less, in fact, than to be willing in some instances to sacrifice an interest of mine, whatever it might be, for yours—as we commonly and correctly say, for your sake. Without that, we are still within the domain of the instrumental. The fact is that each of these efforts to resolve the tension in Aristotle’s view of good will faces difficulties of its own. Contrary to the first, Aristotle does not think that the good will of perfect philia is sharply or categorically different from the good will exhibited in its imperfect forms. Contrary to the second two, he seems to believe that he is giving a single account of both forms without, however, thinking that they are all instrumental or all other-regarding. What we need is an interpretation that attributes to Aristotle a unified view of good will that does not, as a result, assimilate the three kinds of philia to one another. We can give such an interpretation. And to do so, we must begin by noticing that Aristotle uses several different expressions when he specifies what well-wishing involves. Philoi wish each other well, he writes when he first introduces the notion, for their philoi’s sake. A few lines later, in his definition of philia, he says that philoi wish each other well on account of virtue, pleasure, or benefit (which one it is depends on what binds them to each other), and just below he tells us that philoi wish each other well in so far as they are virtuous, pleasant, or beneficial; neither of these, moreover, is to wish them well ‘in or for themselves’ (kath’ hautous). All these expressions appear together in his account of the well-wishing of the virtuous: 

This will have to be qualified; see below, p. . For the other’s sake: ἐκείνου ἕνεκα (e.g. b, b); in so far as the other is virtuous, pleasant, or beneficial: ᾗ (e.g. a, , ); on account of being virtuous, pleasant, or beneficial: διά (e.g. a; a, ). See also b–, where wanting what is good for the ϕίλος is given as a necessary feature of ϕιλία (only to be apparently denied later on). Cooper’s interpretation rests on an ingenious distinction between ἕνεκα and διά. He takes the διά-construction to be retrospective, meaning something like ‘in recognition of’ and expressing the grounds on which each particular ϕιλία has been based so far. By contrast, he reads the ἕνεκα construction (Aristotle’s most common way of designating the final cause, the purpose for which things occur as they do) prospectively, as expressing the direct object of the well-wishing: ‘The pleasure-friend will now be said to wish well to his friend for his friend’s own sake, in consequence of recognizing him as someone who is and has been an enjoyable companion, and the advantage-friend wishes his friend well for his friend’s own sake, in consequence of recognizing him as someone who regularly benefits him and has done so in the past’ (‘Aristotle on the Forms of Friendship’, ). The interpretation I propose does not require that distinction, which, moreover, does not meet the objection I make in the main text. 

Aristotelian Philia, Modern Friendship?



The virtuous wish each other well in the same way and in so far as they are virtuous, and they are virtuous in themselves, and those who wish each other well for their own sake are philoi above all, for they are such on account of themselves and not incidentally. (b–)

And although he uses the same expressions in connection with the lesser philiai, he also points out a major difference between them and the philia of virtue: Those who care for each other on account of benefit do not care for [their philoi] in themselves but only in so far as they obtain some benefit from them; so too in the case of pleasure: these people do not care for those they find entertaining in themselves, but because they are pleasant to them. Both those who care on account of benefit and those who care on account of pleasure care on account of what is beneficial or pleasant to them—not in so far as the people they care for are who they are but in so far as they are beneficial or pleasant. Such philiai, then, are incidental: people are not cared for in so far as they are who they are but only in so far as they provide either benefit or pleasure. (a–)

Although Aristotle does not use the expression ‘for the friend’s sake’ in this passage, he does not seem to want to withhold it from the lesser philiai. In his general statements regarding good will (e.g. b–) he applies all three expressions to each kind without qualification. In that respect, therefore, the three kinds of philia are strictly parallel: whether they care for virtue, pleasure, or benefit, philoi always wish each other well for their philos’s sake, in so far as, or on account of, virtue, pleasure, or benefit respectively. The difference is that if we are drawn to each other’s virtue, I wish you well on account of your character and therefore on account of who you really are; but if we are bound by the pleasure or benefit we can provide for each other, my good will is based on and directed towards accidental features of yours: your sense of humour, your looks, your financial resources, and your social position are all, as far as Aristotle is concerned, incidental to your rational essence. At various points, though, Aristotle seems to draw a much sharper distinction between higher and lower philiai. He sounds, that is, as if he believes not simply that the good will of pleasure and benefit is focused on the other’s accidental features but that,  ‘In so far as’: ᾗ; ‘in themselves’ or, depending on the context, ‘for themselves’: καθ᾿ αὐτούς; ‘for their own sake’: ἐκείνων ἕνεκα; ‘on account of themselves’: δι᾿ αὐτούς.



Alexander Nehamas

in fact, it is not focused on the other at all: ‘Those who are philoi on account of benefit part as soon as their interests separate, for they were not philoi of each other but of gain’ (a–). Later on, he says that erotic philiai end ‘as soon as that for the sake of which they were involved is no longer there, for they did not care for each other but for things they happened to have—things that are not permanent’ (a–). Does that, then, not take us back to where we started, making the philia of benefit or pleasure purely instrumental? Is it possible to insist that every philia, not just that of the virtuous, requires wishing well for the sake of one’s philos? Yes, it is—if we recall that Aristotle thinks that one can refer, say, to Alcibiades, who happens to be handsome, as ‘the handsome Alcibiades’, ‘the handsome’, or ‘Alcibiades in so far as he is handsome’. These are all, as he puts it, ‘in a way’ the same (Phys. . , b–a). And so, if I wish my barber, Tomas, well in so far as he is beneficial to me, I also wish him well for the sake of the beneficial Tomas, or simply on account of the beneficial, which is an accidental feature of his. But if I wish him well in so far as he is virtuous, I wish him well for the sake of the virtuous Tomas, or simply for the sake of the virtuous, which belongs to his essence and establishes who he really is (I am concerned with Tomas, as we would put it today, ‘as a person’ and not for this or that feature of his). In the first case, where the philia and the well-wishing disappear when Tomas is no longer of benefit to me, Aristotle can say that I cared not for Tomas himself but for the beneficial instead. Not so in the second: to care for Tomas in so far as he is virtuous is not to care for the virtuous instead of caring for Tomas, since his virtue makes him who he is: Tomas, the virtuous Tomas, Tomas in so far as he is virtuous, and the virtuous are now equivalent, and I care for Tomas himself. And, therefore, although everyone wishes their philos well for the philos’s sake, and not only for their own, only those who wish others well for their virtue wish them well for their own sake in a strict Aristotelian sense. Aristotle can make that claim because virtue belongs to the soul and is the excellent exercise of the natural abilities, the fulfilment of the function (ergon), of human beings. And, as Aris Aristotle seems to claim something similar at b–, where he distinguishes virtue-ϕίλοι from others because ‘they wish good things for those they care for [not for themselves but] for their sake, because of a state they are in and not a feeling they have’.  I offer an expanded version of this argument in an appendix to this essay.

Aristotelian Philia, Modern Friendship?



totle argues, among other places in Metaphysics Ζ , a thing and its nature or essence (ousia), what it is to be (to ti ēn einai) such a thing, are identical. Is the difference between wishing you well for your own sake and wishing you well simply for your sake more than a linguistic trick? Does it actually make a difference? I believe it does. If our relationship is based on benefit (or pleasure), all sorts of things may be good for you but not for me. Some of these may be contrary to my own interests: if you make a new philos, for example, you may be less willing to help me when I need you; if you get elected to public office, you may stop favouring me above the rest of the citizens. And since I am concerned with what benefits you only to the extent that it either increases or at least does not conflict with my own interests, I will neither wish such things for you nor will I help you obtain them. Since my own interests determine whether I will do so or not, it is clear that I do not wish you well for your own sake: that is the difference between benefit and pleasure, on the one hand, and virtue on the other. Other benefits, though, may be irrelevant to me: as long as you continue to fulfil your obligations towards me, how rich you become will make no difference to our relationship and I may very well wish such a thing for you. That shows that I wish you well not only for my own sake but also for yours: that is the ‘similarity’ to the philia of virtue that allows the lesser relationships to be considered species of philia as well. For if our relationship is based on virtue, nothing that is good for one of us can be bad for the other. When I wish you well in so far as you are virtuous, everything I wish will augment, or at least not interfere with, your virtue. But nothing that happens as a result of your becoming a better person can ever harm me or be incompatible with my interests or pleasures. And since anything that serves virtue, if done for the right reasons, 

Metaph. Ζ , a–a; cf. Ζ , b–a. According to A. W. H. Adkins, in ‘ “Friendship” and “Self-Sufficiency” in Homer and Aristotle’, Classical Quarterly,   (), –, it does not. He argues that all ϕιλία is ultimately selfish: both pleasure- and benefit-ϕίλοι ‘feel affection of a selfish kind on account of their own ἀγαθόν; but this is true of ϕιλία κατ᾽ ἀρετήν as well’ (). He claims that Aristotle is forced into ‘contortions’ that ‘show the gravity of the problems which faced the Greek moral philosopher in his efforts to create a civic morality out of the primitive and intractable materials which lay to hand’ ().  Aristotle is careful to emphasize that the ϕιλία of the virtuous, since it is perfect and complete, always brings with it both benefit and pleasure, e.g. b–. He does not offer a separate argument for that idea here, which seems to me plainly true, but he does, in connection with pleasure, at NE . , a–. 



Alexander Nehamas

is itself virtuous (NE ., a–), anything I do for you will be, in addition to any beneficial or pleasant consequences it may have for us, itself good—a virtuous act in its own right.

III Despite the formal similarities between them, several practical differences separate the philia of the virtuous from its two lesser varieties. One is that the former is permanent or at least longlasting: character, according to Aristotle, is unlikely to undergo radical changes, while interest, power, beauty, and everything else that may provide benefit or pleasure are quick to come and go (b–). Another is that although in pleasure- and benefitphilia I will not wish you well if doing so interferes with my interests, everything I wish for you on account of your virtue will tend to make you an even better person and will therefore be good not only for you but for me as well. No wonder this is a rare species of philia: even among the small group of upper-class Athenian men who were Aristotle’s main audience, such relationships must have been hard to come by. Aelred believed that ‘friendship is eternal if it is true friendship; but, if it should ever cease to be, then it was not true friendship, even though it seemed to be so’. That is an exaggeration. Aristotle recognized that any philia—in certain circumstances, even that of the virtuous—may end, but did not think that such an end would show it to have been a sham (b–). At the same time, though, when friends or lovers abandon us at an hour of need, the genuineness of their feelings always comes into question, no matter how they had treated us up to that point. Exactly that question is raised in Paul Anderson’s  film There will be Blood. The oilman Daniel Plainview assumes responsibility for the infant son of one of his workers, who died in an accident. He brings up the child on his own and the boy seems to be the only being to whom this rough and wild man is truly attached. One day, though, when a rig explodes and the blast leaves the boy deaf, Plainview’s attitude towards him undergoes a serious change: he becomes indifferent to, almost ashamed of, the boy, and eventually tricks him cruelly and has him deposited at a special boarding school. At that point, it is 

Aelred of Rivaulx, Spiritual Friendship, . ,  Laker.

Aristotelian Philia, Modern Friendship?



hard to know whether his feelings for the boy had been genuine up until the accident or whether, as he screams, murderously drunk many years later, he only kept the boy with him in order to appear respectable—a family man—to the people whom he was cheating out of their land so that he could dig his wells. No cases of this sort are easy to interpret—we argue about them constantly—and every one introduces the suspicion that what had appeared as a real friendship was all along a mask for someone’s selfishness. It seems to us essential that if I wish you to do well for your own sake, as friendship requires, I will be willing to sacrifice my own interests for your sake if that becomes necessary—and the extent of my sacrifice is often a measure of the depth of our friendship. But no such sacrifice is possible in Aristotle’s scheme. If our philia is based on pleasure or benefit, I will end it as soon as I realize that your good is incompatible with mine. And if it is based on virtue, there is never anything to sacrifice: as we have seen, every good thing I might wish for you promotes your virtue and nothing that happens as a result of your becoming a better person can be either painful or harmful to me. The instrumental nature of pleasure- and benefit-philia on the one hand and the fact that self-sacrifice is never an issue either for them or for that of the virtuous establish a large gap between Aristotelian philia and our own understanding of friendship. And that is not all. Aristotle is particularly interested in the obligations philia generates, and he devotes large parts of books  and  of the Nicomachean Ethics to a long and detailed specification of what philoi owe one another. The result is that, to modern ears, much of his discussion sounds out of place in a treatise on moral philosophy. He distinguishes, for example, ‘ethical’ benefit-philia, which is based on the parties’ knowledge of each other’s character, from a ‘legal’ sort, which involves explicit agreements between them. Legal philia, in turn, can be either ‘liberal’ and allow its participants to exchange goods as they become able to do so or ‘completely mercenary’, requiring a ‘hand-to-hand’ exchange. And since this last relationship is without a doubt a case of philia, it makes the gap between Aristotle’s notion and modern friendship even bigger (b–). Aristotle also points out how difficult it can be to establish the value of the goods one philos provides another so that one can know how much is owed in return. How should we establish their value,



Alexander Nehamas

‘by the benefit that accrues to the recipient . . . or instead by the good turn done by the benefactor’ (a–)? He answers that in benefit-philia the standard should be the value of the benefit to the recipient while in virtue-philia (which, we should recognize, is not exempt from such legislation) the standard is determined by the cost to the benefactor (a–). When, again, two philoi are unequal in some respect that prevents one from making an exact repayment to the other, ‘the superior person should get more honour, and the needy person more profit, since honour is the reward of virtue and beneficence, while profit [assists in times of] need’ (b–). When the aims of philoi conflict, disputes are difficult to resolve: if a man expects that the boy he pursues will give him pleasure while the boy expects the man to be useful to him and help him assume the duties of an adult citizen, as was the Athenian custom, the likely outcome of their disagreement is the end of their relationship (a–). And although in general one should first repay a debt before doing favours for one’s philoi, the rule is not without exceptions. Imagine, for example, that when you were abducted by pirates a philos ransomed you: are you now obliged to ransom your philos, who was recently abducted himself, in return? Or suppose that your philos has not been abducted but needs his money for some other reason: should you discharge your debt or ransom your father, who is now in the pirates’ hands? ‘It would seem’, Aristotle writes in a judicious mood, ‘that you should ransom your father, even in preference to yourself.’ Yet another exception to the rule occurs when making a gift to a third party appears ‘nobler or more necessary’ than repaying someone who had helped you earlier (b–a). If this small selection of the many cases Aristotle discusses reminds you of an introductory text on the law of contracts, that is exactly as it should be: such a text is just where such passages belong and such a text is just what this part of the Nicomachean Ethics is. Nothing could be further from Aristotle’s mind than Adam Smith’s fundamental distinction between the self-interested, rulegoverned exchanges that control the workings of the market and the personal friendships that arise from ‘natural sympathy’. And since the fundamental bond that holds each city together and unites  On the mutual obligations assumed by those connected through ϕιλία see M. W. Blundell, Helping Friends and Harming Enemies: A Study in Sophocles and Greek Ethics (Cambridge, ), –.

Aristotelian Philia, Modern Friendship?



its citizens is a species of benefit-philia (often called ‘civic friendship’), Aristotle would not have been able even to imagine how someone might believe, as C. S. Lewis did, that ‘Authority frowns on Friendship . . . Every real Friendship is a sort of secession, even a rebellion.’ Conversely, it is equally hard for us to imagine that contemporary accounts of friendship might provide principles for resolving commercial disputes or advise their readers to ‘accord to every older person the honour befitting his age, by standing up, giving up our seats, and similar actions’ (a–).

IV Perhaps, then, we should leave benefit-philia aside and turn instead to relationships that are based on pleasure. Surely, they seem closer to modern ideas of friendship, since friends so clearly enjoy each other’s company. And so they might, had Aristotle limited himself to the narrow range of examples that regularly appear in the many interpretations of his views: drinking companions, squash partners, amusing acquaintances, or temperate pleasures such as enjoying the sight of ‘a vigorous colt prancing about a field on a fine spring morning, or the look of delight on a child’s face as he rides on a swing’. But Aristotle’s own examples are much more often cases of passion, erotic and sexual—primarily pederastic or matrimonial—than the witty conversationalists so common among his interpreters. Aristotle’s straightforward inclusion of erōs under philia contrasts sharply with the common contemporary suspicion that sex makes  The Four Loves, . On civic ϕιλία see Cooper, ‘Aristotle on the Forms of Friendship’, sect. , and ‘Political Animals and Civic Friendship’, in id., Reason, –.  This last example comes from Pakaluk, Aristotle’s Nicomachean Ethics, , who also mentions the amusing companion (); enjoying each other’s company is cited by D. Bostock, Aristotle’s Ethics (Oxford, ), , drinking companions are Cooper’s example in ‘Aristotle on the Forms of Friendship’,  (though see  n. ), and squash partners appear in S. Broadie (comm.) and C. Rowe (trans.), Aristotle: Nicomachean Ethics (Oxford, ), .  Pederastic: b–, a–, a–, b–, a ff.; matrimonial: a ff.; pleasant companions (εὐτράπελοι): a, a, a. Attention is paid to ἔρως by L. S. Pangle, Aristotle and the Philosophy of Friendship (Cambridge, ), –, who also speculates that ‘we may well wonder whether erōs and friendship, in their best and fullest manifestations, might not converge, so that the strongest friendship would be confined to two and would even itself be erotic’ (). The position has merit, but it is unlikely to have been Aristotle’s own.



Alexander Nehamas

friendship ambiguous and dangerous and is the cause of complex and harmful self-deceptions. The relationship between philia and friendship, then, is very complicated. To understand some of its complexity, and see some of the differences between them, we need to begin by moving far afield—as far from both Aristotle and ourselves as early modern England and Edmund Spenser’s Faerie Queene. Spenser’s ‘friendship’ is an expression of Aristotle’s understanding of philia—an understanding that, even as Spenser was writing, was being replaced by a new idea, which gradually became characteristic of modernity. Each of the poem’s six books is devoted to an allegory of a different virtue, and the fourth book addresses friendship. There is an (imprecise) echo of Aristotle in Spenser’s distinction between three kinds of love, to weet The deare affection vnto kindred sweet, Or raging fire of loue to woman kind, Or zeale of friends combynd with vertues meet.

What is important, however, is to note the place of friendship within Spenser’s catalogue of virtues—it is a place a modern reader may find surprising. As one of Spenser’s readers has written, In the first three books of the Faerie Queene Spenser is primarily interested in the virtue of the individual as such: his holiness, his temperance, his chastity—the harmony of the whole nature controlled by reason. Beginning with the Fourth Book he turns to a more definite study of the individual in relation to other individuals. With his classical background it was reasonable that he should put friendship next. Aristotle had upheld the thesis that all social relationships grow out of friendship.

This is surprising because, far from thinking of friendship as the springboard of all social relationships, we are much more likely to 

See M. Vernon, The Philosophy of Friendship (New York, ), ch. . This new understanding emerges in Spenser’s own time. Its most eloquent expression is in Montaigne’s ‘Of Friendship’.  Edmund Spenser, The Faerie Queene, ed. T. Roche, Jr. (London, ), . ix. .  C. G. Smith, ‘Spenser’s Theory of Friendship’, PMLA,  (), – at . I first became aware of this way of reading Spenser when I attended one of the lectures of my Princeton colleague Jeffrey Dolven. Paul Alpers was kind enough to share some of his knowledge of Spenser while I was revising the manuscript of this essay. 

Aristotelian Philia, Modern Friendship?



draw a sharp contrast between our friendships and our public interactions. In the essay that first focused the attention of contemporary philosophers on friendship, for example, Elizabeth Telfer specified the affection friends feel for one another as a desire for another’s welfare and happiness as a particular individual. This desire is thus to be distinguished both from the sense of duty and from benevolence. For these motives prompt us to seek others’ good in general, whereas we want to say that those who feel affection feel a concern for another which they do not feel for everyone.

The impartial point of view of duty or benevolence requires the effort to treat each individual as we treat every other, while the affection of friends, which is partial and preferential, is ‘not for each individual, but for this individual rather than others’. This contrast, as we have seen, is central to modern moral philosophy, whose main concern with friendship has been to show that it is consistent with morality’s universal demands. For us moderns, it is fundamental that friendship belongs to the realm of the private and the personal, not to the public domain where moral and political values regulate how we should interact with the world at large. But friendship was not part of that domain for most of its history: it was only in the late sixteenth century—ironically, just when Spenser was composing The Faerie Queene, writing ‘for an aristocratic and upper-middle-class audience in a self-consciously archaizing manner, thereby participating in the decorative revival of feudal trappings that characterized Elizabethan courtly ritual’— that friendship began to withdraw from the public realm where it had lodged till then. To which realm, then, does philia belong—the public or the private? And what—we must ask first—is its place in Aristotle’s ethical scheme? Why does Aristotle spend more time on this one subject than he does on any other? The reason, I believe, is this. The overall purpose of the Nicomachean Ethics is to determine  Telfer, ‘Friendship’, Proceedings of the Aristotelian Society,  (–), – at , repr. in Pakaluk (ed.), Selves, – at .  The proposal by feminist philosophers to use friendship rather than instrumental relations as the basis of a new understanding, and organization, of society, important as it is in itself, is a subject for another occasion. See M. Friedman, What are Friends For? Feminist Perspectives on Personal Relationships and Moral Theory (Ithaca, NY, ).  S. Greenblatt, Renaissance Self-Fashioning: From More to Shakespeare (Chicago, ), .



Alexander Nehamas

the kind of life that is the best and happiest for human beings. Most of the book is concerned with the nature and development of the soul—the internal, psychological structure—that one must possess in order to lead such a life. Aristotle opens the Ethics with a general account of happiness; he then turns to the nature of virtue and the relationship of virtue to emotion, decision, and action; the cultivation and character of the individual virtues and their corresponding vices; obstacles to their development; the role of pleasure in life; and, after philia, the nature of happiness once again. Throughout these investigations, his attention is focused on the internal condition of the individual. Apart from his examination of justice, which, he tells us repeatedly, is the virtue most closely connected with it, it is only in his account of philia that Aristotle turns from the psychology of individuals to a serious examination of their attitudes, feelings, and behaviour towards the other people with whom, in various degrees, they all share their lives. It is only here, in these two books of the Ethics, that Aristotle assumes an interpersonal point of view and addresses what ‘we owe to each other’ and so, to a certain extent, the duties and obligations that are the primary concern of modern moral philosophy. The relationships that Aristotle examines originate within the narrow circle of the family (a–) and gradually come to encompass everyone who is relevant to his ethical investigations. They stop far short, though, of extending to everyone who is relevant to modern moral philosophy. The duties and obligations of morality apply equally to every human being, but Aristotle did not think that rationality—which of course he considers essential to every human being—is enough, as it later was for Kant, to ground the principles of ethics. The range of ethical subjects that matters to him is by and large exhausted once he reaches the limits of the polis—the city-state that was the basic political unit of classical antiquity. He notes, in passing, that members of the same species, especially human beings, are bound by philia and that we admire those who love human beings in general (‘philanthropists’), but he has nothing more to say about them. He also points out that ‘in our travels we can see how every human being is akin and philon to every other’ (a–), but that is an anthropological observation, a description of how the 

e.g. a–, a–, b–. The phrase comes from T. M. Scanlon’s What we Owe to Each Other (Cambridge, Mass., ). 

Aristotelian Philia, Modern Friendship?



members of different societies treat one another and not, at a time when people were as likely to harm as to befriend strange visitors, a reference to the reception travellers were likely to find abroad. And while he believes that there is no philia or justice between human beings and animals or slaves ‘in so far as they are slaves’ (how, he asks with frightening equanimity, can we care for a tool, even if, as in this case, it is ‘a tool with a soul’?), he also allows that ‘to the extent that slaves are human beings . . . some justice . . . and philia’ towards them are appropriate. It is unlikely, however, that Athenian slaves could really take comfort in that thought. ‘Affection’ (philēsis), which is necessary to philia, ranges from the lovers’ most intense passion through the measured feelings of the virtuous to the anaemic attitude of people towards those they happen to run across in their travels or the contemptuous regard of a master for a slave. Philia may extend to all human beings, in so far as they are human beings, but if it does, it reaches most of them in a very attenuated form. So, most surprisingly, does justice, which applies differently to people to whom one is related in different ways. To ask how philoi should behave towards one another is the same as asking how they can behave with justice: ‘How it is just for a philos to treat a philos is different from how it is just to treat a stranger or a companion or a schoolmate’ (a–). For us, it is essential that justice apply to everyone in the same way, and we distinguish sharply between its universal demands and the discriminating commitments of friendship; Aristotle, by contrast, sees no difference between them. That is, again, because for him the circle of ethical relevance reaches no further than the borders of the polis. There is no hint in Aristotle’s thought of the Enlightenment conception of a common human nature that demands universal and unconditional respect for every fellow being  See Blundell, Helping Friends and Harming Enemies, –. The dangers of travel were to some extent mitigated by the institution of ξενία or ξενικὴ ϕιλία, as Aristotle refers to it (a–), the host–guest relationship that joined individuals from different social units. Aristotle has nothing to say about it beyond classifying it among the species of benefit-ϕιλία. But fascinating information about the institution and the complex rules that governed it is contained in Gabriel Herman’s Ritualized Friendship and the Greek City (Cambridge, ), kindly brought to my attention by G. E. R. Lloyd.  Could this be an allusion to Plato, who writes that in the ideal city the artisans, whose reason is weak, submit to the reason of their rulers, becoming slaves through their submission but ϕίλοι through the reason they now have in common (Rep.  )?



Alexander Nehamas

any more than there is a trace of the Christian ideal of a human brotherhood that requires universal and unconditional love for the neighbour (agapē). Aristotle knew that the Athenians had certain well-delineated obligations towards their allies (to the extent that their allies were ready to reciprocate). But he shows little if any ethical concern for those who live beyond the borders of Attica and none for its enemies—Sparta or Thebes, at various times—or for distant barbarians—Persians or Hypeboreans. No ethical demands can come from such places. By contrast, the essential intention behind agapē, universal respect, and the general welfare is to bind every human being to every other. Although who counts as a neighbour or a human being has always been, and still is, manipulated in all sorts of ways and exploited for the worst purposes, the notions (unlike, say, ‘Athenian’ or ‘Spartan’) are vague and flexible enough to allow us to extend them to new individuals or groups: it was, after all, the language of the rights of man that eventually justified extending property rights and the vote to women. But philia is not flexible in that way: it dissolves where the city ends, and it joins only those whose interests affect one another directly. In no case, moreover, could the interests of another party take precedence over one’s own welfare. One of Aristotle’s most surprising claims about philia, as we have seen, is that even in the very best of cases one’s own interests come before those of one’s philos. Tempted to say that one wishes the greatest goods for one’s philos, he finally holds back: ‘But perhaps not exactly all: for everyone wishes the greatest goods for oneself above all’ (a–). And although he recognizes that the virtuous are willing to sacrifice money, honours, and even their life for philoi and country, he does not think of their choice as a genuine sacrifice: in doing so, he points out, they obtain something much more valuable instead—nobility (to kalon): ‘they therefore award the greatest good to themselves’ (a–b).  That does not prevent commercial interactions between different cities or their citizens, although trade occurred mostly between allies.  My sense is that the attitude Aristotle describes here is not nearly as complacent as the account Gregory Vlastos offered in ‘The Individual as Object of Love in Plato’. According to Vlastos, ‘The only love of persons as persons that really interests [Aristotle] is that between the members of a social élite, each of whom can afford disinterested affection for his peers, assured in advance that he will normally have theirs in return’ (). But the love Aristotle discusses is both more discriminating (one can only have a few ϕίλοι, he argues in . ) and depends on such an

Aristotelian Philia, Modern Friendship?



V Throughout this essay, I have left philia, philos, and philēsis untranslated, although I realize that doing so must have made for cumbersome reading. I did so in order not to prejudge the question of their connection to friendship. I hope I have given some reason to think that Aristotelian philia is a very broad genus of relations that take radically different forms and includes, among many others, family connections, erotic love, mutual arrangements and commercial exchanges (explicit or tacit), social contacts and political alliances, and personal commitments of the most various sorts. To the extent that these are governed by profit or pleasure, and although they are not egoistic—such philoi are not motivated purely by the wish to promote their own interests—they place clear and explicit limits to the extent to which one can go in wishing another’s good: well-wishing stops as soon as one’s own interests are harmed. That they are kinds of philia is simply not enough to make them kinds of friendship. They are not the sorts of relationships that are relevant to an ethical investigation of friendship. But the philia of the virtuous is. It has a deep connection to our relationship with our closest friends. For both the virtuous and close friends love one another ‘for themselves’, both are attracted to one another’s character, and both expect their relationships to be, if not permanent, at least long-lasting. But there are other similarities as well. Although one does not enter into it for what one can get out of it, virtue-philia brings with it both benefits and pleasure (b– ); so does friendship: one can rely on one’s friends, and their company is always something one enjoys—although Aristotle does not seem to focus his discriminating eye on the sorrows and troubles that friends can cause one another. Like virtue-philia, friendship assurance (which does not obtain ‘in advance’—in advance of what?) only because ϕιλία is a mutual relationship.  A ϕιλία can begin for some particular reason and develop into one of a different kind. For example, the ϕιλία between men and women is ‘natural’, owing to our fundamental need for procreation, which we share with animals. But, unlike animals, families also work together in order to secure life’s necessities and their members have different tasks (Aristotle elsewhere makes it clear that women’s capacities— and virtues—are inferior to men’s, e.g. b), and their ϕιλία can come to involve both benefit and pleasure. In some cases the relation between husband and wife can even come to be based on virtue—‘if they are good’ (a–).  Cooper’s gentler ‘character friendship’, which, unlike Aristotle’s perfect ϕιλία,



Alexander Nehamas

is immune, or at least resistant, to slander: we know our friends well and it takes much to undermine our faith in their goodness; virtue-philoi (we might as well call them friends from now on) trust one another (a–). Friends influence the shape of one another’s life: they spend much of their time together, do things they would not have done had they been alone, and affect each other’s character deeply (b–). Theirs is a complex and demanding relationship, which one can have only with a few people: ‘It is impossible to have the perfect sort of philia with many people’ (a–). Friends do not dispute about who has done what for whom (a–b), and their relationship makes claims on them in its own right: one should still care for one’s friends even if they have undergone serious changes—provided, though, that they have not become ‘incurably’ vicious (b–). Friends, finally, ‘are disposed towards each other as they are disposed towards themselves: a friend is another self’ (a–). ‘is not the exclusive preserve of moral heroes’, can much more easily accommodate such imperfections; see ‘Aristotle on the Forms of Friendship’,  ff.  An interesting story is connected with this Aristotelian idea. Both Montaigne and Nietzsche refer to an otherwise unknown (and, for him, quite unusual) Aristotelian text: ‘Oh, my friends; there is no friend!’ (ὦ ϕίλοι, οὐδεὶς ϕίλος)—a statement of which Jacques Derrida makes extensive use in The Politics of Friendship (New York, ). Montaigne and Nietzsche derive the text from Diogenes Laertius’ Lives of the Philosophers, . . , which, in its present editions, actually quotes Aristotle as saying, ᾧ ϕίλοι οὐδεὶς ϕίλος (‘He who has [many] friends has no friend’). In his article ‘Friendship’ (Contretemps,  (), –), Giorgio Agamben traces this, the most reasonable reading of Diogenes’ text, to Isaac Casaubon’s  edition of Diogenes. The textual difference between the two versions is slight—Casaubon simply changed the breathing and added an iota subscript to the initial omega of the phrase, and so attributed to Aristotle a reasonable view that conforms to the spirit of the Nicomachean and, more or less, to the letter of the Eudemian Ethics: οὐθεὶς ϕίλος ᾧ πολλοὶ ϕίλοι (b). Montaigne’s text, of course, pre-dated Casaubon’s edition but Nietzsche, who had written on Diogenes, should have known better (see Beiträge zur Quellenkunde und Kritik des Laertius Diogenes (Basel, )).  At another point Aristotle’s dry prose, ‘The defining features that are found in ϕιλίαι towards one’s neighbours [οἱ πέλας] would seem to be derived from the features of ϕιλία towards oneself’ (a–), is an uncanny anticipation—of no more, I think, than the language—of the Christian command to ‘love your neighbour [ὁ πλησίον] as yourself’ (Matt. : –). Only the virtuous can have such an attitude: Aristotle holds the bizarre view that the vicious, whose souls are in constant conflict and who are always regretting their deeds, hate their life and bear no ϕιλία towards themselves since (they know that) nothing about them can be the object of ϕίλησις; the vicious hate themselves. And so, most likely, do the many, who are all base: although Aristotle considers the possibility that they may be satisfied with themselves, thinking that they are decent people, he finally decides that they too lack anything that may attract ϕίλησις and cannot be ϕίλοι to themselves (b–).

Aristotelian Philia, Modern Friendship?



This, Aristotle’s most famous expression regarding philia and a staple of the philosophical tradition, occurs twice more in his argument that friendship should be part of the life of even the happiest human being, who by definition possesses every good and is to the highest extent ‘self-sufficient’ (b; b; see also a–). Aristotle believes that ‘no one would choose to have all other goods and yet be alone, since human beings are political and born to live together with others’. And since we have no choice about living with others, it is ‘better to spend one’s days among friends rather than strangers and in the company of decent people rather than anyone who happens to come along’ (b–). Aristotle concedes that, being self-sufficient, a happy person will not need friends for benefit or pleasure; but a happy life must include virtuous friends, first, because observing their actions will add to the pleasures that life already contains and, second, because friends provide more opportunities for exercising one’s virtue and so, once again, more pleasure as well. His third argument, by far the most complex passage in these two books of the Ethics, aims to show that ‘a virtuous person would naturally choose to have a virtuous friend’ (a–). It seems to begin with the idea that the virtuous derive great pleasure from their awareness of being virtuous: they enjoy being who they are. And since they ‘are disposed towards themselves as they are disposed towards their friends (for a friend is another self)’ (b–), they are, in addition, almost as pleased by their awareness of their friends’ virtue as of their own and enjoy their friends’ being who they are: it is almost as if the actions of our friends are our own.  The argument seems to start with the thesis that the life of virtue, which like all human life consists essentially in the activities of perception and thought, is both good and pleasant to those who lead it. It goes on to claim that when we perceive or think we are aware of our perceiving or thinking, and in being aware of perceiving or thinking we are aware of being alive. Being aware we are alive is pleasant in itself, for life is by nature good, and it is pleasant to be aware of something good within us. Now life is especially worthy of choice for the virtuous, for being is good and pleasant to them as well as in itself. But ‘the virtuous are disposed towards themselves as they are disposed towards their ϕίλος (for the ϕίλος is another self)’ (b–). And since the virtuous person finds being (a person like) himself choiceworthy, he will find being (a person like) his ϕίλος choiceworthy in the same or in a similar way. As we saw, such a life is worthy of choice because one perceives oneself to be good, and such a perception is in itself pleasant. The virtuous person will then perceive his ϕίλος’s being along with his own (and so derive greater pleasure than otherwise), something that is achieved through spending their life together and engaging with each other’s words and thoughts. In short, a virtuous person, who is happy to be who he is, is in addition happy that his ϕίλος, as he comes to know him through their many interactions, is himself who he is. [Note  continues overleaf



Alexander Nehamas

But precisely because virtue-philia is so close to friendship, we should pause before we follow Aristotle, as he has been followed by so many, in emphasizing the connection between friendship and virtue. That is partly because Aristotelian aretē is much broader than virtue as we understand it, but that is not the main reason. More important, Aristotle holds the bizarre view that the vicious (hoi mochthēroi, perhaps in a moral sense here) live in constant internal conflict, always regretting their deeds, hating their life and themselves. He adds that in all likelihood so do the many, hoi polloi, who are (merely) base or vulgar. He actually wonders whether perhaps the many might not think of themselves as decent people and so manage to be satisfied with themselves, but he resists the temptation to think so and concludes that since, as with the vicious, nothing about the many is lovable, they can neither love themselves nor have any friends (b–). Perhaps La Rochefoucauld was right after all: true friendship may indeed be rarer than true love. But he was not. And neither was Aristotle when, thinking that virtue is absolutely essential to friendship, he allowed only a small group of Athenian men to love one another (and sometimes, in a qualified sense, their wives, a–) as friends love one another. Friendship is both much more common and more fraught with risk than he imagined. George Eliot was closer to reality when she asked: Among our valued friends is there not some one or other who is a little too self-confident and disdainful; whose distinguished mind is a little spotted with commonness; who is a little pinched here and protuberant there with native prejudices; or whose better energies are liable to lapse down the wrong channel under the influence of transient solicitations?

Neither the virtuous nor the vicious nor, for that matter, the many are drawn from life in Aristotle’s scheme: they are unrealistic and one-dimensional constructs. We are all of us mixtures of virtues The issues raised by Aristotle’s various statements to the effect that ϕιλία is originally derived from our feelings towards ourselves and then applied to others (e.g. a–, b), and the importance of self-love for Aristotle generally, are extremely complex and I have to leave them aside. See, among others, J. Annas, ‘Plato and Aristotle on Friendship and Altruism’, Mind,  (), –; J. M. Cooper, ‘Friendship and the Good in Aristotle’, in id., Reason, –; C. H. Kahn, ‘Aristotle and Altruism’, Mind,  (), –; and, especially, J. Whiting, ‘The Nicomachean Account of ϕιλία’, in R. Kraut (ed.), The Blackwell Guide to Aristotle’s Nicomachean Ethics (Malden, Mass., ), –. 

George Eliot, Middlemarch, ch. ,  vols. (Boston, ), i. –.

Aristotelian Philia, Modern Friendship?



and vices and all of us are aware that even our best friends have their shortcomings. True, it is unlikely that we can be friends with people we consider evil. But we are very often willing to excuse our friends’ failings, moral and otherwise. And yet in Aristotle’s bond between friendship and virtue there is a deep insight that it is important not to lose. In Euripides’ Iphigenia in Tauris Orestes and his friend Pylades are sent by Apollo to bring the sacred statue of Artemis from Tauris (the land of the savage Taurians), in the Crimea, to Athens. It turns out that Iphigenia, Orestes’ sister, who everyone thought had been sacrificed by her father Agamemnon to Artemis so that the Greek ships could sail for Troy, was actually snatched from the altar by the goddess and is now her priestess in Tauris. The two friends are captured by the locals and brought to Iphigeneia, who does not recognize her brother any more than he knows who she is. According to ancient custom, one of them is to be sacrificed to Artemis, but when Orestes is given the choice, he refuses to allow Pylades, his ‘best and dearest friend’ (philtatos philōn, ), to die in his place— partly because, he explains, ‘I am the captain of this misadventure, | and he the loyal shipmate who stayed by my side’ (–), revealing that their friendship, whatever else it involves, is also based on Pylades’ courage and loyalty—his virtues. In the first instance, this represents Orestes’ own view of Pylades, whose friend he would not be if he found nothing admirable in his character. Aristotle makes just this point when he writes that we are all drawn to what is good or pleasant ‘for us’, what we, as a matter of fact, find good and pleasant. But, he continues, what is good or pleasant for us is not necessarily what is good or pleasant in itself— really or objectively good or pleasant, we might say. And since he believes that virtue is the only thing that is good and pleasant in itself, he claims that only those who find virtue attractive are drawn to what is really good and pleasant. These are the virtuous; the rest of us, drawn to wealth, honour, or some other of ‘the goods that people usually fight over’—or, we might add, taking the wrong feature to be a virtue (a), only believe that we are friends of one another. But we are not, and our relationship ends as soon as wealth is lost, honour undermined, or pretend-virtue exposed for what it  Iphigeneia in Tauris, trans. W. Bynner, in Euripides, ii. Four Tragedies (Chicago, ), , . Pylades’ loyalty is manifested in his own refusal to allow Orestes to die alone: ‘If you meet knife and flame, then so do I’ ().



Alexander Nehamas

is. Orestes, then, cannot truly be Pylades’ friend unless Pylades’ loyalty really is a virtue and not something that merely seems like a virtue to him (leaving aside for now Plato’s thought that nothing depicted in tragedy could ever be a real virtue). The first part of Aristotle’s view is absolutely right: we cannot be friends of people in whom we find nothing to admire or appreciate. But because we can admire all sorts of things, he insists that we cannot be friends unless what we admire is in fact one of the relatively few human virtues. And as aretē gradually became even more restricted and finally almost exclusively moral, friendship came to seem the sole possession of the morally good: ‘Friendship is a moral good by its own nature [and] the life with such a good is, to that extent, an intrinsically moral life.’ But that cannot be right. First, because experience makes it absolutely clear that morality has little to do with our love for our friends—not to mention the fact that what constitutes a moral virtue is not something most of us are really clear about. Second, because not only do we love our friends despite their shortcomings: sometimes we love them on their account: think of the self-importance or the forgetfulness that makes your friend so dear to you and so irritating to everyone else. That, in fact, is part of what we mean when we say that we love our friends for themselves: for Aristotle, that is to say we love them for their virtues (b–). I think he is wrong, but a reversal of his view can point us in the right direction. Aristotle believes that the virtues exist objectively and that recognizing them in another is the first step towards friendship: my desire to become your friend depends on and is directed to features that you already possess. If these are truly virtues, my feelings may gradually develop into genuine friendship. Now let us agree, first, that if I am your friend I will love you on account of features of yours that I admire, although these need not be restricted to the moral virtues or even to the broader range of virtues Aristotle had in mind: your sense of humour could well be among them, although for Aristotle it would provide a basis only for a pleasure-philia; so could your taste in music, books, clothes, and who knows what else.  N. K. Badhwar, ‘Why it is Wrong to Always be Guided by the Best: Friendship and Consequentialism’, Ethics,  (), – at .  Although Aristotle knows well that people’s character changes and that friends affect one another, he says almost nothing about how virtuous one must be in order to enter into a virtue-ϕιλία and about whether, or to what extent, such a relationship can make its participants more virtuous.

Aristotelian Philia, Modern Friendship?



Second—here we move further from Aristotle’s view—let us say that these features are for me, and not necessarily for others or in actual metaphysical fact, part of who you are: losing them would turn you, as far as I am concerned, into a significantly different person, although others may well not have the same reaction. I may find that my friend’s changed political views make it impossible for our friendship to continue while you may consider them irrelevant to your relationship. Third—this is where we reverse Aristotle’s view—let us say that we are friends not because I recognize virtues that you already have but because my reasons for liking you are features of yours that I take to be virtues, whether or not they are such in the abstract: your willingness to take orders and your ability to execute them unquestioningly might well make us friends if we belong to the same criminal gang; even the vicious can be friends. In that way, countless features—not just the short list of the canonical virtues, whatever they happen to be at a given historical or cultural moment—can provide grounds for friendship. And in saying that they may not be virtues ‘in the abstract’, what I mean is that if I say that I like you for your generosity, I need not like every generous person I run into; and if I say that I like you for the intensity with which you pursue the issues that interest you, I may still actively dislike such intensity in someone else. What attracts me to you is, as we say, your generosity or your particular generosity or the way intensity manifests itself in you. It is for the very same reason that what draws me to you may be just what someone else finds indifferent or even unattractive. The question Flaubert asked regarding his two hapless clerks is also our own: ‘Why is a certain peculiarity, a certain imperfection, indifferent or baneful in one person but enchanting in someone else?’ Friendship may generate virtue as well as recognize it: do I love you because you are generous or do I admire your generosity because I love you? This transformation of Aristotle’s view may preserve part of its structure and much of what he says about the best kind of philia, but it raises more questions than it answers. Aristotle can say that we love our friends for themselves because he has a substantive conception of human nature, which accounts for what each one of us really is; and since our nature consists of our virtues, to love your friends for their virtues is by definition to love them for themselves.  Gustave Flaubert, Bouvard and Pécuchet, trans. M. Pollizotti (Urbana, Ill., ), .



Alexander Nehamas

But without such a conception, are we entitled to speak that way? What is the difference between loving you for yourself and loving you for your generosity, your good looks, your sense of humour, or (a favourite philosophical example) for your yellow hair? Why can I not say that I love you for your money? Why is it that whenever I try to explain why I love you, no matter how much I say, I always feel that I have left the most important thing unspoken? And what is that most important thing, the self, which seems to be the only real object of our love but is always left unspoken? Aristotle could answer these questions. We are still trying.

APPENDIX

‘For your sake’ vs. ‘For your own sake’ Aristotle gives us at least three ways of understanding the structure of benefit-philia: () Alcibiades is a philos of Socrates because Socrates is beneficial to him and wishes Socrates good things for his sake (a–); () Alcibiades is a philos of Socrates because Socrates is beneficial to him and wishes Socrates good things in so far as Socrates is beneficial to him (a–); () Alcibiades is not a philos of Socrates but of the beneficial instead (a–). These look incompatible, but recall that Aristotle finds it perfectly natural to refer to, say, Socrates, who is educated, or to Socrates in so far as he is educated, as ‘the educated’ or ‘the educated Socrates’. See, for example, Phys. . , b–a: ‘It is possible for a man to become educated, for the not-educated to become educated, and for the not-educated man to become an educated man.’ These, as he goes on to explain, are ‘one in number though not in form’ (a–). He makes a similar point, applied to ‘Socrates’ and ‘the sitting Socrates’ as well as to ‘being in so far as it is being’ and ‘being in so far as it is numbers or fire’, at Metaph. Α , b–, and Κ , b–a. This last passage contains some general remarks on the sense of ‘in so far as’ (hēi). With that in mind, we can rewrite sentences ()–() as follows: (a) Alcibiades is a philos of the beneficial Socrates and wishes the beneficial Socrates good things for the beneficial Socrates’ sake; (a) Alcibiades is a philos of the beneficial Socrates because Socrates is

Aristotelian Philia, Modern Friendship?



beneficial to him and wishes the beneficial Socrates good things in so far as Socrates is beneficial to him; (a) Alcibiades is not a friend of Socrates (we want to add: himself) but of the beneficial instead. All three now appear to be saying the same thing and that makes it clear that, when the ground of the philia no longer obtains, the philia, its object (the beneficial Socrates), and that for the sake of which Alcibiades wishes good things (also the beneficial Socrates) all disappear at once. In regard to virtue-philia, the first two sentences remain the same: (b) Alcibiades is a philos of the virtuous Socrates and wishes the virtuous Socrates good things for the virtuous Socrates’ sake; (b) Alcibiades is a philos of the virtuous Socrates and wishes the virtuous Socrates good things in so far as the virtuous Socrates is virtuous. But in this case we cannot also say: (b) Alcibiades is not a friend of Socrates but of the virtuous Socrates instead. The reason is that although Socrates is beneficial (or pleasant) accidentally, he is virtuous in himself (b–). And while Socrates can continue to be although he has ceased to be beneficial, it is very difficult if not impossible for Socrates to lose his virtue, which is permanent (b; but see b–). In Aristotelian terms, whereas Socrates and the beneficial Socrates are one in number but different in form, Socrates and the virtuous Socrates are one in both. So there is at least a logical or metaphysical difference between ‘for the beneficial philos’s sake’ and ‘for the virtuous philos’s sake’—a difference we can express by speaking of ‘the philos’s sake’ in the first case and ‘the philos’s own sake’ in the latter.

Princeton University

BIBLIOGRAPHY Adkins, A. W. H., ‘ “Friendship” and “Self-Sufficiency” in Homer and Aristotle’, Classical Quarterly,   (), –. Aelred of Rievaulx, Spiritual Friendship, trans. M. E. Laker (Kalamazoo, Mich., ). Agamben, G., ‘Friendship’, Contretemps,  (), –. Annas, J., ‘Plato and Aristotle on Friendship and Altruism’, Mind,  (), –.



Alexander Nehamas

Badhwar, N. K., ‘Why it is Wrong to Always be Guided by the Best: Friendship and Consequentialism’, Ethics,  (), –. Blum, L. A., Friendship, Altruism, and Morality (London, ). Blundell, M. W., Helping Friends and Harming Enemies: A Study in Sophocles and Greek Ethics (Cambridge, ). Bostock, D., Aristotle’s Ethics (Oxford, ). Broadie, S. (comm.), and Rowe, C. (trans.), Aristotle: Nicomachean Ethics (Oxford, ). Cicero: De amicitia, trans. F. Copley (Ann Arbor, ); repr. in Pakaluk (ed.), Selves. Cooper, J. M., ‘Aristotle on the Forms of Friendship’, in id., Reason and Emotion: Essays on Ancient Moral Psychology and Ethical Theory [Reason] (Princeton, ), –. ‘Friendship and the Good in Aristotle’, in id., Reason, –. ‘Political Animals and Civic Friendship’, in id., Reason, –. Derrida, Jacques, The Politics of Friendship (New York, ). Eliot, G., Middlemarch (Boston, ). Euripides, Iphigeneia in Tauris, trans. W. Bynner, in Euripides, ii. Four Tragedies (Chicago, ). Flaubert, Gustave, Bouvard and Pécuchet, trans. M. Pollizotti (Urbana, Ill., ). Friedman, M., What are Friends For? Feminist Perspectives on Personal Relationships and Moral Theory (Ithaca, NY, ). Greenblatt, S., Renaissance Self-Fashioning: From More to Shakespeare (Chicago, ). Heath, M., The Poetics of Greek Tragedy (Stanford, Calif., ). Herman, G., Ritualized Friendship and the Greek City (Cambridge, ) Jollimore, T., Friendship and Agent-Relative Morality (New York, ). ‘Impartiality’, in E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy 〈http://plato.stanford.edu/archives/fall/entries/impartiality/〉. Kahn, C. H., ‘Aristotle and Altruism’, Mind,  (), –. Konstan, D., Friendship in the Classical World (Cambridge, ). La Rochefoucauld, François, duc de, Maxims, trans. S. D. Warner and S. Douard (South Bend, Ind., ). Lewis, C. S., The Four Loves (Orlando, Fla., ). Montaigne, Michel de, ‘Of Friendship’, in The Complete Works of Montaigne, ed. and trans. Donald M. Frame (Stanford, Calif., ), –. Nietzsche, F., Beiträge zur Quellenkunde und Kritik des Laertius Diogenes (Basel, ). Pakaluk, M., (trans. and comm.), Aristotle: Nicomachean Ethics Books VIII and IX (Oxford, ). Aristotle’s Nicomachean Ethics: An Introduction (Cambridge, ).

Aristotelian Philia, Modern Friendship?



(ed.), Other Selves: Philosophers on Friendship [Selves] (Indianapolis, ). Pangle, L. S., Aristotle and the Philosophy of Friendship (Cambridge, ). Railton, P., ‘Alienation, Consequentialism and the Demands of Morality’, Philosophy and Public Affairs,  (), –. Rorty, R. M., ‘Human Rights, Rationality, and Sentimentality’, in S. Shute and S. Hurley (eds.), On Human Rights: Oxford Amnesty Lectures (New York, ), –. Scanlon, T. M., What we Owe to Each Other (Cambridge, Mass., ). Silver, A., ‘Friendship in Commercial Society: Eighteenth-Century Social Theory and Modern Sociology’, American Journal of Sociology,  (), –. Smith, C. G., ‘Spenser’s Theory of Friendship’, PMLA,  (), –. Spenser, Edmund, The Faerie Queene, ed. Thomas Roche, Jr. (London, ). Telfer, E., ‘Friendship’, Proceedings of the Aristotelian Society,  (– ), –; repr. in Pakaluk (ed.), Selves, –. [Tucker, Abraham], The Light of Nature Pursued by Edward Search, Esq., iii/ () [quoted from K. Thomas, The Ends of Life: Roads to Fulfilment in Early Modern England (Oxford, )]. Vernon, M., The Philosophy of Friendship (New York, ). Vlastos, G., ‘The Individual as Object of Love in Plato’, in id., Platonic Studies, nd edn. (Princeton, ), –. Whiting, J., ‘The Nicomachean Account of ϕιλία’, in R. Kraut (ed.), The Blackwell Guide to Aristotle’s Nicomachean Ethics (Malden, Mass., ), –. Williams, B., Ethics and the Limits of Philosophy (Cambridge, Mass., ).

This page intentionally left blank

SENS A T I O N A N D SCEPTICI S M I N P L O T I N U S SARA MAGRIN

P’ epistemology is centred on the idea that knowledge is not a disposition of our soul but a principle of reality: Nous (see esp. . . –). Human beings can come to share in this self-subsisting knowledge because since birth they have latent within their intellectual faculty all the forms (. . ), which they can reactivate by using their ‘discursive reason’ (dianoia). For Plotinus, Nous is both knowledge of forms and the forms themselves, and in it there is no distinction between subject and object. His conception of Nous as a form of self-knowledge reflects Aristotle’s observations in Metaphysics Λ , but is also presented as a solution to some sceptical threats. In two places, . . – and . . , Plotinus suggests, in fact, that the lack of distinction between subject and object is a necessary condition for knowledge. Whenever the object is other than the knowing subject, he argues, the latter can only grasp an ‘image’ of it, rather than ‘the thing itself’, but knowledge, he observes, has to be of ‘the things themselves’. It has often been remarked that this contrast between ‘images’ and ‘things themselves’ is inspired by the ancient sceptical debates on the cognitive power of the senses. But © Sara Magrin  There are several people who helped me enormously in writing this paper. I would especially like to thank Stephen Menn for many stimulating discussions and several rounds of comments. I am very grateful to Cristina D’Ancona, Eyjólfur Emilsson, Brad Inwood, Paul Kalligas, Claude Panaccio, Brian van den Broek, and an anonymous referee for their challenges and their invaluable suggestions. Earlier drafts of this paper were presented at the annual meeting of the ACFAS held in Quebec City in May , and at the annual meeting of the International Society for Neoplatonic Studies held in New Orleans in June . I am grateful to those audiences for their comments, and, in particular, to Luc Brisson and the late Steven Strange.  On Nous as self-subsisting knowledge see S. Menn, Plato on God as Nous (Carbondale, ), –.  See e.g. S.E. PH . ; . . The remark goes back to Émile Bréhier’s Notices to Enn. .  and .  in his Budé edition of the Enneads (É. Bréhier (ed. and trans.),



Sara Magrin

the nature of Plotinus’ engagement with the ancient sceptical tradition is still controversial. The problem, in a nutshell, is the following: granted that Plotinus manifests some interest in this tradition, can we actually say that he builds his epistemology as an answer to scepticism, or should we rather conclude that he uses scepticism merely as a polemical foil? R. T. Wallis, who was the first to draw attention to the role of scepticism in Plotinus, argued that Plotinus’ epistemology was a direct answer to what he called ‘the Sceptics’ epistemological critique’, by which he meant the epistemological critique which Charlotte Stough regarded as the basis of the ancient sceptical tradition. For Stough, ancient scepticism was basically an attack against empiricist theories of knowledge, and it rested on the view that the senses do not reveal ‘the things themselves’, but only the ‘affections’ (pathē) that sensible things cause in a perceiver, which are like ‘images’ of those things. Taking Stough’s thesis for granted, Wallis suggested that Plotinus endorsed the sceptics’ critique, that he extended it to all the theories of knowledge in which the object known was viewed as external to the knowing subject, and that he answered it by introducing a conception of knowledge as self-knowledge. The problem with this suggestion is that Stough’s reading of ancient scepticism is questionable, and it is no longer the standard one. We have come to conceive of the ancient sceptical tradition as dealing with several different epistemological challenges rather than with a single epistemological critique, and we no longer interpret that tradition as centred on the view that the senses perceive only ‘images’ of their objects. This is not to say that the ancient sceptics Plotin: Ennéades,  vols. (Paris, –)), and the point has been examined in a number of more recent studies that I am about to introduce.  I shall consider only those studies that, though starting from .  and . , offer an overall reconstruction of the role of scepticism in Plotinus’ epistemology. For a detailed analysis of .  in the light of specific sceptical challenges on the possibility of self-knowledge see I. Crystal, ‘Plotinus on the Structure of Self-Intellection’, Phronesis,  (), –, and W. Kühn, Quel savoir après le scepticisme? Plotin et ses prédecesseurs sur la connaissance de soi (Paris, ). Kühn suggests that Plotinus’ conception of Nous as self-knowledge is an answer to sceptical challenges, but his very rich analysis of .  and . , like Crystal’s, focuses exclusively on the issue of self-knowledge and self-intellection.  R. T. Wallis, ‘Scepticism and Neoplatonism’, in ANRW ii. .  (), –.  C. L. Stough, Greek Skepticism: A Study in Epistemology (Berkeley and Los Angeles, ).  For an overview of the problems with Stough’s intepretation see M. Frede, ‘Review of Greek Skepticism’, Journal of Philosophy,  (), –.

Sensation and Scepticism in Plotinus



never put forward such a view; they did, but it is now thought that they used it, together with other observations, only in the interest of polemics, without ever endorsing it or turning it into their main weapon. Now, if there is no such thing as ‘the Sceptics’ epistemological critique’, in contrast to what Wallis assumed, it becomes rather difficult to maintain that Plotinus built his epistemology in an attempt to answer it. A different interpretation of the role of scepticism in Plotinus has been put forward by Eyjólfur Emilsson. Rather than attacking Wallis’s conception of ‘the Sceptics’ epistemological critique’ directly, Emilsson suggests that, had Plotinus endorsed that critique, he should have maintained that the senses reveal only ‘subjective affections’ of things, whereas he maintains exactly the opposite, for he explicitly claims that the senses reveal ‘the things themselves’ rather than ‘images’ (. . . –). For Emilsson, Plotinus’ theory of sensation is a direct realist theory that rests on the view that we directly perceive sensible things as they are. Plotinus, he suggests, has a ‘common sense’ conception of sensible things (the expression is Emilsson’s), in the sense that he holds that they are ‘objectively’, i.e. by themselves and independently of any perceiving subject, just such as they are perceived. From these remarks Emilsson concludes that Plotinus has no reason to be genuinely interested in sceptical challenges against the cognitive power of the senses. Plotinus is interested in sceptical arguments, he argues, and he does use the sceptical distinction between ‘images’ and ‘things themselves’, but views these arguments and this distinction only as an aid to his polemics against competing philosophical schools. Although I entirely agree that for Plotinus we know the sensible objects themselves in sensation, rather than the affections they cause on our sense-organs, I think that this does not exclude on his part an interest in scep See G. Striker, ‘The Ten Tropes of Aenesidemus’ [‘Ten Tropes’], in ead., Essays on Hellenistic Epistemology and Ethics (Cambridge, ), – at .  Emilsson deals extensively with the issue of scepticism in Plotinus in his ‘Plotinus on the Objects of Thought’, Archiv für Geschichte der Philosophie,  (), –, where he relies on his analysis of Plotinus’ theory of sensation in his Plotinus on Sense-Perception: A Philosophical Study [Sense-Perception] (Cambridge, ). The main thesis of his  work on Plotinus’ theory of sensation is defended again in his more recent ‘Plotinus on Sense Perception’, in S. Knuuttila and P. Kärkkäinen (eds.), Theories of Perception in Medieval and Early Modern Philosophy (Dordrecht, ), –, and the main thesis of his  article is defended again in his Plotinus on Intellect [Intellect] (Oxford, ), –. There are no substantial differences in the arguments.



Sara Magrin

tical challenges that goes beyond the needs of polemics. It all depends on whether perceiving ‘the things themselves’ really means for Plotinus perceiving them just as they are by themselves. This is Emilsson’s opinion, but there is room for disagreement, as has been pointed out by Dominic O’Meara. O’Meara argues that Plotinus’ epistemology is not only an answer to scepticism, but is actually grounded on scepticism. Like Augustine and Descartes, he says, Plotinus uses sceptical arguments to discard sense-based theories of knowledge, and he grounds on these arguments the view that knowledge can only rest on intelligible principles, i.e. the forms. In keeping with Wallis’s reading, O’Meara suggests that for Plotinus knowledge cannot depend on the senses because the senses reveal only ‘affections’ or ‘images’ of their objects, but he adds a crucial remark about the meaning of ‘affection’. He observes that by ‘affections’ or ‘images’ Plotinus, in contrast to the sceptics, does not mean ‘subjective affections’, viz. something like private ‘feelings’, but means ‘objective affections’, namely affections that, while falling short of revealing their objects, are none the less the expression of those objects’ causal power. I think that O’Meara’s reading of the role of scepticism in Plotinus is right, but I do not find his argument entirely convincing. For he finds support for the ‘objective’ interpretation of what counts as an affection in Plotinus only in a fairly obscure aspect of the latter’s metaphysics—namely, the analysis of so-called ‘affections’ that the One produces on Nous when Nous thinks of it (. . ). O’Meara does not show that his ‘objective’ interpretation holds also in the case of specifically sensory affections. But unless this is shown, his reading of the role of scepticism in Plotinus’ epistemology remains questionable. Today, as far as I can see, the debate on scepticism in Plotinus is at a stalemate. All the scholars who have dealt with this topic have concentrated almost exclusively on . . – and . . , and not much remains to be said about these passages. Furthermore, even if Stough’s reading of ancient scepticism is no longer the standard one, her ‘images vs. things themselves’ theme is still the only scep D. J. O’Meara, ‘Scepticism and Ineffability in Plotinus’, Phronesis,  (), –.  In speaking of ‘feelings’, O’Meara has especially in mind the analysis of the Cyrenaics’ ‘affections’ offered by V. Tsouna in The Epistemology of the Cyrenaic School (Cambridge, ), .  I say ‘almost exclusively’ because O’Meara’s remarks on affections rest on . . .

Sensation and Scepticism in Plotinus



tical theme used to assess Plotinus’ views. Finally, given the centrality of this theme in the debate, it would seem that, before discussing Plotinus’ approach to scepticism, one should examine his theory of sensation. But there is very little discussion of this subject in the literature, and the only relevant work is Emilsson’s  essay Plotinus on Sense-Perception. I think that the role of scepticism in Plotinus needs to be reconsidered afresh, without presupposing any interpretation of his theory of sensation, without prioritizing the ‘images vs. things themselves’ theme, and without focusing primarily on the two loci classici in the debate, namely . . – and . . . This is what I intend to do here. Since we lack any hard evidence concerning the extent of Plotinus’ knowledge of the ancient sceptical tradition, and, more generally, concerning his knowledge of the Hellenistic epistemological debates, with which that tradition is mostly associated, my remarks will inevitably involve some amount of speculation, and my argument will ultimately be hypothetical in nature. But the reader should be aware that, while I intend to argue that Plotinus is deeply engaged with the sceptical tradition, I do not want to suggest that this engagement simply runs parallel to his more obvious interest in Plato and Aristotle, nor do I want to maintain that some aspects of his epistemology can be better understood by looking at the ancient sceptics rather than at Plato and Aristotle. Plato and Aristotle are Plotinus’ two most important philosophical interlocutors, and I do not question their status here. Rather, my suggestion is that Plotinus uses material from the sceptical tradition to interpret Plato and Aristotle, and that he proceeds in this way either because he detects the presence of such material in their texts or because he thinks that sceptical arguments can be appropriately employed to show the relevance of some of their observations for his own epistemological project. I want to start by presenting the context in which Plotinus’ engagement with scepticism is to be addressed in my view: his anti This is not to say that everyone agrees with Emilsson’s direct realist reading, but no one has tried to refute it either. There are some critical remarks against it in T. S. Ganson, ‘The Platonic Approach to Sense-Perception’, History of Philosophy Quarterly,  (), – at . An interpretation of Plotinus’ theory of sensation that points in a different direction from Emilsson’s is suggested by A. C. Lloyd, The Anatomy of Neoplatonism [Anatomy] (Oxford, ), –, and S. Rappe, ‘SelfKnowledge and Subjectivity in the Enneads’, in L. P. Gerson (ed.), The Cambridge Companion to Plotinus (Cambridge, ), –. Both Lloyd and Rappe tend towards an idealist reading, but neither of them aims to present an account of Plotinus’ theory of sensation.



Sara Magrin 

Stoic polemics. It is fair to say that Plotinus tends to present the Stoics as his most threatening adversaries. On the surface, the reasons for this are obvious. He holds that real being is intelligible, and that the soul is incorporeal, whereas the Stoics maintain exactly the opposite: they think that bodies are the only real beings, and that soul is a body. One can find entire batteries of arguments against Stoic corporealism scattered throughout the Enneads, but I would like to draw attention to one passage in particular: . . . –. This passage is part of an attack against the Stoic categories, and it is important because in it Plotinus isolates the core philosophical commitment that, in his view, leads the Stoics to assign real being to bodies. They [i.e. the Stoics] give non-being [τὸ µὴ ὄν] the first rank as that which is most of all being and so rank the last first. The cause of this is that sensation became their guide and they trusted it for placing the principles and the rest [αἴτιον δὲ ἡ αἴσθησις αὐτοῖς ἡγεµὼν γενοµένη καὶ πιστὴ εἰς ἀρχῶν καὶ τῶν ἄλλων θέσιν]. For they believed that bodies were the real beings [τὰ γὰρ σώµατα νοµίσαντες εἶναι τὸ ὄν], and, since they were afraid of their transformation into each other, they thought that what persisted under them was reality. (Enn. . . . –, trans. Armstrong, slightly modified)

These remarks are to be interpreted against the background of the Stoic reading of Tim.   –  . There Plato describes the pre-cosmic state of the universe. He says that, since water, earth, air, and fire at this stage change continuously into each other, it is difficult to say what each of them is, and one can at best point to their substrate, i.e. the receptacle, and say that they are ‘This’. The Stoics, who identify the receptacle with matter, argue that matter has to be the substance of bodies. Plotinus agrees that the receptacle represents matter (see . . –, and my observations below), but rejects the view that matter is bodies’ substance. Matter, he says, is far from being substance, since it is non-being (see . . , and my  I shall often speak of ‘Stoics’, and this broad classification might seem dubious, but in dealing with Plotinus one is often forced to adopt it, for he pays little or no attention to differences of opinion among Stoic thinkers.  On this see S. Menn, Descartes and Augustine (Cambridge, ), –.  The passages in which Plotinus engages with the Stoics have been collected by A. Graeser, Plotinus and the Stoics: A Preliminary Study (Leiden, ).  A. H. Armstrong (ed. and trans.), Plotinus: The Enneads,  vols. (Cambridge, Mass., –).  For a summary of this reading see D. N. Sedley, ‘Stoic Physics and Metaphysics’, in K. Algra, J. Barnes, J. Mansfeld, and M. Schofield (eds.), The Cambridge History of Hellenistic Philosophy (Cambridge, ), – at –.

Sensation and Scepticism in Plotinus



observations below). The Stoics think that matter is substance, he suggests, only because they believe that bodies are real beings, and, worried that their intertransformation could undermine their ontological status, they ground their existence on the only thing that persists throughout their changes, namely, matter (SVF i.  = LS ). This is not to say that matter is not a body. We know that for the Stoics matter, too, is a body; but they view it as a simple, passive, and completely malleable body, capable of acquiring certain corporeal qualities or hexeis which intrinsically dispose it in this or that way so as to form a compound body, i.e. a natural body (SVF ii.  = LS ; SVF ii.  = LS ). The most important point in the passage, however, is Plotinus’ remark that, had the Stoics not ascribed real being to bodies, they would not have ‘confused’ matter with substance. The reason that leads the Stoics to ascribe real being to bodies, Plotinus says, is their trust in sensation. The point is compressed, but we can unfold it thus: inspired by Soph.  , the Stoics hold that the mark of being is the capacity to act and be acted upon (SVF i.  = LS ; SVF ii.  = LS ). Since they believe that truth is to be reached through the senses, and sensation seems to indicate that this capacity belongs to bodies, they conclude that real being belongs to bodies only. Plotinus, then, thinks that at the origin of the Stoics’ corporealism there is an epistemological presupposition concerning the cognitive power of sensation, and he maintains that it is precisely this presupposition that leads them to draw ‘wrong’ conclusions on the ontological status of bodies and matter. However, in .  he provides no real argument in support of his criticism. To find this argument, I suggest, we have to look at . , the treatise On the Impassivity of the Bodiless. Plotinus’ most immediate concern there is to show that no incorporeal entity can suffer ‘affections’ (pathē), and the discussion is clearly divided into two parts; the first part (chapters –)  A. A. Long and D. N. Sedley, The Hellenistic Philosophers [LS],  vols. (Cambridge, ).  Plotinus does defend his views in . , but his criticisms of the Stoics’ theory of categories in . . – are always very compressed and too often biased to shed light on his position. On Plotinus’ analysis of the Stoics’ theory of categories see M. Mignucci, ‘The Stoic Notion of Relatives’, in J. Barnes and M. Mignucci (eds.), Matter and Metaphysics (Naples, ), –, and S. Menn, ‘The Stoic Theory of Categories’, Oxford Studies in Ancient Philosophy,  (), –.



Sara Magrin

focuses on the impassivity of soul; the second part (chapters –) on that of matter. At the beginning of .  Part II, Plotinus claims that being is not as most people believe [ὡς οἱ πολλοὶ νοµίζουσιν]. For being, that which can be truly called being, is ‘the real being’ [ἔστι γὰρ τὸ ὄν, ὃ καὶ κατ᾿ ἀλήθειαν ἄν τις εἰποι ὄν, ὄντως ὄν]. (Enn. . . . –)

The context shows that ‘real being’ (ὄντως ὄν) is the being of soul, Nous, and forms and that it is not ‘as most people believe’, because it is intelligible, whereas ‘the many’ take it to be sensible. This rather dogmatic claim about being invites an obvious objection, which runs as follows: How can the nature of bodies be non-being? Or matter, on which bodies are based, mountains, rocks, the whole solid earth, all things that offer resistance [ἀντίτυπα] and by their impact force what they strike to acknowledge their being [ταῖς πλαγαῖς βιαζόµενα τὰ πληττόµενα ὁµολογεῖν αὐτῶν τὴν οὐσίαν]? (Enn. . . . –, trans. Fleet)

If bodies are not real beings, an opponent could ask, how do you explain the evidence with which they seem to be so to our senses? Plotinus does not reveal the identity of his opponent here, but he probably has the Stoics in mind, for the claim that bodies ‘force what they strike to acknowledge their being’ (ταῖς πλαγαῖς βιαζόµενα τὰ πληττόµενα ὁµολογεῖν αὐτῶν τὴν οὐσίαν) recalls a passage in Sextus Empiricus, M. . , where Sextus discusses the Stoic criterion of truth, namely the cataleptic representation. On Stoic doctrine, a ‘cataleptic representation’ (katalēptikē phantasia) is a representation that may or may not be sensory, but that, in any case, is true,  Plotinus’ analysis of the impassivity of the soul in .  is generally read against the background of the Stoics’ theory of passions rather than in relation to their theory of sensation, in contrast with my procedure here. While I do not deny that Plotinus engages with the issue of the passions of the soul in . . –, I hope to show that, by focusing on the epistemological rather than on the ethical aspects of his discussion, we can perhaps find a deeper link between the two parts of the treatise than that provided by the general theme of the impassivity of the bodiless. Moreover, that Plotinus’ focus in .  is on epistemology, and in particular on the cognitive value of sensation, seems to be suggested by the opening lines of the treatise, where the discussion of the impassivity of the soul is introduced in order to further develop and support the view that sensation is not an affection of the soul (. . . –).  B. Fleet (trans. and comm.), Plotinus: Ennead III. . On the Impassivity of the Bodiless [Ennead III. ] (Oxford, ).  That the Stoics are Plotinus’ opponents here is argued also by Fleet, Ennead III. , in his commentary ad loc. However, Fleet does not point to S.E. M. . .

Sensation and Scepticism in Plotinus



accurate, and endowed with a special, distinctive mark that makes it infallibly distinguishable from a non-cataleptic one (S.E. M. . – = LS ). Sensation is the assent to a cataleptic representation, and it is an ‘apprehension’ (katalēpsis, SVF ii. ). At M. .  Sextus reports that for the Stoics, whenever a cataleptic representation is not obstructed by contingent factors (such as a poor condition of the sense-organs, a false and prejudicial belief, and so on), it is always ‘evident and striking’ (ἐναργὴς . . . καὶ πληκτική), so that, as if it were pulling one by the hair, it literally forces one to assent to it. Plotinus may not have read Sextus, but seems familiar with the views he reports. In the light of these views, in . .  he concedes that the evidence of our sensory representations could make us believe that bodies are real beings, but issues a warning: although bodies seem to be real beings, you should refrain from thinking that they are so. This warning introduces a long discussion of the impassivity of matter, which is inspired by Plato’s description of the receptacle in the Timaeus. Let me briefly reconstruct the background against which Plotinus works. After distinguishing three orders of reality (the intelligible paradigm, its visible copy, and the receptacle), at Tim.    ff. Plato describes the state of the receptacle before the constitution of the cosmos. At this stage, the four elemental bodies—earth, water, air, and fire—have not yet fully come into being, but the receptacle already contains some ‘traces’ (ἴχνη) of them by which it is irregularly ‘inflamed’, ‘moistened’, and so on. It is by giving order to these ‘traces’ through some ‘forms’ and ‘numbers’, i.e. some ‘shapes’ (  ), that the demiurge produces the elemental bodies and the world. Plotinus thinks that the world is eternal, and that the creation myth of the Timaeus is only a narrative device, used for clarity’s sake. He interprets Plato’s receptacle as matter; he takes the ‘traces’ to represent sensible qualities (see e.g. . . . –); and he identifies the ‘forms’ and ‘numbers’ that the demiurge imposes on the ‘traces’ with some intelligible powers that he calls logoi, and that he views as the hidden  I draw from M. Frede, ‘Stoics and Skeptics on Clear and Distinct Impressions’ [‘Stoics and Skeptics’], in id., Essays in Ancient Philosophy (Minneapolis, ), –.  On the Platonist interpretations of the Timaeus before and after Plotinus see M. Baltes, Die Weltentstehung des platonischen Timaios nach den antiken Interpreten,  vols. (Leiden, –), and D. N. Sedley, Creationism and its Critics in Antiquity  (Berkeley, ), –. See Lloyd, Anatomy, .



Sara Magrin

causes of sensible qualities (. . . –). In .  Part II he observes that Plato presents the receptacle as the universal substrate of all the things that come to be (Tim.   ), and he reasons that, if it is to fulfil this role, it must be impassive, which means that it cannot possess as its own ‘affections’ (pathē) any of the traits that the things that come to be in it display. Plotinus introduces this observation as the conclusion of some philosophical arguments of his own, but he has no doubt that this is also Plato’s considered view in the Timaeus (cf. Tim.   – ). However, he is aware that to show that his interpretation is correct, he needs to engage in the exegesis of some difficult passages which, if taken at face value, do not seem amenable to his reading. The most problematic of these is   –, where Plato suggests that whatever comes to be in the receptacle, does so in virtue of the ‘affections’ (pathē) caused by some ‘forms’ (µορϕαί). The affections in question are the receptacle’s ‘traces’, and they are characterized as ways in which the receptacle ‘is moistened’ (ὑγραινοµένη) and ‘inflamed’ (πυρουµένη). On the other hand, the forms that cause these affections are the ‘copies [µιµήµατα] of the forms’ (Tim.   –), which, Plato claims, ‘enter and exit’ the receptacle ‘shaping it through’ (διασχηµατίζεσθαι). But how could anything be ‘moistened’, ‘inflamed’, etc. without being affected? And how are some ‘forms’ supposed to enter and exit the receptacle, thus ‘shaping it through’, if not by affecting it? Plotinus deals with these difficulties in . . . I shall begin with the first one, for it is more tractable, even though this inverts the order in which they are examined in the text. When Plato claims  See L. Brisson, ‘Logos et logoi chez Plotin: leur nature et leur rôle’, Cahiers philosophiques de Strasbourg,  (), –. For a comprehensive analysis of the notion of λόγος in Plotinus see also P. Remes, Plotinus on Self: The Philosophy of the ‘We’ (Cambridge, ), –, to be compared with P. Kalligas, ‘Logos and the Sensible Object in Plotinus’, Ancient Philosophy,  (), –. The differences between Remes’s and Kalligas’s positions mainly pertain to the nature of the λόγοι of complex entities (e.g. the λόγος of ‘human being’ vs. that of ‘white’). Here, however, I shall try to focus on sensible qualities and their λόγοι, and I shall deal only tangentially with the issue of bodies’ λόγοι.  Plotinus maintains that matter must be impassive for the following reasons: qua universal substrate it must be an eternal reservoir, but if it was affected, in the course of time it would have to be consumed (. . ); contraries are the only things that can suffer affections from each other because they have a common genus, but matter has no contrary, being completely other than anything else (this conclusion depends on Aristotle’s notion of contrariety: . . . –); matter must be available for the reception of any form, but if it was affected, it would acquire forms that would obstruct any further reception (. . . –).  Plotinus, too, suggests that this is the correct order in which the two difficulties

Sensation and Scepticism in Plotinus



that the receptacle ‘is inflamed’, Plotinus says, he does not mean to say that it is ‘set on fire’ (πυροῦσθαι), for if this was the case, the receptacle would indeed suffer some affections, i.e. some physical changes. Plato, he goes on, wants to maintain that the receptacle ‘becomes’ (γίνεσθαι) fire, and this is not a way of being affected. Plotinus arrives at the conclusion that ‘becoming’ fire is not a way of being affected by noticing how Plato describes the genesis of the ‘traces’: the receptacle, Plato says, is inflamed, moistened, and so on ‘receiving the forms of air and water’ (καὶ τὰς ἀέρος καὶ ὕδατος µορϕὰς δεχοµένην, . – = Tim.   –). This precision shows somehow in Plotinus’ view that it is not with affections that we are dealing here. The observation is peculiar, but allows us to infer that Plotinus rests the view that the ‘traces’ are not affections of the receptacle on the kind of relation that the copies of the forms entertain with it, and on the special kind of ‘shaping’ they cause. Let us examine, then, the second difficulty, i.e. how Plato can maintain that some forms ‘enter’ matter without affecting it. This is the way in which the issue is addressed at the beginning of . . : He [i.e. Plato] wants to show . . . how matter itself would remain unaffected when receiving the forms [i.e. the copies of the forms], seeking an example [παράδειγµα] of unaffected participation. It would not be easy in any other way to explain what things exactly, when they are present, allow the substrate to remain the same. He raised many aporiai in his eagerness to achieve his aim [i.e. showing matter’s impassivity] and, furthermore, in his desire to show the deficiency of substantial being in the objects of perception, and that the space of semblances is vast [τὴν χώραν τοῦ εἰκότος οὖσαν πολλήν]. He therefore framed a hypothesis [ὑποθέµενος] that it is by shapes that matter produces affections in soulless things, although matter itself has none of those affections; in this way he indicates its permanence and allows us to draw the conclusion that matter does not even undergo affection and alteration from the shapes [τὴν οὖν ὕλην σχήµασιν ὑ ποθέµενος τὰ πάθη ποιεῖν τοῖς ἀψύχοις σώµασιν οὐδὲν αὐτὴν ἔχουσαν τούτων τῶν παθηµάτων τὸ µένον ταύτης ταύτην ἐνδείκνυται διδοὺς συλλογίζεσθαι, ὡς οὐδὲ παρὰ τῶν σχηµάτων ἔχει τὸ πάσχειν αὐτὴ καὶ ἀλλοιοῦσθαι]. For one might say that altershould be examined. He explains what it means to be ‘moistened’ or ‘inflamed’ without suffering affections after discussing how matter is ‘shaped’ by the ‘forms’, but says that the former point is to be analysed before the latter (. . . ).  Note that Plotinus maintains that it is the receptacle that ‘becomes’ (γίνεται) the ‘traces’, whereas Plato says that ‘we seem to see the thing that we have just now called water [ὃ δὴ νῦν ὕδωρ ὠνοµάκαµεν] condensing and becoming [γιγνόµενον] stones and earth’ (Tim.   – ).



Sara Magrin

ation takes place in those bodies when they receive one shape in succession to another—using the term ‘alteration’ homonymously to mean a change of shape. But since matter has no shape, nor even size, how could one even homonymously say that the presence of shape in any degree was an alteration? (Enn. . . . –, trans. Fleet, slightly modified)

Here Plotinus reconstructs from his own perspective the way in which Plato explains the unaffected participation of the receptacle in the copies of the forms, i.e. the kind of participation that brings about the ‘traces’. He suggests that Plato proceeds as a geometer: he has a problem to solve—to show that the receptacle participates in the copies of the forms without being affected by them—and he sets out a hypothesis—that it is by shapes that matter produces the affections in soulless things. Then, with this hypothesis at hand, Plato reasons that, if he can show that the affections in soulless things are caused through shapes, he can conclude that matter remains unaffected when receiving the copies of the forms. The suggestion that Plato proceeds by hypotheses in the Timaeus has solid textual grounds. But Plotinus is likely to have in mind also other texts in which Plato stresses the crucial role of hypotheses in philosophical investigations, such as Phaedo   –  , Rep.   –  , and especially Meno  – , where the use of hypotheses specifically in geometry is invoked to explain recollection. In any case, for Plotinus, Plato’s reasoning in the Timaeus rests on the implicit identification of the copies of the forms  Cf. Arist. Phys. . , b–, where we read that change of shape is not a type of alteration.  The claim, in the passage above, that ‘matter produces affections in soulless things’ should not be taken to mean that matter is actually the proper cause of those affections. Matter seems to be a necessary, but not a sufficient, condition for the affections of soulless things. Matter’s ποιεῖν is an echo of Plato, Rep.  – (cf. also Soph.  –  and   –), where Plato speaks of mirrors ‘producing’ things. Plotinus, who claims in .  that matter is like some sort of invisible mirror, ascribes to it the same quasi-causality (cf. . . . –). However, we should refrain from taking the mirror analogy at face value. For Plotinus’ rejection of a straightforward  application of the analogy see . . . See esp. Tim.  – .  An interesting account of the use of geometry in the Timaeus can be found in Plut. Quaest. conv. . , where the question is: ‘What Plato meant by saying that God is always doing geometry’. Plutarch shows that geometry could be understood both as a means to train the mind in the grasp of the intelligible and as a divine method for imposing order upon matter and bringing about the world, with no contrast between the two. Here Plotinus seems to allude to geometry as a means to improve our understanding of the world; but the possible allusion to recollection suggests that the ‘geometer-style’ hypotheses of the Timaeus are not arbitrary, but are meant to lead us to the true order of things. I come back to this point below. For

Sensation and Scepticism in Plotinus



with some ‘shapes’ (σχήµατα), and, as Paul Kalligas has remarked, the passage of the Timaeus that he has in mind here is   –   . After explaining the nature of the four primary elements in terms of shapes and introducing the inanimate bodies that derive from their transformations, Plato enquires into the causes of ‘the affections’ (pathēmata) of these bodies (  –), i.e. the causes of their sensible qualities. He claims that the pathēmata of soulless bodies cannot be explained independently of the nature of our own bodies, sensation, and mortal soul, and that, conversely, we cannot grasp the latter without an account of the pathēmata. Finding it impossible to give an account of all these things at once, he decides to take the nature of sensation, body, and mortal soul for granted, and to explain first the genesis of sensible qualities (  – ). It is at this point that the shapes to which Plotinus refers are introduced. For Plato seems to suggest that soulless bodies have the sensible qualities they have primarily in virtue of the shapes of their constitutive particles. Thus, for instance, fire’s heat consists in ‘the sharp experience’ produced by its pyramids on our flesh. Since Plato mentions other causes of sensible qualities besides the shapes (e.g. the swiftness of the particles’ motion in relation to heat; their quantity and position in relation to heavy and light), one wonders why Plotinus neglects them. The reason is that he wants to simplify the Timaeus’s account of the pathēmata, because he believes that even the shapes, despite the primary causal role that Plato assigns them, are ‘an example’ used for the sake of clarification. For Plotinus, Plato in the Timaeus is not really committed to the idea that the pathēmata come about through shapes, but introduces the shapes to make his notion of unaffected participation easier to grasp. But Plotinus’ reinterpretation of the notion of ‘recollection’ see R. A. H. King, Aristotle and Plotinus on Memory (Berlin, ). 

See P. Kalligas (ed.), Πλωτίνου ᾿ Εννεὰς Τρίτη (Athens, ), . I call the παθήµατα of soulless bodies ‘sensible qualities’ only because Plato says that they are things such as ‘hot’, ‘cold’, and so on, but I do not intend to make any special claim about their ontological status and their nature at this point.  Cf. . . . –: ‘And indeed [we must also ask] where and how about plane and solid and all the shapes: for it is certainly not we who merely think [ἐπινοεῖν] the shapes. The shape of the universe, which was before us, is evidence of this, and the other natural shapes in the things which exist by nature, which must exist before the bodies as shapes without a shape there in the intelligible [ἃ δὴ ἀνάγκη πρὸ τῶν σωµάτων εἶναι ἀσχηµάτιστα ἐκεῖ], and primary shapes. For they are not shapes in something else, but since they are themselves belonging to themselves there was no need for 



Sara Magrin

why should shapes make things easier? Consider the way in which Plotinus goes on: Thus in this instance it would not be out of place to use the term ‘colour by convention’ and ‘the other things by convention’, because the underlying nature ‘has’ nothing the way it is usually believed to [εἴ τις οὖν ἐνταῦθα τὸ νόµῳ χροιὴ καὶ τὰ ἄλλα νόµῳ λέγοι τῷ τὴν ϕύσιν τὴν ὑ ποκειµένην µηδὲν οὕτως ἔχειν, ὡς νοµίζεται, οὐκ ἂν ἄτοπος εἴη τοῦ λόγου]. But in what way does matter ‘have’ the forms, if we are not even satisfied with the suggestion that it has them as shapes? Well, Plato’s hypothesis [ὑπόθεσις] at least gives us such indication as can be given of the impassivity of matter and of the apparent presence of a kind of image which is not really present [ἔχει ἔνδειξιν ἡ ὑπόθεσις ὡς οἷόν τε τῆς ἀ παθείας καὶ τῆς οἷον εἰδώλων οὐ παρόντων δοκούσης παρουσίας]. (Enn. . . . –, trans. Fleet, slightly modified)

The first lines of this passage suggest that Plotinus detects a Democritean influence at Tim.   ff. In fact, when he says that, all things considered, it would not be out of place to suggest that for Plato colour and ‘the other things’ are ‘by convention’, he is pointing to Democritus B / DK: ‘For he says: “By convention sweet, by convention bitter, by convention hot, by convention cold, by convention colour; but in reality atoms and void”’ (“νόµῳ” γάρ ϕησι “γλυκὺ καὶ νόµῳ πικρόν, νόµῳ θερµόν, νόµῳ ψυχρόν, νόµῳ χροιή· ἐτεῇ δὲ ἄτοµα καὶ κενόν”). This allusion to Democritus allows us to see why Plotinus interprets the identification of the copies of the forms with some shapes as a move that Plato makes for clarity’s sake. It is because the shapes are meant to remind readers of the Timaeus of them to be extended: the extended shapes belong to other things. Shape, then, is always one in real being, but it has distinctions in it . . . But I mean “has distinctions” not in the sense that it has acquired size, but because it has been divided, each part of it in correspondence to each being, and given to the bodies there in the intelligible, as to fire there, if you like, to the pyramid there’ (trans. Armstrong, slightly modified).  For this unusual rendering of µηδὲν οὕτως ἔχειν, ὡς νοµίζεται, which intuitively should be translated as ‘is not as it is usually believed to be’, see below.  Plotinus’ use of Democritean material has been mostly overlooked so far. T. Gelzer, ‘Plotins Interesse an den Vorsokratikern’, Museum Helveticum,  (), –, does not examine Democritus, and the suggestion is that Plotinus has no interest in him. The only study I am aware of that considers the use Plotinus makes of Democritean material is G. Stamatellos, Plotinus and the Presocratics (Albany, NY, ), –. However, Stamatellos thinks that Plotinus read at least some of Democritus’ works, whereas I can find no evidence for this conclusion. Furthermore, Stamatellos does not attempt to reconstruct how Plotinus interprets Democritus’ epistemology and ontology.  See the Index fontium in P. Henry and H.-R. Schwyzer (eds.), Plotini opera,  vols. (Oxford, ), iii. .

Sensation and Scepticism in Plotinus



a notion they are already familiar with, i.e. the Democritean notion of the atomic shapes. Briefly, to show how matter can remain impassive if the affections of soulless bodies are caused by shapes, Plotinus recommends that we look to Democritus’ account of sensible qualities at B / . This suggestion is of course problematic; for what Democritus might have meant in B / is still a matter of dispute. But for our purpose we do not need to sort out what Democritus thought; we only need to figure out what Plotinus takes him to have thought. From the passage quoted above (. . . –) we can see that for Plotinus, to say that colour and ‘the other things’ (i.e., probably, the other sensible qualities) are ‘by convention’ is to maintain that ‘the underlying nature’, i.e. the receptacle, ‘ “has” nothing the way it is usually believed to [νοµίζεται]’. It follows that Plotinus reads Democritus’ claim that colour, sweet, etc. are ‘by convention’ as meaning that we ‘believe’ things to possess these qualities, whereas in fact they have none of them. Hence, for Plotinus, Democritus holds that sensible qualities are merely subjective appearances, in the sense of illusions to which nothing corresponds in the world. In the light of these observations, we can reconstruct Plotinus’ reasoning as follows. Plato believes that the receptacle, i.e. matter,  In other words, Plotinus identifies the copies of the forms with some shapes on the grounds of Plato’s remarks in the Timaeus, but suggests that the shapes are ‘an example’ used for clarity’s sake (a suggestion that Plato does not make in the Timaeus) because he thinks that they are meant to stand for Democritus’ shapes. If we assume that Plotinus has in mind Democritus’ shapes here, we can make good sense of his claim that the shapes of the Timaeus are only ‘an example’.  What Plotinus means here by ‘the other things’ is unclear. He could be referring to sensible qualities, but also to sensible qualities and bodies in general. However, since he is interpreting Democritus, and since Democritus does not seem to have held that bodies are ‘by convention’, it is perhaps more prudent to suggest that Plotinus means ‘other sensible qualities’ here. For this interpretation of Democritus see P.-M. Morel, Démocrite et la recherche des causes (Paris, ), and M.-K. Lee, Epistemology after Protagoras [Epistemology] (Oxford, ), –. We know that Democritus might not have believed that all sensible qualities are ‘by convention’ either (the status he ascribes to weight is especially controversial), but this falls outside our interest, for Plotinus analyses every quality in the same way.  Someone might object that, since Plotinus is examining matter here, what he holds is that we believe matter rather than ‘things’ to have qualities that in fact it lacks. But note that, even though Plotinus is indeed discussing the nature of the receptacle, he examines the relation between receptacle and qualities starting from Tim.   –  , where what is at stake are the qualities of soulless bodies. Thus, his focus is on matter as the substrate of bodies, and, in my view, this is why he points out that we believe sensible things and not merely matter in general to have this or that quality.



Sara Magrin

is impassive, and it cannot be affected by the copies of the forms. To explain how it comes to acquire its ‘traces’, that is, the sensible qualities that it displays, Plato appeals to Democritus. He identifies the copies of the forms with some shapes that are meant to recall the atomic shapes in Democritus, and he claims that sensible qualities appear to be on the matter of this or that soulless body (cf. Tim.   ) when these shapes encounter our body (flesh and sense-organs). Plato does not really believe in shapes (nor in void and atoms), but uses a familiar notion to convey in simpler terms his own theory of the unaffected participation of the receptacle in the copies of the forms. For this theory is too remote from our ordinary way of thinking to be grasped directly, because it does not rely on shapes or other familiar notions, but on some special, intelligible entities, i.e. the logoi. It is in terms of logoi that the copies of the forms must be understood. But the logoi are fairly obscure entities, and their modes of operation are bound to be obscure as well. It is better to think of them as shapes, and to find in the shapes’ behaviour as close an approximation as possible of their operations. As Democritus thinks that sensible qualities are merely subjective appearances, so does Plato. For Plato, we only believe the matter of a body to be hot or cold, and so on, when in fact matter has none of the qualities it displays, for no quality ever ‘enters’ it, and, hence, no quality is ever intrinsic to it. You might think, then, that the upshot of Plotinus’ analysis of the impassivity of matter is that the sensible world is merely an illusion, some sort of ‘stage-painting’. In a sense, this is precisely what Plotinus wants to suggest. He argues that, viewed merely as series of sensible things supposedly made only of matter and qualities, the sensible world is only an illusion. Bracket the logoi, he suggests, and all there is around you is merely a subjective appearance to which nothing corresponds in the world, and that has no stability, like the ‘traces’ in the pre-cosmic universe of the Timaeus. This suggestion, however, is polemical. That the sensible world is merely an  It is true that Plotinus’ account of the ‘traces’ of the Timaeus presupposes the existence of perceiving subjects at the pre-cosmic stage, whereas humans and animals in general do not yet exist at this stage for Plato. But Plotinus discards this problem as one that pertains to a literal reading of the dialogue. For him the pre-cosmic stage represents what the world would be in the absence of intelligible principles. Plotinus perhaps reads Plato’s words at   –, ‘we seem to see the thing that we have just now called water [ὃ δὴ νῦν ὕδωρ ὠνοµάκαµεν]’, as meaning ‘the thing that we have now conventionally called water’.

Sensation and Scepticism in Plotinus



illusion is not Plotinus’ view; it is the view to which he wants to reduce the Stoic conception of bodies as matter–quality compounds. Hints of Plotinus’ anti-Stoic polemic, in fact, can be found throughout the discussion of the impassivity of matter in . . Consider the way in which B / is introduced. Plotinus claims that it is not out of place to say that colour is ‘by convention’ (. . . –) because in fact matter ‘“has” nothing the way it is usually believed to’. The Greek runs: µηδὲν οὕτως ἔχειν, ὡς νοµίζεται. The most intuitive way of reading this phrase would be to take οὕτως ἔχειν as periphrastic and translate: ‘It [i.e. matter] is not as it is usually believed to be.’ But notice what Plotinus asks next: ‘In what sense does matter “have” [ἔχει] the forms?’ (ll. –). This question suggests that he wants to put some unusual emphasis on the ἔχειν in µηδὲν οὕτως ἔχειν, ὡς νοµίζεται, and his persistent use of ἔχειν throughout his discussion of the participation of matter in the copies of the forms in . . – confirms the impression (e.g. . . . ; . –; . ; . –). Plotinus wants to know whether matter quite literally ‘has’ or does not ‘have’ the sensible qualities that it displays, because the Stoics maintain that qualities are hexeis, namely bodies that matter ‘has’, and by which it is intrinsically disposed so as to form other, compound, bodies, i.e. the natural bodies that we perceive. By saying that matter does not ‘have’ any quality, Plotinus wants to suggest that the Stoics’ interpretation of the receptacle in terms of a passive body ready to be shaped by qualities is ‘wrong’, and it leads to the likewise ‘incorrect’ view that matter could have intrinsic qualities, so as to form with them a bodily compound. Then, recall Plotinus’ claim in . .  that being is not ‘as most people believe’ (ὡς οἱ πολλοὶ νοµίζουσιν), namely sensible, but ‘truly’ (κατ᾿ ἀλήθειαν) is intelligible. In the light of his use of B / in . . , it is natural to suggest that this claim too reflects a Democritean influence. For the contrast between what ‘truly’ is, and what is merely a matter of ‘belief’ is regularly used by some authors, most prominently Sextus and Aetius, to explain the νόµῳ–ἐτεῇ opposition  This is not to say that the Stoics are Plotinus’ only polemical target; for in . . – he also attacks the Aristotelian conception of bodies as form–matter compounds. However, note that Plotinus tends to criticize the Aristotelian conception of bodies by crudely assimilating it to the Stoic one, which, thus, emerges as the conception that he most strenuously opposes. This strategy is particularly evident at the beginning of . . , where Plotinus misleadingly presents a form–matter compound as something that comes to be when form and matter are ‘so to say’ (οἷον) mixed together and co-affected (. . . –). Forms, for Aristotle, do not mix with matter; it is the Stoics’ qualities that mix with matter.



Sara Magrin 

in Democritus. In . .  Plotinus, then, draws on some paraphrases of B / (perhaps by Sextus, or by Aetius, or by some other unknown thinker—we cannot know); he develops the νόµῳ– ἐτεῇ opposition in terms of a contrast between what ‘truly’ is and what is a matter of ‘belief’; and, finally, he uses this contrast to convey his own opposition between sensible and intelligible being. But, as in . .  he is concerned with the Stoics, we can conclude that he uses B / to suggest that the Stoics, like most people, confuse what really is with what is merely a matter of belief, i.e. sensible qualities and bodies (when these latter are viewed as matter–quality compounds). Thus, through the analysis of the receptacle in .  Part II Plotinus argues that the Stoic conception of matter and bodies leads to the conclusion that the sensible world is only a series of illusions. Bodies, on this view, are whatever they are only extrinsically and in relation to a perceiver, he suggests. Now, if bodies are whatever they are only relatively to a perceiver, they certainly cannot be real beings, for real beings, both for Plotinus and for the Stoics, must  See Lee, Epistemology, –. Lee has drawn a map of the ancient families of interpretation of B /. Quotations and/or paraphrases of the fragment can be found in D.L. . , , , ; in Galen, On Medical Experience . ; On the Elements according to Hippocrates . ; in Plutarch, Adv. Col.  –; in Aetius, . . ; in Sextus, PH . –, M. . . Lee remarks that Sextus and Aetius have very similar interpretations that differ substantially from those of Galen and Plutarch. Since Plotinus in . .  reveals that he relies on paraphrases for his knowledge of Democritus, perhaps he never read Democritus’ own works.  Cf. . . . –, where Plotinus summarizes his argument in . .  by saying that it was directed against those who place real beings in the class of bodies, and who rest this view on the contents of their sensations. Another anti-Stoic element in the discussion of the impassivity of matter can be found, perhaps, in . . . –. There Plotinus excludes the possibility of interpreting Plato’s claim that matter is inflamed at Tim.    as meaning that matter is ‘set on fire’. He argues that this cannot be what Plato wants to suggest, because otherwise Plato would be committed to the view that matter is turned by fire into some fiery stuff, which corresponds to the sensible fire that we ordinarily experience. This idea that matter would be turned into fire by some other fire, Plotinus holds, is absurd, for it amounts to saying that a statue comes to be when the statue itself ‘takes a walk’ through some bronze (ll. – ). This remark, I take it, is Plotinus’ version of a criticism of the Stoic theory of categories that is voiced most famously by Plutarch at Comm. not.  –  (= LS ). Plutarch observes that, on Stoic doctrine, every natural body is both matter, since matter is the substance of all bodies (and it is the first category), and ‘qualified’ (ποιόν, the second category), and he polemically concludes from this that the Stoics turn each thing into two objects that mysteriously share the same place, and the same properties. The observation is crude, because for the Stoics matter is a component of a ‘qualified object’ rather than a distinct body sharing the same space with it; but Plotinus seems happy to echo Plutarch’s criticism none the less.

Sensation and Scepticism in Plotinus



have at least some intrinsic differentiations. But, to maintain that on the Stoics’ account bodies are whatever they are only relatively and extrinsically, Plotinus needs some further argument. So far he has argued that sensible qualities cannot be intrinsic to matter only on the grounds of his own view that matter must be impassive. But obviously the Stoics have no reason to accept such a counter-intuitive metaphysical position, and thus they are entitled to reject the conclusions it leads to. Plotinus must show why, on the Stoic account, sensible qualities can only be subjective appearances without relying on his own metaphysics. It is to this end that he engages with the notion of cataleptic representation. Let us consider again Plotinus’ claim that Plato in the Timaeus formulates a hypothesis about sensible qualities, namely that they are produced by shapes. Plotinus must think that Plato has some grounds for formulating this specific hypothesis, but what are they? Since Plotinus takes Plato’s shapes to have the same function as Democritus’, if we can see why in Plotinus’ view Democritus posits his own shapes, we should have also some indication of how he reconstructs Plato’s reasoning. In the light of the sources available to us, it seems that Democritus posits the shapes on the basis of an argument from conflicting appearances. He observes that the appearances of one and the same thing conflict, and, thinking that one and the same thing cannot really have contrary qualities at the same time, he concludes that sensible things by themselves have none of the qualities that we perceive. Then, he infers that the nature of things consists in some shapes that cause different and even opposite effects on differently disposed perceivers. That Plotinus reads Democritus along these lines is suggested by his description of matter at . . . –. I have noted that for Plotinus matter is nonbeing rather than the substance of bodies, in contrast to what the Stoics think, but consider the way in which he describes this nonbeing: Truly non-being [ἀληθινῶς µὴ ὄν], an image and a phantasma [ϕάντασµα] of mass, a yearning for substantial existence; . . . it always displays contrary images on itself [τὰ ἐναντία ἀεὶ ἐϕ᾿ ἑαυτοῦ ϕανταζόµενον], small and great,  Our best source in this regard might be Sextus. See Lee, Epistemology, . Lee observes that Sextus is the source that most clearly presents Democritus’ discussion of sensible qualities and their relation to shapes in the context in which it was probably developed by Democritus himself, i.e. as part of an enquiry about the limits of the cognitive power of the senses.



Sara Magrin

less and more, deficiency and excess, . . . and its being in a semblance is no being at all, but like some elusive illusion. (Enn. . . . –, trans. Fleet, slightly modified)

Here Plotinus draws on several sources: Aristotle’s description of the Indefinite Dyad (Metaph. Α , b; Ν , b–); the ὄντως µὴ ὄν of Soph.   ; and the ‘others’ of Hypothesis VII of the Parmenides. But what interests me is his own description of matter’s non-being, which rests on the claim that matter ‘always displays contrary images on itself’ (τὰ ἐναντία ἀεὶ ἐϕ᾿ ἑαυτοῦ ϕανταζόµενον). This phrase certainly picks up Tim.   –  , but sums up the content of that passage by echoing a formula that at least from Aristotle’s time is used to refer to the phenomenon of conflicting appearances—τὰ ἐναντία περὶ τὸ αὐτὸ ϕαίνεσθαι—and that is used by Aristotle himself to describe Democritus’ account of sensible things. Plotinus’ appeal to this formula reveals that he does want to discuss the phenomenon of conflicting appearances, and that he is engaged in the epistemological issues Democritus deals with according to our sources. As the ‘contrary images’ that appear on matter are the sensible qualities that we believe matter to possess, the passage shows, I think, that for Plotinus the illusory appearances that matter displays are conflicting ones. Let us see, then, whether Plotinus could have found any grounds for thinking that Plato posits the shapes starting from the premiss of conflicting appearances. Is there anything in the way in which the pathēmata are introduced at   ff. that might lead one to infer that Plato works on this implicit premiss? Plato claims that we cannot explain the nature of the pathēmata independently of our sensations, body, and mortal soul. But, then, he goes on to describe the pathēmata by taking our knowledge of these last three things for granted. This poses a problem: since the assumption is that we do not know these three things yet, how are we to make any sense of the account of the pathēmata? Plato, one could say, must rely on at least some previous clues about the nature of sensation, body, and mortal soul, and these clues are to be found at Tim.  – . There sensations are identified with some kind of violent and contrary motions that enter the soul through the body and end up disrupting the circles of the rational, immortal soul. These violent motions, which  

See Fleet, Ennead III. , . Fleet does not mention Hypothesis VII. Metaph. Γ , b–; cf. S.E. PH . –.

Sensation and Scepticism in Plotinus



at    Plato calls pathēmata, are produced by the encounter of our body with external fire, earth, water, and air ( ), and when they reach the rational soul they cause its circles to move in twisted and opposite ways. As a consequence, we begin to see things as the opposite of what they are: what is on our left suddenly appears on our right, and vice versa; and the soul, confused, says that this is ‘the same as that’ or that this is ‘different from that’, when in fact the matter is the other way round. At  –  Plato is particularly interested in the contrast that arises between the motions of the irrational soul and those of the rational one. But he might be taken to suggest that some conflict is present also among the motions of the irrational soul alone, because he says that these motions contribute to the ‘tumult’ of the irrational soul (  –). Plotinus, I suggest, seizes upon this remark, and he reads Tim.  –  as Plato’s avowal of the conflicting nature of sensory appearances, which he identifies with the violent movements that cause the rational soul to see things as the opposite of what they are. Having done so, he concludes that Plato posits the shapes in the light of the same reasoning from conflicting appearances that leads Democritus to posit his own shapes. This emerges if we take a broader look at the structure of the argument for the impassivity of matter in . . As I have said, the discussion of matter begins at . . , and it is preceded by the analysis of the impassivity of the soul in . . –. In Part I of the treatise Plotinus shows that there are no alterations in the soul nor motions, but only ‘activities’ (energeiai). Naturally, he is fully aware that his position might be taken to conflict with Plato’s own, since Plato very often speaks of ‘motions of the soul’, and the ‘violent motions’ of which sensations consist in the Timaeus are a case in point. Thus, Plotinus needs to reinterpret Plato’s language and to show that what Plato means by ‘motions of the soul’ are not physical or quasi-physical changes, but activities. At . . . – he suggests, by echoing Plato, that what ‘enters’ the soul in sensation is a phantasma. A phantasma, I take it, is simply the content of the ‘sensory phantasia’ (αἰσθητικὴ ϕαντασία) that he mentions a few lines before, in . . . , which, in turn, can only be identified with the ‘undecidable phantasia’ (ἀνεπίκριτος ϕαντασία), and the ‘obscure quasi-belief’ (ἀµυδρὰ οἷον δόξα) he also mentions in . . , at line . 

Cf. Tim.   – ;   – ;   –  ; Phileb.   –  .



Sara Magrin

Now, Plotinus is not innovating when he claims that our sensory phantasiai are ‘obscure quasi-beliefs’, but most likely has in mind a passage from Alexander’s De anima (p.  Bruns) where we find a classification of sensory representations that is at least in part of Stoic origin. Alexander claims that sensory ‘representations’ (phantasiai) are primarily of two kinds: ‘obscure’ (ἀµυδραί) and ‘intense’ (σϕοδραί). The most important and unifying characteristic of our ‘obscure representations’, he says, is that they are never cataleptic, because they always represent their object in a confused and indistinct manner. As other sources inform us, the Stoics define an ‘obscure representation’ by opposition to a ‘clear’ one. If clear representations are such as to report all the characteristics of a sensible object, or at least all those characteristics that are most relevant for claiming that that object is the type of thing it is, an obscure representation is, in contrast, one that does not report all or the most relevant traits of its object, so that it represents it in a confused way. A typical obscure representation for the Stoics is one that is caused by a distant object, when we cannot figure out what it is that we are looking at. But Plotinus is far from innovating even when he describes our sensory representations as ‘undecidable’ (anepikritos). In fact anepikritos is a very common adjective in the Hellenistic and imperial epistemological debates, where it is used in contexts in which a sceptical position is introduced or discussed. We can find it in Aristocles’ report on Pyrrho ap. Euseb. PE . . –, and it is employed in a somewhat technical sense by Sextus and Galen to describe the ‘disagreement’ (diaphōnia) that dogmatic opinions fall into, and sometimes to characterize representations (cf. PH . ), as in Plotinus. A disagreement of opinions is said to be ‘undecidable’ when we cannot arrive at a ‘judgement’ (krisis) to sort it out; that is, when we cannot ‘evaluate’ (epikrinein) which opinions are true and which are false. The case of an ‘undecidable representa Cf. Frede, ‘Stoics and Skeptics’, . I say that Alexander’s classification is ‘at least in part’ Stoic because it starts with a distinction between ‘intense’ (σϕοδραί) and ‘obscure’ (ἀµυδραί) representations, and, while the Stoics did view some representations as ‘obscure’, it is not clear that they ever labelled a representation as ‘intense’. It is Carneades who, influenced by Stoic vocabulary, divides what he calls the ‘apparently true representation’ (ἡ ϕαινοµένη ἀληθὴς ϕαντασία) into two types: the ‘obscure’ one (ἀµυδρά), and the ‘intense’ one (σϕοδρά); see S.E. M. . .  On this see Frede, ‘Stoics and Skeptics’, .  e.g. S.E. PH . ; Galen, On the Sects for Beginners .  and .

Sensation and Scepticism in Plotinus



tion’ is analogous. A representation is ‘undecidable’ when we cannot judge whether it is true or false, because it conflicts with other representations of the same object. A typical example of an ‘undecidable representation’ in Sextus is the Stoics’ so-called ‘cataleptic representation’. It is certainly true that the Timaeus is not the only Platonic text that Plotinus has in mind in the first part of . . But, since it is the chief text with which he engages in the second part of the treatise, it seems reasonable to suggest that the ‘undecidable’ and ‘obscure’ representations introduced in . .  are the violent and contrary motions that Plato describes at Tim.  – . These motions are ‘violent and contrary’ for Plotinus because they are obscure and conflicting beliefs; and they are ‘undecidable’ because they confuse the rational soul so as to make it see things in twisted and opposite ways, thus rendering it unable to judge the truth. I therefore suggest that in .  Plotinus reconstructs Plato’s discussion of sensation and sensible qualities in the Timaeus. In his view, at   Plato introduces the shapes of the elements as part of a hypothesis that is used to show the impassivity of matter, after explaining that sensory appearances conflict at  – . Plotinus does the same: he explains why matter is impassive in the second part of . , after analysing our sensory phantasiai in terms of conflicting appearances in the first part of the treatise. In so doing, he makes explicit the reason why, in his view, Plato frames the hypothesis that the pathēmata of soulless bodies are produced through shapes: it is on the grounds of an inference to the best explanation that aims to account for the nature of things starting from the phenomenon of conflicting appearances. In the light of the background against which Plotinus works when he says that sensory representations are ‘undecidable’ and ‘obscure’, it seems safe to conclude that he wants to make a polemical point against the Stoic notion of cataleptic representation. He maintains that the Stoic distinction between cataleptic and non-cataleptic sensory representations cannot hold, for all our sensory representations are ‘obscure’ and ‘undecidable’, and thus non-cataleptic. Interestingly, Tim.  – , which Plotinus reads as presenting an account of the ‘obscure’ and conflicting nature of sensory representations, is a text where Plato examines children’s  On ‘disagreement’ as a form of ‘undecidable’ conflict that depends on contradiction see Striker, ‘Ten Tropes’, –.



Sara Magrin

sensations. This is significant because the Stoics maintain that cataleptic representations are available only to adult human beings as opposed to children and animals. By alluding to that passage, Plotinus polemically suggests that there are no ‘special’ sensory representations that belong exclusively to adults, but that, as far as sensory representations are concerned, adults are on a par (or at least almost on a par) with children. He does not explain why no sensory representation can be cataleptic, but this is probably because he implicitly relies on Academic and Pyrrhonist arguments. In particular, he seems to presuppose and use Arcesilaus’ ‘Indiscernibility thesis’ (ἀπαραλλαξία), namely the thesis that there are indiscernible objects in the world. It is a basic tenet of Stoic ontology that each body in the world (except matter) is uniquely differentiated or ‘peculiarly qualified’ (ἰδίως ποιόν), and the Stoics rest on this tenet the very possibility of cataleptic representations. For it is because each object is uniquely differentiated from all the others on their view that it can cause a distinct representation through which it can be infallibly grasped. By pointing at cases such as that of indistinguishable twins or grains of sand, Arcesilaus argues, in contrast, that there are indiscernible objects in the world, and that in their case a true and a false representation are indistinguishable in point of evidence, that is in terms of phenomenal content. Plotinus endorses Arcesilaus’ critique, I think, when in .  Part II he suggests that on the Stoic account of matter and bodies not only can there be no uniquely qualified object but there can be no intrinsically differentiated object at all. To be sure, the Stoics answer Arcesilaus’ challenges (see, for instance, Sphaerus’ story in D.L. .  = LS ), and they engage Academics and Pyrrhonists in a long and complex debate on the cognitive value of sensory representations that lasts for centuries. Plotinus does  Cf. Sen. De ira . . , where animal representations are said to be turbidae and confusae. For the confusion that appearances cause in the soul see also Epictetus’ Discourses, . . Plotinus’ scattered remarks about animal sensation are most obscure. In this paper I shall deal exclusively with his observations on human sensation.  The fortune of this thesis among Plotinus’ Platonist predecessors is well attested: see C. Brittain, ‘Middle Platonists on Academic Scepticism’ [‘Middle Platonists’], in R. W. Sharples and R. Sorabji (eds.), Greek and Roman Philosophy  –  ,  vols. (London, ), ii. –.  See D. N. Sedley, ‘The Stoic Criterion of Identity’, Phronesis,  (), –.  Plotinus’ position can be helpfully examined in the light of the reconstruction of this debate offered by R. J. Hankinson, ‘Natural Criteria and the Transparency

Sensation and Scepticism in Plotinus



not ignore this debate, but when in .  he says that all sensory representations are ‘obscure’ and ‘undecidable’, he suggests that, despite all the moves and counter-moves of the Stoics, the problems raised by Arcesilaus against the notion of cataleptic representation have never been solved in his view. What is distinctive of Plotinus’ anti-Stoic polemics, however, is that it rests on a sceptical reading of Democritus that reflects Aristotle’s observations in Metaphysics Γ  and the Democritean views of the sceptical circles of the late fourth century . Plotinus develops this reading in the light of Academic and Pyrrhonist reflections on the nature of sensible objects and sensory representations, and he argues that the Stoics’ reliance on the cognitive power of the senses leads them, paradoxically, to a form of Democritean scepticism. The Stoics argue that bodies are real beings, he suggests, because they trust sensation, and more precisely, because they think that there are cataleptic representations. But, through Democritus, one can show that these representations are in fact a red herring, and that the only conclusion to which sensation can lead is that bodies are illusions. But Plotinus’ use of Democritean scepticism cannot be merely polemical; for even if he views Democritus as a sceptic, he suggests that the atomic shapes, qua antecedents of the Platonic ‘shapes’ or logoi, are an important step in the direction of a ‘correct’ understanding of the sensible world. Since Plotinus is committed to the existence of the logoi, and he infers their existence from sceptical views about sensible qualities and sensation, he has to endorse these of Judgement: Antiochus, Philo and Galen on Epistemological Justification’, in B. Inwood and J. Mansfeld (eds.), Assent and Argument: Studies in Cicero’s Academic Books (Leiden, ), – at –.  The most prominent figures in these sceptical circles are Anaxarchus of Abdera and Metrodorus of Chios—who thought that the world was only a ‘stage-painting’ (Cic. ND . ;  A  DK)—and Pyrrho, who apparently held Democritus in high esteem (D.L. . ).  There is no evidence that the Stoics were ever engaged with the late th-cent.  strand of Democritean scepticism in which Plotinus is interested, but we know that Aristo, at least, was inclined to assimilate Arcesilaus’ views to Pyrrho’s (D.L. . – = LS ). Even though Aristo held unorthodox views, this could mean that, independently of any real historical affiliation, Arcesilaus was at least perceived by some Stoics as the heir of Pyrrho and of the late th-cent. sceptics. If this is the case, then Plotinus’ goal in attacking the Stoics by means of Democritean material could be viewed as an attempt to reduce their position precisely to that form of scepticism that they tried to oppose by attacking Arcesilaus. However, the relation between Pyrrho, other late th-cent. sceptics, and Arcesilaus is controversial: see R. Bett, Pyrrho, his Antecedents, and his Legacy (Oxford, ),  n. .



Sara Magrin

views. In other words, he has to maintain that qualities are, at least to some extent, a matter of belief, and that sensory representations are all ‘obscure’ and ‘undecidable’, because it is from these premisses that he can infer the presence of logoi in the world. But, and this is his problem, he also wants to show that he can go beyond the scepticism to which these premisses led Democritus. His strategy is roughly as follows. He argues that, in contrast to atomic shapes and corporeal qualities, the logoi can provide bodies with some intrinsic differentiations. Then he maintains that, in so far as they are caused by logoi, sensible qualities are not illusions, but actually manifest the intrinsic differentiations of bodies, which can therefore be apprehended in sensation, albeit only in an ‘obscure’ manner. Finally, he claims that the ‘obscure’ and ‘undecidable’ character of our sensory representations cannot prevent us from knowing in a clear way what things really are, because the nature of sensible things is not sensible but intelligible. To know what sensible things really are by themselves, Plotinus suggests, to know their nature, we need to stop looking at them, and start considering our own representations as objects of enquiry in their own right. Let me explain, first, how Plotinus argues for the presence of intrinsic differentiations in bodies. He maintains that every natural body has an intelligible structure that consists in a special kind of logos, namely, the logos of a natural kind. This logos is like a ‘seed’, and it ‘contains’ in an organized way other, subordinate logoi, those of the sensible qualities that belong by nature, i.e. essentially, to that type of body, including its size. The logos of a natural kind is the real locus of a body’s qualities, which, properly speaking, are intelligible entities. Take white-lead: Plotinus holds that white-lead is not white because it appears so to a perceiver’s eyes, in contrast to what Democritus held in his view, but because the logos of white is part of its intelligible nature, white being among the essential properties of white-lead. The logos  This is what emerges especially from the following passage: ‘So when the form [εἶδος] comes to the matter it brings everything with it; the form has everything, the size and all that goes with [µετά] and is caused by [ὑπό] the λόγος. Therefore, in every natural kind the dimensions are determined along with [µετά] the form; the dimensions of a man are different from those of a bird, and those of different kinds of birds from one another. Is there anything more surprising in the bringing of quantity to matter as something different from itself than in the addition to it of quality? It is not the case that quality is a λόγος and quantity is not, since quantity is form [εἶδος] and measure [µέτρον] and number [ἀριθµός]’ (Enn. . . . –, trans. Armstrong). For the seed simile see esp. . . .

Sensation and Scepticism in Plotinus



of white is an intelligible ‘white’ that of course can never be instantiated in any sensible particular, but its presence ‘in’ white-lead, i.e. in the intelligible nature of white-lead, makes white-lead not only appear to be white, but also intrinsically possess and really display whiteness. One could say, in other words, that the logos of white makes white-lead ‘manifest’ sensible whiteness. Thus, some qualities, i.e. the essential qualities of a body, are intrinsic to the body in which they are perceived to inhere for Plotinus, by being in that body in the form of a logos. The relation between these intelligible qualities and their sensible manifestation is explained through the so-called ‘double activity theory’. This is a theory of metaphysical causality according to which each thing (be this sensible or not) has two ‘activities’ (energeiai), one of which is internal to the thing and is the activity of its intelligible substance, while the other is external to the thing and is the activity caused by its substance (. . . –). To see how this theory works in the case of sensible things we can take the example of fire. Fire for Plotinus contains in its intelligible structure the logos of heat, which is its internal activity, and in virtue of this logos it also produces a sensible heat through which it warms any nearby body of suitable nature—your hand, for instance, or the air in a room. It has been suggested, and persuasively in my view, that the doubleactivity theory should be interpreted against the background of two Aristotelian passages: DA . , b–, and Physics . . The distinction between internal and external activity, in this perspective, is analogous to that drawn in DA . , b–, between the state of having knowledge, a first actuality, and its exercise. Starting from Aristotle’s remarks in DA . , Plotinus argues that fire’s internal activity is a disposition to heat analogous to a first actuality, and he contrasts this disposition to heat with its actual exercise, which corresponds to the production of heat. Then he appropriates one of the main conclusions of Physics . , viz. the thesis that the activity of the agent is the same as the activity of the patient, and in  For the presence of λόγοι ‘in’ bodies see C. D’Ancona, ‘Le rapport modèleimage dans la pensée de Plotin’, in D. De Smet, M. Sebti, and G. De Callataÿ (eds.), Miroir et savoir: la transmission d’un thème platonicien, des Alexandrins à la philosophie arabo-musulmane (Leuven, ), – at –.  A. C. Lloyd, ‘Plotinus on the Genesis of Thought and Existence’, Oxford Studies in Ancient Philosophy,  (), – at –. Lloyd’s remarks on Plotinus’ debt to Aristotle’s observations in Physics .  rest on C. Rutten, ‘La doctrine de deux actes dans la philosophie de Plotin’, Revue philosophique,  (), –.



Sara Magrin

the light of it he argues that the external activity of the agent is the same as the effect produced by that activity on a patient. Thus, fire’s sensible heat, for instance, is the same as the heat of the nearby air or of your hand. Precisely this conception of the external activity of a thing is explicitly introduced in . , in the context of a discussion about light and luminosity: The activity within the luminous body, which is like its life [ἔστι δὲ ἡ µὲν ἐν αὐτῷ ἐνέργεια καὶ οἷον ζωὴ τοῦ σώµατος τοῦ ϕωτεινοῦ], is greater and is a kind of source and origin of its [external] activity [οἷον ἀρχὴ τῆς ἐνεργείας καὶ πηγή]; that which is adjacent to the limit of the body [ἡ δὲ µετὰ τὸ πέρας τοῦ σώµατος], an image [εἴδωλον] of that within, is a second activity which is not separated [οὐκ ἀϕισταµένη] from the first. For each thing that exists has an activity, which is a likeness [ὁµοίωµα] of itself, so that while it exists that likeness exists, and while it stays in its place the likeness goes far out [ϕθάνειν εἰς τὸ πόρρω], sometimes a longer, sometimes a lesser distance; and some activities are weak and dim, and some even indiscernible [αἱ δὲ καὶ λανθάνουσαι], but other things have greater activities which go far; and when an activity goes far, one must think that it is there where the active and powerful thing is [ὅπου τὸ ἐνεργοῦν καὶ δυνάµενον], and again there at the point it reaches [καὶ αὖ οὗ ϕθάνει]. (Enn. . . . –, trans. Armstrong, slightly modified)

A luminous body (e.g. the sun), Plotinus says here, is a body that has light within in the form of a power or disposition to illuminate the air and other bodies. We never see the internal light of this body, as it is intelligible, but we can see its ‘light-like’ product, namely the luminosity that the body manifests. This luminosity is an external activity, as it is the exercise of a disposition to illuminate, and it is present both all around the luminous body and on any other body that it can reach in virtue of its strength. This view that the external activity is both in a sensible thing and in the patient it reaches is not precisely in harmony with what Aristotle’s says in Physics . ; for Aristotle claims there that the activity of the agent is actually in the patient, and only in the patient. But I shall come back to this shortly. In .  Plotinus uses the account of light and its external activity he gives in the passage above to explain colour transmission in vision. In vision, he suggests, a sensible colour is present both in  This is never explicitly maintained in .  nor is it said elsewhere, but the problem with which Plotinus struggles in .  is precisely that of accounting for colour transmission. Thus, it must be the case, I think, that the double-activity theory that

Sensation and Scepticism in Plotinus



the body in which it is perceived and in the eyes, as it is the external activity of a logos, and external activities are both in the active agent and on any other body they can reach. But how is this presence of a sensible quality in two things at once to be explained? First of all, I think, we must keep looking to Aristotle, and in particular to DA . , b–. There Aristotle uses Physics .  to explain the transmission of sensible qualities in sensation, and he claims that in sensation the activity of the sensible (e.g. sound) and that of the sense (e.g. hearing) are always one and the same. Plotinus, I submit, follows his lead: after explaining the relation between logoi and sensible qualities on the basis of DA . , b–, and Physics . , he uses the observations of Physics .  to point out that the activity of the sensible (e.g. a colour) must be the same as that of the sense (e.g. sight). Once again, however, while, in line with Physics . , Aristotle maintains in the De anima that the activity of the sensible is in the sense (a–), Plotinus, in contrast, holds that it is both in the sensible and in the sense. Thus, if white-lead appears white to me, for Plotinus this is because the whiteness it manifests on its surface is at once on its surface and in my eyes. Why, then, does Plotinus depart from Aristotle on this point? I suggest that, by speaking of a sensible quality that is both in the sensible object and in the sense, Plotinus reveals his dependence on Plato’s conception of the pathēmata of soulless bodies in the Timaeus. On some interpretations, Plato holds that the pathēmata are something midway between the qualities that inhere in a body and the awareness of them in a perceiving subject. Plotinus agrees, I think, and he behe uses to explain the radiation of light from bodies is supposed to explain primarily the transmission of colours from the sensible objects to the eyes: cf. . . . –, and the remarks in Emilsson, Sense-Perception, . Emilsson stresses that the analysis of colours is strictly connected to that of light in Plotinus, and he observes that colour for Plotinus must have or be some kind of light.  However, in contrast to Aristotle, in .  Plotinus argues that there is no need for a medium in vision. His criticism of Aristotle’s notion of medium occupies a large part of the treatise, but is not relevant for my argument here. For this criticism see Emilsson, Sense-Perception, –. Emilsson, however, does not point out that Plotinus’ account of colour transmission is influenced by DA . . Finally, it should be noted that Plotinus often presents sight as a paradigmatic sense, and colours as paradigmatic sensible qualities.  See Denis O’Brien’s suggestion that the παθήµατα are not ‘characters of the bodies themselves’, but ‘occupy a position midway between the features which are inherent in an object and which give rise to sensation and the actual sensible awareness which is registered in a human or an animal percipient’ (id., Theories of Weight in the Ancient World,  vols. (Leiden, ), ii. ). Cf. L. Brisson, ‘Plato’s Theory



Sara Magrin

lieves that he can explain this peculiar status of the pathēmata by using Aristotle’s view that the activity of the sensible is the same as that of the sense, and by suggesting, in contrast to Aristotle, that this activity is both in the sense and in the object. In any case, what we can infer from these observations is that Plotinus invites us to look to Democritus if we want to understand what a sensible quality is when it is considered only qua sensible, i.e. independently of its logos, but then suggests that we look to Aristotle to understand what a sensible quality is when its logos is taken into account. Let us examine DA .  more carefully, then. I assume that when Aristotle says that the activity of the sense is the activity of the sensible in the sense, what he means is that if you take a sensible quality, white, say, this very colour, not a power nor any more basic quality in the object, is the cause of your eyes seeing white. Whiteness for Aristotle is a quality that an object has by itself, and it is what causes any perceiver with the appropriate sense-organ to see that object as white and only as white (save for minor exceptions, cf. DA . , b–). It is because Aristotle thinks that objects have sensible qualities by themselves that in DA . , a– he criticizes the thesis, which he ascribes to the natural philosophers, that nothing is either black or white—and so on for any sensible quality—in the absence of sight—or, in general, of the appropriate sense. The natural philosophers in question are probably thinkers such as Protagoras and Democritus, and Aristotle’s response to them, as I understand it, is that they fail to distinguish between the apparent quality that constitutes the content of a particular act of sensation and the sensible quality that an object possesses. The latter is the cause of the quality that appears in sensation, but for Aristotle only the apparent quality cannot exist independently of its being perceived. Thus, white-lead, for instance, is white even if nobody is there to perceive it, in his view, but the white that fills your sight when you look at white-lead, i.e. the apparent white, is there only so long as you keep looking at it. of Sense Perception in the Timaeus: How it Works and What it Means’, Proceedings of the Boston Area Colloquium in Ancient Philosophy,  (), – at .  See esp. S. Broadie, ‘Aristotle’s Perceptual Realism’ [‘Perceptual Realism’], in J. Ellis (ed.), Spindel Conference : Ancient Minds (Southern Journal of Philosophy, , suppl.; Memphis, ), –; and H. Lorenz, ‘The Assimilation of Sense to Sense-Object in Aristotle’, Oxford Studies in Ancient Philosophy,  (), – at  n. .  Cf. Broadie, ‘Perceptual Realism’, –, for a detailed analysis of this point.

Sensation and Scepticism in Plotinus



To understand how Plotinus works through these remarks we need to recall that in his view the real locus of an object’s qualities is an intelligible power or logos. This means that, even if he grants that the activity of the sense and that of the sensible are the same, he implicitly rejects Aristotle’s thesis that the cause of seeing white is the sensible quality ‘whiteness’ in the object. In contrast to Aristotle, Plotinus maintains that the cause of seeing white is the power of white ‘in’ the intelligible structure of the object. As this power is only a disposition to appear white, the object with that disposition will tend to appear white to perceivers with suitable sense-organs, but will not necessarily appear in the same way to every perceiver or in every circumstance. This does not mean, however, that for Plotinus an object is white, or hot, and so on, only if it is perceived, as if whiteness were merely the content of a subjective experience. Plotinus, too, wants to answer the philosophers of nature, and Democritus in particular, and he does so by means of the double-activity theory. On the grounds of that theory he maintains, like Aristotle, that an object has whiteness independently of its being perceived, so that its being white is not exhausted in the apparent white that constitutes the content of a particular act of vision. But, unlike Aristotle, he identifies this whiteness with an intelligible logos, and thus with a disposition or dispositional power to appear white, rather than with an actual sensible whiteness that the object would display by itself independently of any perceiver and in every, or almost every, circumstance. Plotinus, then, an This is maintained also by Emilsson, Intellect, , , but Emilsson seems to think that this point makes no difference for Plotinus’ theory of sensation. He suggests, that is, that even though Plotinus locates the cause of sensible qualities in powers, none the less he can have a ‘common sense’ view about the presence of qualities in objects, where the ‘common sense’ view is that qualities are in the objects just as we perceive them. But I fail to see how this could be the case: it is one thing to say that I see white because the sensible whiteness of an object causes me to see white, and it is a different thing to say that I see white because of a power in the object. Both positions are compatible with a realist theory of sensation, but the latter is incompatible with the form of direct realism in the light of which Emilsson interprets Plotinus’ theory of sensation, namely one in which sensible qualities are ‘objective’ features of bodies that exist independently of any perceiving subject and are in the objects exactly as they are perceived.  Thus, I disagree with Emilsson when he claims that sensible qualities for Plotinus are ‘objective’ features of bodies. It is true that bodies display sensible qualities independently of any perceiver for Plotinus, but in his view we cannot know if they really preserve these qualities even when they are not perceived. Plotinus is not a ‘common sense’ realist, because he thinks that the locus of qualities is the intelligible structure of bodies, and because he believes that only some of the qualities that we



Sara Magrin

swers Democritus through the double-activity theory, which allows him to grant bodies some intrinsic differentiations without having to renounce the view that sensible qualities are to some extent subjective—in the sense of being relative to a perceiver, however, and not in the sense of being illusions. Relative though they are, he argues, the essential qualities of bodies are also in those bodies in which they are perceived to inhere in the form of a dispositional power that is active also in the absence of any perceiver. The problem, he goes on, is that, since a power may have different effects in different circumstances and on differently disposed perceivers, one cannot tell which of the qualities that a body manifests are intrinsic to it and which are not, and this is precisely why all our sensory representations must be deemed ‘obscure’ and ‘undecidable’. To avoid falling into a form of scepticism, however, in addition to providing an account of the intrinsic differentiations of bodies, Plotinus must also tell us how those differentiations can be known. He must tell us, that is, how we are to reach the real nature of things by overcoming the ‘obscure’ and ‘undecidable’ character of our sensory representations. Let us turn, then, to his theory of sensation. Although he endorses Arcesilaus’ criticism of cataleptic representations, Plotinus wants to maintain that sensory representations are to an extent cognitive. Sensation, he argues in . . , is a form of ‘cognition’ (γνῶσις). Thus, his views in this respect are not very different from those of some of his Platonist predecessors, such as Plutarch and Galen. Like them, Plotinus thinks that the Stoics have overestimated the cognitive power of sensation, but does not want perceive as inhering in a body (i.e. the essential qualities) are actually intrinsic to that body. In general, a direct realist, in contrast to Plotinus, thinks that the sensible structure of bodies is the locus of qualities, and maintains that all the qualities that we perceive as inhering in a body are intrinsic to that body, in the sense that they really belong to that body.  ‘This sensible substance is not being simpliciter, but this whole is something sensible; since we maintained that its seeming existence [τὴν δοκοῦσαν ὑπόστασιν] is a bundle [σύνοδον] of things relative to sensation [τῶν πρὸς αἴσθησιν], and the conviction that they exist comes from sensation [ἡ πίστις τοῦ εἶναι παρὰ τῆς αἰσθήσεως αὐτοῖς]’ (Enn. . . . –). It is true that Plotinus does not contrast the subjective affection an object produces on the sense-organs with the sensible object itself, but in my view this is not because he is not interested in sceptical issues, but because for him there is no sensible object that, qua sensible, could be such or such by itself. Plotinus turns the sceptical contrast between the affection caused by an object and the object itself into a contrast between sensible objects and intelligible natures, but the contrast remains; for more on this see below.

Sensation and Scepticism in Plotinus

 

to claim that sensations have no cognitive value at all. In sensation, he maintains, you actually do know a sensible thing because, even if this thing is grounded in the intelligible power of the logoi, there is nothing more to it qua sensible than the way in which it appears to you, since its sensible traits are external activities, and external activities are one and the same thing in the object and in the perceiving subject. Thus, say that something appears white to you, you can actually assert that that thing is white, and you can commit yourself to this claim, as there is no other sensible colour that the thing could have by itself, as opposed to the one it manifests to you. This sensory apprehension, however, is an ‘obscure’ apprehension for Plotinus. This, at least, is what I take him to suggest in . . , where he describes sensation as follows: Sensation is for the soul an ‘apprehension’ [ἀντίληψις], in which the soul grasps the quality that approaches bodies and is stamped with their [i.e. bodies’] forms [τὴν προσοῦσαν τοῖς σώµασι ποιότητα συνιείσης καὶ τὰ εἴδη αὐτῶν ἀποµαττοµένης]. (Enn. . . . –, trans. Armstrong, slightly modified)

The soul, Plotinus says here, apprehends a quality in sensation, and it is ‘stamped’ with the form of a body. The claim that the soul is ‘stamped’ with forms clearly echoes Tim.   –, where Plato says that the receptacle is ‘stamped’ with the copies of the forms. By echoing Tim.   –, Plotinus reveals that he implicitly relies on an analogy between soul and matter. In fact, he has introduced and developed this analogy in . . , where he observes the following: The reason why matter remains as it is is as follows: that which enters in  On Plutarch’s and Galen’s views on the cognitive value of sensory representations see Brittain, ‘Middle Platonists’.  Emilsson, Sense-Perception, , suggests that Plotinus here distinguishes two moments in sensation. At first, the soul would passively receive ‘qualities inherent in bodies’; then it would ‘mark itself’ with the forms of the bodies. The soul could mark itself with the form of a body for Emilsson only by fitting qualities into one or more of the forms it has latent within itself. Emilsson does not explain what these forms are, but he seems to suggest that they are things such as the form of the sky, say. I find this view that there are two moments in sensation problematic, for Plotinus nowhere actually says that receiving qualities and grasping the form of a body are two different things, and Emilsson is aware of this, of course (p. ). Furthermore, I do not think that for Plotinus we actually use forms such as that of the sky in sensation. This issue is taken up below.  This was remarked by H. J. Blumenthal, ‘Plotinus’ Adaptation of Aristotle’s Psychology’, in R. Baine Harris (ed.), The Significance of Neoplatonism (Norfolk, Va., ), – at .



Sara Magrin

no way gains anything from it—nor matter from what enters. In just such a way beliefs and appearances in the soul do not mix with it [ὡσπερ αἱ δόξαι καὶ αἱ ϕαντασίαι ἐν ψυχῇ οὐ κέκρανται], but depart from it, each retaining its own individuality intact, taking nothing, leaving nothing [οὐδὲν ἐϕέλκουσα οὐδὲ καταλείπουσα], in that it had not been commixed . . . Here [i.e. in the soul] the representation [ϕαντασία] is an image, but the soul does not have the nature of an image; although the representation seems to pull the soul in all directions at will, in fact it uses it none the less as if it were matter or something analogous; but since the representation was often driven out by the activities within the soul, it failed to obliterate the soul . . . For the soul has activities and logoi within itself that stand in opposition [to the representations] [ἔχει γὰρ ἐν αὐτῇ ἐνεργείας καὶ λόγους ἐναντίους], with which it repels any approach. (Enn. . . . –, trans. Fleet, modified)

Plotinus’ main point here is that the soul in sensation receives sensory phantasiai in a way analogous to that in which matter receives the ‘traces’. As matter displays sensible qualities without being affected by them, so the soul receives representations of sensible objects without being affected by these representations. Since Plotinus speaks of sensory representations that ‘pull the soul in all directions’, it is natural to infer that he has in mind the ‘obscure quasi-beliefs’ of . . , i.e. those beliefs that represent the turbulent motions that confuse the soul in Tim.   ff. Apparently, for Plotinus the soul gains control over these ‘obscure quasi-beliefs’ by using some logoi that it has within itself. These logoi stand for the forms in the soul, but I shall come back to this later. By comparing . . . – and . . . – we can conclude that the soul in sensation receives the representation of a sensible  The analogy is built with the help of some Aristotelian material, for Plotinus uses Aristotle’s description of νοῦς in DA . . Aristotle says that νοῦς has to be capable of receiving the form of an object while remaining ‘impassive’ (a), that it must be ‘unmixed’ (a), for the presence in it of what is alien to its nature would be a hindrance to the reception of further forms, that νοῦς is ‘the place of forms’ (a–). Inspired by these remarks, Plotinus maintains that matter is ‘impassive’ (ἀπαθής, . . .  and passim), that the λόγοι never mix with matter (οὐ κέκρανται; µὴ ἐµέµικτο, . . . –), that matter is a ‘place of forms’ (τόπος εἰδῶν, . . . ), that matter has no form of its own, because if it did, this form would prevent other forms from ‘entering’ it (. . . –). Then Plotinus applies these observations to the case of soul by means of an analogy between matter and soul. Notice also that Aristotle, at DA . , a–, says that νοῦς, like matter, is capable of ‘becoming all things’ [πάντα γίνεσθαι]. We have seen that Plotinus maintains that matter ‘becomes’ in the sense that, for any sensible quality, it appears to be that quality. It is reasonable to think that the soul, in his view, ‘becomes all things’ when sensory representations ‘enter’ it. Only matter, for Plotinus, has no power, whereas the soul has activities.

Sensation and Scepticism in Plotinus



thing. This representation is the way in which the form of a body becomes present in the soul. It is a ‘quasi-belief’ in which the sensible qualities of a body are confusedly apprehended. By ‘quasibelief’ Plotinus means a rudimentary, simple belief of the form ‘This is F’. This belief gives only a confused idea of its object because it is inaccurate, and it conflicts with other rudimentary beliefs about that object. But how do we arrive at having a ‘quasi-belief’ about a sensible thing? Plotinus claims that it is through a ‘judgement’. Time and again, in fact, he repeats that sensation is a ‘judgement’ (krisis) of the soul, and a judgement about sensible objects and their characteristics (. . . –; . . . –; . . . –; . . . –). By saying that sensation is a judgement, he does not mean that sensation is or involves an inferential process. What he wants to say is that sensation is a critical ‘activity’ (energeia) of the soul as opposed to a passive affection. Thus, for Plotinus, the soul ‘receives’ sensory representations, but this reception comes about through an act of judgement. Let us examine, then, how the soul judges sensible things. The most relevant passage is . . : Sensations must take place through bodily organs [δι᾿ ὀργάνων σωµατικῶν]. . . . The organ must be either the body as a whole or some member of it set apart for a particular work; an example of the first is touch, of the second, sight. And one can see how the artificial kind of organs [or tools] are intermediaries between those who judge and what they are judging [καὶ τὰ τεχνητὰ δὲ τῶν ὀργάνων ἴδοι τις ἂν µεταξὺ τῶν κρινόντων καὶ τῶν κρινοµένων γινόµενα], and inform the judge of the special characteristic of the objects under consideration [καὶ ἀπαγγέλλοντα τῷ κρίνοντι τὴν τῶν ὑποκειµένων ἰδιότητα]: for the ruler acts as link between the straightness in the soul and that in the wood [ὁ γὰρ κανὼν τῷ εὐθεῖ τῷ ἐν τῇ ψυχῇ καὶ τῷ ἐν τῷ ξύλῳ συναψάµενος]; it has its place between them and enables the craftsman to judge his work [ἐν τῷ µεταξὺ τεθεὶς τὸ κρίνειν τῷ τεχνίτῃ τὸ τεχνητὸν ἔδωκε]. (Enn. . . . –, trans. Armstrong, slightly modified)

Let us call the judgement that takes place in sensation ‘sensory judgement’. A sensory judgement is a judgement that the soul carries on through some bodily organ, which functions as a tool or, more precisely, ‘as a ruler’. As an artisan has a notion of straightness and uses a ruler to know whether something is straight or not,  That sensation does not involve inference is suggested at . . . –, and it is persuasively argued, in my view, in Emilsson, Sense-Perception, .



Sara Magrin

thus the soul has some notion of straightness and uses the senseorgans intentionally to verify whether a sensible object is straight or not. The ruler is the standard against which the straightness of the wood is tested by the artisan, and likewise, the sense-organs are the standards against which the soul tests the characteristics of a sensible thing. How exactly the sense-organ functions is a rather obscure issue. But notice that Plato, in the Timaeus (  ), speaks of at least one sense-organ, the tongue, as having ‘vessels’ that function as ‘testers’ (δοκίµια). Thus, even if the details remain obscure, what Plotinus here has in mind, probably, is the view that the soul judges that something is sweet, for instance, when an object causes the appropriate, physical reaction on the tongue. The tongue suffers some kind of affection, and the soul perceives something as sweet by means of that affection, much as we judge—to take the example of another tool used as ‘tester’—that something is hot when the mercury bar of a thermometer rises. However, the most problematic point in the passage above is the role ascribed to the notion of straightness. According to the ruler analogy, this notion must be in the soul before sensation takes place, but how are we to explain its presence there? Since Plotinus holds that all human beings have latent within their intellectual faculty all the forms that are in Nous (see . . ), he probably has these forms in mind when he speaks of a straightness in the soul. His point, I take it, is that we could not perceive things the way we do, unless we were somehow naturally predisposed towards them in certain ways by latent, a priori notions. Say that you perceive a portrait as beautiful, for instance: this capacity for Plotinus requires you to be  This point, I think, is not directly relevant for my purposes here. The standard account of the function of the sense-organ in Plotinus’ theory of sensation is Emilsson, Sense-Perception, . Emilsson maintains that the ensouled organ is not affected in sensation, but receives sensible qualities that become present in it according to a mode of existence which differs from the one they have in bodies. The organ, he suggests, is ‘phenomenally coloured, hot’, and so on, and the soul judges a sensible thing via such a ‘phenomenal experience’. The problem with this reading is, as I have said, that it rests on a distinction between reception of qualities and sensation that is not unambiguously present in the texts. Furthermore, it is not clear to me that this interpretation fits Plato’s remarks on the sense-organs in the Timaeus, which, as I am about to suggest, are the starting-point for Plotinus’ account of the role of the sense-organs in . . .  Plotinus never denies that the sense-organ is affected; in fact, in . . . – he claims that sensation is a judgement of the soul that ‘has to do with affections’ [περὶ παθήµατα], and he stresses that the judgement belongs to the soul, whereas the affections belong to what he calls the ‘qualified body’, i.e. the living body.

Sensation and Scepticism in Plotinus



naturally predisposed to see things as beautiful. Beauty is in fact one of just two examples of a priori notions that Plotinus provides in the Enneads; the second example is goodness. But beauty and goodness cannot be the only a priori notions with which our soul operates, I think. Consider again how Plotinus describes the function of the ‘ruler’ in the passage quoted above. He says that the ruler ‘acts as link between the straightness in the soul and that in the wood’ (ὁ γὰρ κανὼν τῷ εὐθεῖ τῷ ἐν τῇ ψυχῇ καὶ τῷ ἐν τῷ ξύλῳ συναψάµενος). This idea of a contact between soul and sensible things, I suggest, reflects Tim.   – , where Plato says that, since the soul is made of being, same, and different bonded together ‘proportionally’ (ἀνὰ λόγον), whenever ‘it comes into contact’ (ἐϕάπτηται) with something sensible or intelligible, it is stirred ‘through the whole of itself’ (διὰ πάσης ἑαυτῆς), and it declares what the thing is the same as or is different from. In . .  Plotinus suggests that what puts the soul in contact with sensible things is the sense-organ, which he defines as ‘a proportional mean’ (µέσον ἀνάλογον) just a few lines before introducing the ruler analogy (. . . –). We have already seen Plotinus engage with the Timaeus in . . ; for he certainly refers to that dialogue when he says that the soul is ‘stamped’ by the forms of bodies, and he probably works against its background when he claims that the sense-organs are rulers. If I am correct in suggesting that he appeals to the Timaeus also when he says that the sense-organs put the soul in contact with sensible things, it is difficult to resist the conclusion that throughout . .  Plotinus examines sensation by presupposing the conception of the soul of the Timaeus. Since Plato in the Timaeus says that the soul is made of being, same, and different, Plotinus could assume that being, same, and different, too, are among the basic a priori notions that predispose the soul in sensation. But we can go further, and say that he does count these notions among the a priori notions of the soul, because their presence in the soul is genuinely required to make his 

On goodness see . . . –, and on beauty see . . . –. . Cf. L. Brisson and J.-F. Pradeau (trans. and comm.), Plotin: Traités – (Paris, ),  n. . Brisson suggests that, in describing the sense-organ as ‘a proportional mean’, Plotinus has in mind Tim.  – . I think that he might actually have both passages in mind,  –  and   – . In any case, he describes the function of the sense-organs on the grounds of Plato’s remarks in the Timaeus.  For being, same, and different as some primary concepts or notions the soul would be made of in the Timaeus see D. Frede, ‘The Philosophical Economy of Plato’s Psychology: Rationality and Common Concepts in the Timaeus’, in M. Frede and G. Striker (eds.), Rationality in Greek Thought (Oxford, ), – at . 



Sara Magrin

theory of sensation work. Recall that for him sensations are rudimentary beliefs. How can the soul say that ‘This is F’ if it has no notion of being? One has to assume that, in predisposing us towards sensible things, some a priori notions become immediately available to our soul and allow it to formulate judgements of a simple type, such as ‘This is F’, ‘This is the same as that’, ‘This is different from that’, ‘This is beautiful’, and so on. Thus, for instance, say that you taste something sweet. The soul has no notion of ‘sweetness’ within itself that can be directly applied in sensation, nor does it have a notion of ‘sugar’. The soul does not have forms corresponding to each sensible quality or to each sensible object. You know that something is sweet in tasting it only because your tongue reacts in the appropriate ‘sweet-typical’ way, and you judge that the object that caused that reaction on your tongue is sweet. What your soul allows you to do in virtue of its notions is to detect some very general properties that things might have, such as beauty, for example, and to build the judgement ‘This is sweet’ by means of a copula in which the notion of being is expressed. With this analysis of sensation at hand, let us see how Plotinus overcomes the conflicting and ‘obscure’ nature of our sensory representations. It is ‘discursive reason’ (dianoia or to logizomenon), he says, that does the job. How reason proceeds is explained in . : Well, then, sensation sees a human being and gives its ‘imprint’ to the reasoning faculty [ἔδωκε τὸν τύπον τῇ διανοίᾳ]. What does reason say? It will not say anything yet, but only takes notice, and stops at that; unless perhaps it asks itself [πρὸς ἑαυτὴν διαλογίζοιτο] ‘Who is this?’ if it has met the person before, and says, using memory to help it, that it is Socrates. And if it unfolds the details of his form [ἐξελίττοι τὴν µορϕήν], it divides [µερίζει] what the phantasia gave it; and if it says whether he is good, its remark originates in what it knows through sensation [ἐξ ὧν µὲν ἔγνω διὰ τῆς αἰσθήσεως], but what it says about this it has already from itself, since it has a kanōn of the  Cf. Plato, Theaet.   ff. Since Plotinus thinks that sensations are beliefs, and thus that they have propositional content, his notion of sensory representation is remarkably similar to the Stoic one; for the Stoics claim that the sensory representations of an adult human are λογικαί (D.L. . ). However, they also maintain that for a representation to have propositional content there is no need of innate concepts in the soul (SVF ii.  = LS ), and this could lead one to think that Plotinus too has no need to appeal to innate concepts, such as being, in order to account for our ability to formulate sensory judgements. But I doubt this is the case. Rather, Plotinus views the Stoics as adversaries, and by positing an a priori notion of being in the soul, he implicitly suggests that the Stoics cannot ascribe propositional content to sensory representations unless they are ready to admit the presence of innate concepts in the soul.

Sensation and Scepticism in Plotinus



good in itself. How does it have the good in itself? Because it is like the good, and it is strengthened for the sensation of this kind of thing by Nous illuminating it: for this is the pure part of the soul and receives the traces [ἴχνη] of Nous coming down upon it. (Enn. . . . –, trans. Armstrong, modified)

This passage is clearly inspired by the account of false judgement in Theaetetus Part II, where Plato describes sensations as ‘imprints’ (tupoi) in the soul (  ). Plotinus sometimes speaks of the contents of sensation in terms of tupoi, as he does here, but since he always maintains that the soul can never be affected, we must assume that the term tupos for him has no physicalist connotation. Since he excludes a direct involvement of reason in sensation, when he says that reason considers the tupoi produced in sensation, he has to refer to something that happens after sensation has already taken place. The scenario is that of someone having an ‘obscure’ representation of a person with a snub nose whom he cannot identify. Reason takes notice of the sensory representation and tries to figure out who that person is. It does this by matching the sensory representation to a memory of that person, if it finds one. Through this match it concludes that ‘This is Socrates’. But reason may also want to figure out some details about Socrates, for example whether he is pale, and so on. Most of the time it can do this by examining the content of some sensory representations (a representation of Socrates, say, and of other pale people), but at times it finds that sensory representations are insufficient for judging certain things, and it appeals to some ‘rulers’ or ‘standards’ (kanones) that are made available to it not by sensation but by Nous. Suppose, for instance, that I want to find out whether Socrates is good; reason will start by considering past sensory representations of Socrates accomplishing this or that deed, but will soon find it impossible to judge Socrates’ goodness only on the basis of sensory representations and memories, for it is notoriously difficult to decide whether someone is good, as the relevant appearances always conflict (cf. Plato, Euthphr.   – ). But this problematic nature of goodness, Plotinus suggests,  Thus P.-M. Morel, ‘La sensation messagère de l’âme: Plotin, V,  [], ’, in M. Dixsaut, P.-M. Morel, and K. Tordo-Rombaut (eds.), La Connaissance de soi: études sur le traité  de Plotin (Paris, ), – at . Morel speaks of a τύπος ‘qui témoigne en lui-même de l’activité de l’âme et qui demeure incorporel’. Plotinus explains what he means by τύπος in . . . –, where he suggests that it is something analogous to the content of a thought. If sensations are beliefs, the τύπος, in sensation, is the obscure content of the belief.



Sara Magrin

turns out to have a positive function, for it acts as a ‘summoner’ (cf. Rep.  – ) that pushes reason to look further. Stimulated by the conflicting nature of some sensory appearances, reason reaches an a priori standard of goodness. Notice, however, that Plotinus explicitly says that the goodness found by reason is only a ‘trace’ of the form of goodness in Nous. This means that the a priori notions that reason uses as standards here are not the forms latent in our intellectual faculty, but only representations of those forms in the soul. Reason, then, before reaching the form of goodness, becomes acquainted with a representation of it in the soul. It is as if in trying to figure out whether Socrates is good, it had first to formulate some working hypothesis about goodness. But then how does reason reach the forms? To reactivate the forms, Plotinus suggests, reason needs to examine their ‘traces’ in the soul. This emerges from the following passage in . . : We could say at once that its [i.e. the soul’s] sensitive part is perceptive only of what is external . . . for it perceives the pathēmata in the body by itself, but the reasoning faculty in it makes its evaluation starting from the sensory representations present to it by combining and dividing them [τὸ δ᾿ ἐν αὐτῇ λογιζόµενον παρὰ τῶν ἐκ τῆς αἰσθήσεως ϕαντασµάτων παρακειµένων τὴν ἐπίκρισιν ποιούµενον καὶ συνάγον καὶ διαιροῦν]. And, as for the things which come to it from Nous, it [i.e. reason] observes what one might call their imprints [τύπους], and has the same power [i.e. of judging by combining and dividing] also in respect to them [περὶ τούτους]; and it continues to acquire understanding as if by recognizing the new and recently arrived imprints and fitting [ἐϕαρµόζον] them to those which have long been within it: this process is what we should call the ‘recollections’ of the soul. (Enn. . . . –, trans. Armstrong, slightly modified)

Reason (here to logizomenon) does not examine only sensory representations, it can also examine the representations of the forms. This is why, after formulating a working hypothesis on the nature of this or that thing, e.g. goodness, it can keep enquiring until it reaches its real form. This process starts from the discovery of an a priori representation, it ends with the uncovering of a form, and it is for Plotinus what Plato called ‘recollection’. Before reaching the forms, then, the soul must be capable of two 

Emilsson examines these representations of the forms in Sense-Perception, . For a different reading of this passage see L. P. Gerson, Plotinus (London, ), . Gerson denies that Plotinus posits representations of forms in the soul. The method of ‘combining and dividing’ to which Plotinus alludes seems to be dialectic, and it can be examined, I think, in the light of Remes’s insightful remarks 

Sensation and Scepticism in Plotinus



types of judgement, sensory judgements and what we might call ‘rational judgements’, i.e. judgements of reason. To form sensory judgements, the soul uses as standards or ‘rulers’ the sense-organs; to form rational judgements it uses as standards or ‘rulers’ some representations of the forms. This constant reference to ‘rulers’ (kanones) reveals that Plotinus, once again, works against a Democritean background. As standards, Plotinus’ ‘rulers’ are some kind of criteria of judgement, but they are not ‘criteria’ in the technical sense in which this term is used in the Hellenistic epistemological debates. Senses and representations of the forms in Plotinus are criteria in the ordinary and pre-Hellenistic sense of a means (a fallible means) of judgement. Now, we know that Democritus wrote a treatise called Canons, which we think was about criteria for judging the truth. It is unclear what counted as a kanōn for Democritus, but a report from Sextus seems particularly illuminating. Sextus (B  DK = S.E. M. . –) informs us that in his Canons (ἐν τοῖς Κανόσι) Democritus distinguished a ‘bastard cognition’ (σκοτίη γνώµη) from a ‘legitimate cognition’ (γνησίη γνώµη), and he says that Democritus identified the former with cognition through the senses and the latter with cognition through the ‘mind’ (dianoia). Sextus’ report, then, can be interpreted as suggesting that the kanones for Democritus were the mind and the senses. Even if we do not know whether Plotinus read Sextus, he appears to be familiar with the things that Sextus says about Democritus’ Canons. Plotinus takes Democritus’ kanones to be ‘reason’ (dianoia) and senses, but, while he maintains that the senses yield a ‘bastard cognition’—which he explains in terms of ‘obscure’ and ‘undecidable’ representations— he holds that such a cognition is possible only if sensible qualities are grounded in logoi. Furthermore, while he agrees that reason yields a ‘legitimate cognition’, he holds that to do so, it must rely on a priori standards, i.e. the forms. Now, no matter what Democritus might have meant by ‘bastard cognition’, undoubtedly this cognition for him was of less value than that obtained through the mind. Can the same be said of Plotinus? At first one is tempted in Plotinus on Self, –. Remes points out that the forms for Plotinus are complex entities composed of intelligible parts, and she examines the relevant notion of part in relation to the Sophist and to Socrates’ dream in the Theaetetus. 

See Lee, Epistemology, . One plausible way to interpret the distinction is to say, perhaps, that whereas the mind strives to grasp the true, imperceptible structure of things, the senses only 



Sara Magrin

to answer with an emphatic ‘Yes’: Plotinus is, after all, a Platonist, and thus he holds that real knowledge is to be found through the mind and not through the senses. But here the issue is not merely knowledge. If my remarks are correct, in . .  and  Plotinus is not only saying that reason yields a ‘legitimate cognition’ when it uncovers the real forms, and thus when it reaches knowledge, but suggests that reason’s cognition can be ‘legitimate’ even when it is not yet knowledge, because it still depends on representations of forms rather than on the forms themselves. What reason does when it uses the representations of the forms is presumably to form true beliefs. Hence, the real issue here is the following: does Plotinus think that rational true beliefs are cognitively superior to the rudimentary beliefs of which sensory representations consist? And if so, how exactly can reason formulate a true belief if all our sensory representations are ‘undecidable’? The first thing to do, I think, is to examine what is the object of a true belief. Consider again the passage quoted above from . . . Nowhere does Plotinus suggest that reason judges sensible objects. In . .  reason examines only and exclusively the contents of sensory representations. Reason’s judgements, then, are about sensory representations rather than about sensible things. This is confirmed by the following passage from . : And soul’s power of sensation need not be sensation of sensible things, but rather it must be apprehensive of the tupoi produced by sensation on the living being; these are already intelligible entities [νοητά]. So external sensation [τὴν αἴσθησιν τὴν ἔξω] is the image of this sensation, which is in its essence truer [ἀληθεστέραν τῇ οὐσίᾳ] and is a contemplation without affection of forms alone [εἰδῶν µόνων ἀ παθῶς εἶναι θεωρίαν]. From these forms, from which the soul alone [µόνη] receives its lordship over the living being, come reasonings, and beliefs and thoughts [διάνοιαι δὴ καὶ δόξαι καὶ νοήσεις]. (Enn. . . . –, trans. Armstrong, slightly modified)

Plotinus here distinguishes two kinds of sensation: one is an apprehension of intelligible things; the other is our ordinary sensation of sensible things. The first kind of sensation is ‘sensation’ only in an improper sense. As we read at the end of the passage, it is the power by means of which we reason, and we have ‘beliefs’, namely provide a starting-point for the enquiry (cf. Plut. Comm. not.  ). See Lee, Epistemology, .  I say that reason ‘can’ yield a ‘legitimate cognition’ because nothing prevents reason from making mistakes in Plotinus’ account of its operations in . .  and .

Sensation and Scepticism in Plotinus



proper δόξαι as opposed to the ‘quasi-beliefs’ we have in sensation. This ‘sensation’ differs from ordinary sensation for several reasons (it belongs to the soul alone—cf. Theaet.   —rather than to ‘the living being’, it uses forms, it uses inferences), but most importantly because in place of sensible things it has as objects the tupoi of sensation, i.e. the contents of our sensory representations. There is nothing particularly remarkable in saying that when we reason, we reflect on the contents of our sensory representations; what is noteworthy is that Plotinus says that this cognitive activity is ‘truer’ than ordinary sensation. This is a remarkable claim because if it is only by reflecting on sensory representations that you are brought closer to the truth, then it follows that if you want to find out whether Socrates is good, you should not pay any attention to the sensible Socrates, but examine the representations you have of him. From these remarks we can conclude that not only does Plotinus maintain that true beliefs are indeed cognitively superior to sensory beliefs or representations, he also maintains that they are cognitively superior because they are about representations of sensible objects rather than about the sensible objects themselves. How reason deals with sensory representations can be gathered from a passage in . .  that I have already quoted; I partially repeat it here for ease of reference: We could say at once that its [i.e. the soul’s] sensitive part is perceptive only of what is external . . . for it perceives the pathēmata in the body by itself, but the reasoning faculty in it makes its evaluation starting from the sensory representations present to it by combining and dividing them [τὸ δ᾿ ἐν αὐτῇ λογιζόµενον παρὰ τῶν ἐκ τῆς αἰσθήσεως ϕαντασµάτων παρακειµένων τὴν ἐπίκρισιν ποιούµενον καὶ συνάγον καὶ διαιροῦν]. (Enn. . . . –, trans. Armstrong, slightly modified)

Reason’s judgement, Plotinus says, is an ‘evaluation’ (epikrisis). This is significant because in the context of the sceptical debates from which Plotinus borrows the adjective anepikritos, a representation is anepikritos precisely because no epikrisis, i.e. no ‘evaluation’, of its truth is possible. Sensory representations conflict, a sceptic argues, and there is no non-question-begging criterion to solve their ‘disagreement’ (diaphōnia). But, apparently,  The reference to ‘thoughts’, I think, alludes to a further stage, namely, the stage at which, through reason, we reach Nous and we share in its thoughts.  This is a standard view in the literature on Plotinus, and the point is examined by Emilsson in Sense-Perception, –.



Sara Magrin

for Plotinus this disagreement can be solved by reason. Sextus, at PH . –, actually considers the possibility of appealing to ‘reason’ (dianoia) to sort out the conflict of sensory representations, but discards it. Reason, he assumes, could judge whether a sensory representation is true only by testing it directly against the sensible object it represents. But reason has no independent access to sensible things, he says, and therefore it cannot solve the disagreement of sensory reports. If we read Plotinus’ description of the evaluative task of reason against the background of these sceptical remarks, we can see that his goal is to answer sceptics such as Sextus. Reason can evaluate whether sensory representations are true, Plotinus maintains, and it can solve their disagreement, because this kind of disagreement is not to be solved by ‘reaching out’ to sensible things and comparing a sensory representation to the object itself. Sensory representations already convey all there is in a sensible object qua sensible, and their truth is to be evaluated not against anything sensible, but against an a priori standard that, precisely because it is a priori, is not involved in the disagreement of their reports. In other words, sensory representations are all ‘undecidable’ for Plotinus, but only to the extent that we cannot know which of them are true and which are false without a priori standards. When these standards are taken into account, sensory representations are merely ‘unevaluated’. ‘Undecidable’ in themselves, sensory representations can be evaluated when reason is ‘summoned’ and it gains access to the representations of the forms in the soul. If these observations are correct, Plotinus’ response to scepticism rests on the suggestion that truth, if anywhere, is to be found in the sensory representations themselves, rather than in any attempt to understand what a sensible object might be by itself. Qua sensible, an object is just whatever it appears to be to you, Plotinus holds, and there is nothing more to it. If you want to know what it really is, you have to grasp its nature, but since this is an intelligible item, to find it you will have to ‘look inside your mind’ and examine your own beliefs. Let me summarize, then, how we reach knowledge for Plotinus. We have a sensation, which is a judgement about a sensible thing and which produces an ‘obscure’ representation that consists in a rudimentary belief, such as, for example, ‘This is pink, with a snub nose, and so on’. This belief, as the sceptics (loosely speaking) point 

For ἀνεπίκριτος as meaning both ‘undecidable’ and ‘unevaluated’ see LSJ s.v.

Sensation and Scepticism in Plotinus



out, does not allow us to identify the object that we perceive as this or that thing, and it always conflicts with other beliefs we have about the same object. But this is not a worrisome situation for Plotinus, for at this point reason can intervene, in his view. Plotinus concedes to the sceptics that reason cannot have first-hand knowledge of sensible things, but suggests that this is not a problem because it can proceed like a judge in a law court (cf. Theaet.  – ). The judge assesses a case by listening to witnesses and applying the rule of law; analogously, reason can evaluate sensory reports and judge them by applying some a priori standards. Often, for the needs of daily life, reason does not even need to appeal to these standards, because it can evaluate sensory reports simply by matching them to memories. But sometimes the content of a report is particularly controversial, and it cannot be dealt with that quickly. In these difficult cases the content of a sensory representation acts as a ‘summoner’ that leads to a ‘higher’ truth. Say that reason needs to assess whether Socrates is beautiful. Naturally predisposed in certain ways by the presence of the form of beauty in the soul, it formulates a definition, something like ‘Beauty is having proportionate parts’. If this definition is false, it leads nowhere. But if it is true, it is a representation of the form of beauty in Nous, and reason can use it to solve the conflict of the relevant sensory representations of Socrates. However, this successful definition is bound to cause as many problems as it solves, for is beauty really nothing more than having proportionate parts? Does this hold for intelligible things as it seems to hold for sensible ones? Reason understands that it needs to look for the nature of beauty itself to answer these questions, and this enquiry leads it to discover a form, which alone can be the ultimate standard of truth. Throughout this investigation, which Plotinus calls ‘recollection’, reason proceeds by using definitions as hypotheses (recall the role of hypotheses in . . ), and each step of the process reflects one of the segments of the line of Republic . The lowest segment of the line, ‘imagination’ (eikasia), corresponds to sensory representations; the second segment from the bottom, ‘belief’ (pistis), corresponds to the proper beliefs that reason formulates about sensory reports; the third segment, ‘rea That reason proceeds by formulating definitions can be gathered from Plotinus’ observation (inspired by Arist. Post. An. . –) that the ‘what it is’ [τί ἐστιν] is the starting-point for every enquiry; see . . . –. Cf. the remarks on forms as definitions in C. D’Ancona, ‘Ἄµορϕον καὶ ἀνείδεον: causalité des formes et causalité de l’Un chez Plotin’, Revue de philosophie ancienne,  (), – at .



Sara Magrin

soning’ (dianoia), corresponds to reason’s examination of sensory reports and to reason’s enquiry into the nature of the a priori notions used to evaluate those reports. The last segment, nous, is evoked by reason’s ability to reactivate the forms in the intellectual faculty of the soul, which for Plotinus is a potential Nous. I conclude that Plotinus is directly and not only polemically engaged with the ancient sceptical tradition, and that he develops his epistemology as an answer to scepticism, as both Wallis and O’Meara have suggested. He uses different strands of scepticism, but ascribes a particular importance to a sceptical reading of Democritus which has its roots in Aristotle and in the sceptical circles of the late fourth century . This prominence of Democritus is to be explained in the light of the influence that, in Plotinus’ view, Democritus’ ontology and epistemology had on Plato’s own ontology and epistemology. It has been argued that no ancient philosopher before Augustine ever thought that truth could be predicated of our mental states considered as objects of enquiry in their own right. But Plotinus, if I am right, does think that truth can be predicated of our mental states, and, like Augustine, he builds on this principle his answer to the sceptics. Plotinus’ response to scepticism, as O’Meara has argued, is the same as that of Augustine and Descartes in the end. As for Augustine and Descartes, scepticism for Plotinus is not simply an obstacle that has to be overcome, but provides the grounds for knowledge. It is because sensory reports are ‘undecidable’ that we are led to ‘turn inside’ and to look for some a priori notions in the mind; and it is because even these notions turn out to be ‘obscure’ in their own way that we have to proceed further until we reach Nous, in which every doubt is answered. Université du Québec à Montréal

BIBLIOGRAPHY Armstrong, A. H. (ed. and trans.), Plotinus: The Enneads,  vols. (Cambridge, Mass., –). Baltes, M., Die Weltentstehung des platonischen Timaios nach den antiken Interpreten,  vols. (Leiden, –).  This is one of the main theses of Myles Burnyeat’s very influential article ‘Idealism and Greek Philosophy: What Descartes Saw and Berkeley Missed’, Philosophical Review,  (), –.

Sensation and Scepticism in Plotinus



Bett, R., Pyrrho, his Antecedents, and his Legacy (Oxford, ). Blumenthal, H. J., ‘Plotinus’ Adaptation of Aristotle’s Psychology’, in R. Baine Harris (ed.), The Significance of Neoplatonism (Norfolk, Va., ), –. Bréhier, É. (ed. and trans.), Plotin: Ennéades,  vols. (Paris, –). Brisson, L., ‘Logos et logoi chez Plotin: leur nature et leur rôle’, Cahiers philosophiques de Strasbourg,  (), –. ‘Plato’s Theory of Sense Perception in the Timaeus: How it Works and What it Means’, Proceedings of the Boston Area Colloquium in Ancient Philosophy,  (), –. and Pradeau, J.-F. (trans. and comm.), Plotin: Traités – (Paris, ). Brittain, C., ‘Middle Platonists on Academic Scepticism’ [‘Middle Platonists’], in R. W. Sharples and R. Sorabji (eds.), Greek and Roman Philosophy  – ,  vols. (London, ), ii. –. Broadie, S., ‘Aristotle’s Perceptual Realism’ [‘Perceptual Realism’], in J. Ellis (ed.), Spindel Conference : Ancient Minds (Southern Journal of Philosophy, , suppl.; Memphis, ), –. Burnyeat, M. F., ‘Idealism and Greek Philosophy: What Descartes Saw and Berkeley Missed’, Philosophical Review,  (), –. Crystal, I., ‘Plotinus on the Structure of Self-Intellection’, Phronesis,  (), –. D’Ancona, C., ‘Ἄµορϕον καὶ ἀνείδεον: causalité des formes et causalité de l’Un chez Plotin’, Revue de philosophie ancienne,  (), –. ‘Le rapport modèle-image dans la pensée de Plotin’, in D. De Smet, M. Sebti, and G. De Callataÿ (eds.), Miroir et savoir: la transmission d’un thème platonicien, des Alexandrins à la philosophie arabo-musulmane (Leuven, ), –. Dixsaut, M., Morel, P.-M., and Tordo-Rombaut, K. (eds.), La Connaissance de soi: études sur le traité  de Plotin (Paris, ). Emilsson, E. K., Plotinus on Intellect [Intellect] (Oxford, ). ‘Plotinus on Sense Perception’, in S. Knuuttila and P. Kärkkäinen (eds.), Theories of Perception in Medieval and Early Modern Philosophy (Dordrecht, ), –. Plotinus on Sense-Perception: A Philosophical Study [Sense-Perception] (Cambridge, ). ‘Plotinus on the Objects of Thought’, Archiv für Geschichte der Philosophie,  (), –. Fleet, B. (trans. and comm.), Plotinus: Ennead III. . On the Impassivity of the Bodiless [Ennead III. ] (Oxford, ). Frede, D., ‘The Philosophical Economy of Plato’s Psychology: Rationality and Common Concepts in the Timaeus’, in M. Frede and G. Striker (eds.), Rationality in Greek Thought (Oxford, ), –.



Sara Magrin

Frede, M., ‘Review of Greek Skepticism’, Journal of Philosophy,  (), –. ‘Stoics and Skeptics on Clear and Distinct Impressions’ [‘Stoics and Skeptics’], in id., Essays in Ancient Philosophy (Minneapolis, ), –. Ganson, T. S., ‘The Platonic Approach to Sense-Perception’, History of Philosophy Quarterly,  (), –. Gelzer, T., ‘Plotins Interesse an den Vorsokratikern’, Museum Helveticum,  (), –. Gerson, L. P., Plotinus (London, ). (ed.), The Cambridge Companion to Plotinus (Cambridge, ). Graeser, A., Plotinus and the Stoics: A Preliminary Study (Leiden, ). Hankinson, R. J., ‘Natural Criteria and the Transparency of Judgement: Antiochus, Philo and Galen on Epistemological Justification’, in B. Inwood and J. Mansfeld (eds.), Assent and Argument: Studies in Cicero’s Academic Books (Leiden, ), –. Henry, P., and Schwyzer, H.-R. (eds.), Plotini opera,  vols. (Oxford, ). Kalligas, P., ‘Logos and the Sensible Object in Plotinus’, Ancient Philosophy,  (), –. (ed.), Πλωτίνου ᾿ Εννεὰς Τρίτη (Athens, ). King, R. A. H., Aristotle and Plotinus on Memory (Berlin, ). Kühn, W., Quel savoir après le scepticisme? Plotin et ses prédecesseurs sur la connaissance de soi (Paris, ). Lee, M.-K., Epistemology after Protagoras [Epistemology] (Oxford, ). Lloyd, A. C., ‘Plotinus on the Genesis of Thought and Existence’, Oxford Studies in Ancient Philosophy,  (), –. The Anatomy of Neoplatonism [Anatomy] (Oxford, ). Long, A. A., and Sedley, D. N., The Hellenistic Philosophers [LS],  vols. (Cambridge, ). Lorenz, H., ‘The Assimilation of Sense to Sense-Object in Aristotle’, Oxford Studies in Ancient Philosophy,  (), –. Menn, S., Descartes and Augustine (Cambridge, ). Plato on God as Nous (Carbondale, ). ‘The Stoic Theory of Categories’, Oxford Studies in Ancient Philosophy,  (), –. Mignucci, M., ‘The Stoic Notion of Relatives’, in J. Barnes and M. Mignucci (eds.), Matter and Metaphysics (Naples, ), –. Morel, P.-M., Démocrite et la recherche des causes (Paris, ). ‘La sensation messagère de l’âme: Plotin, V,  [], ’, in M. Dixsaut, P.-M. Morel, and K. Tordo-Rombaut (eds.), La Connaissance de soi: études sur le traité  de Plotin (Paris, ), –. O’Brien, D., Theories of Weight in the Ancient World,  vols. (Leiden, ).

Sensation and Scepticism in Plotinus



O’Meara, D. J., ‘Scepticism and Ineffability in Plotinus’, Phronesis,  (), –. Rappe, S., ‘Self-Knowledge and Subjectivity in the Enneads’, in L. P. Gerson (ed.), The Cambridge Companion to Plotinus (Cambridge, ), –. Remes, P., Plotinus on Self: The Philosophy of the ‘We’ (Cambridge, ). Rutten, C., ‘La doctrine de deux actes dans la philosophie de Plotin’, Revue philosophique,  (), –. Sedley, D. N., Creationism and its Critics in Antiquity (Berkeley, ). ‘Stoic Physics and Metaphysics’, in K. Algra, J. Barnes, J. Mansfeld, and M. Schofield (eds.), The Cambridge History of Hellenistic Philosophy (Cambridge, ), –. ‘The Stoic Criterion of Identity’, Phronesis,  (), –. Stamatellos, G., Plotinus and the Presocratics (Albany, NY, ). Stough, C. L., Greek Skepticism: A Study in Epistemology (Berkeley and Los Angeles, ). Striker, G., ‘The Ten Tropes of Aenesidemus’ [‘Ten Tropes’], in ead., Essays on Hellenistic Epistemology and Ethics (Cambridge, ), –. Tsouna, V., The Epistemology of the Cyrenaic School (Cambridge, ). Wallis, R. T., ‘Scepticism and Neoplatonism’, in ANRW ii. .  (), –.

This page intentionally left blank

AI MING AN D D E T E R M I N I N G A Discussion of Iakovos Vasiliou, Aiming at Virtue in Plato

C. C . W . T A Y L O R

T challenging and original book presents a new interpretation of the ethics of the Socratic dialogues, especially Apology, Crito, and Euthyphro, and of the Gorgias and Republic. The key to this interpretation is a distinction between two types of ethical principle, aiming principles, which specify aims to be pursued, and determining principles, which specify what counts as doing the kind of action which one’s aiming principles prescribe. Specifically, Socrates is represented in all these works as committed to (indeed as knowing) a single aiming principle, that of the supremacy of virtue (SV), which ‘says that doing the virtuous action trumps any other aim one may have in acting’ (). SV functions sometimes as an explicit aim, sometimes (perhaps more often) as a ‘limiting condition’ restricting one’s pursuit of other aims, so that ‘[w]hen acting for some other end than virtue (for example, pleasure or financial gain), SV requires that the agent nevertheless not act in a way that is contrary to virtue. The role of SV as a limiting condition is expressed in Socrates’ well-known statement that “it is never right to do wrong”’ (ibid.). Adherence to SV leaves open determining questions about what counts as acting as SV requires; while Socrates claims (according to Vasiliou) to know SV, he does not claim to know (indeed disclaims knowledge of) the answers to determining questions. Several dialogues are devoted to determining questions; in the Crito, while Socrates and Crito are agreed in their commitment to SV, they disagree on whether Socrates’ trying to escape from prison would count as acting rightly, as Crito urges, or as act© C. C. W. Taylor   Iakovos Vasiliou, Aiming at Virtue in Plato (Cambridge: Cambridge University Press, ), pp. x+.



C. C. W. Taylor

ing wrongly, as Socrates and the Laws insist. Similarly, while Socrates and Euthyphro both accept SV in its application to the actiontype of piety, they are unclear about what kind of action counts as acting piously, and it is that determining question which is the topic of the dialogue. The Gorgias, by contrast, is concerned with SV itself, since Socrates’ interlocutors, especially Polus and Callicles, reject the supremacy of virtue in favour of the supremacy of other goods such as wealth and desire-satisfaction, while Socrates argues against them that the supreme aim should be the good of the soul, i.e. virtue, which is irreducible to the goods which they espouse. Thus far the interpretation is in my view fairly uncontroversial, though one might quibble about some details (for quibbles, see below). Much more controversial is its application to the Republic, which takes up four of the eight chapters. Standardly, the account of justice in the soul is taken to be an account of what justice is, an ‘agent-centred’ as opposed to an ‘act-centred’ account. This is taken as involving a conceptual shift from a conception of the property of justice as the disposition to act in certain ways, e.g. to accord others their due, to a conception of the property as consisting in a state of psychic harmony, from which kinds of actions such as according others their due are supposed to issue. Against this orthodoxy Vasiliou argues that the state of psychic harmony is not presented as what justice is; rather, it is not the essence, but an essential property of justice (an Aristotelian ‘necessary coincident’, ) that it promotes psychic harmony, i.e. the health of the soul. That is not to say what justice is, but to show why we should value justice above all: that is, it is an argument in favour of SV. The ‘determining’ question what justice is is left open, and indeed is not answered in the Republic at all, since it can be answered only by someone who knows the Form of Justice, which the readers of the dialogue do not. What the Republic tells us is that we should act justly because doing so will perfect our souls, but in answer to the determining question ‘What counts as acting justly?’ all that it tells us is that acting justly is acting in ways that relate appropriately to the Form of Justice, whatever those ways may be. I have no quibbles about Socrates’ commitment to SV; it is expressly attested by Crito  – and Ap.  , and I know of no text which qualifies that commitment. But Vasiliou claims repeatedly that Socrates knows SV, and at least once () that SV is the only item of moral knowledge he has. I am less clear that the texts estab-

Aiming and Determining



lish these claims. The key passage is Ap.   –, τὸ δὲ ἀδικεῖν καὶ ἀπειθεῖν τῷ βελτίονι καὶ θεῷ καὶ ἀνθρώπῳ, ὅτι κακὸν καὶ αἰσχρόν ἐστιν οἶδα. This may be read as ‘I know that it is bad and shameful to do wrong, and (consequently) that it is bad and shameful to disobey my superior, whether god or man.’ On that reading Socrates says that he knows that it is bad and shameful to do wrong: that is, he makes a general claim about the badness of wrongdoing. But even that falls short of SV’s claim that acting virtuously trumps any other aims; from the fact that it is bad and shameful to do wrong it does not follow that one should never do wrong in any circumstances, since there might be circumstances in which doing something else was worse and more shameful than doing wrong. Further, the Greek allows another reading, in which Socrates’ claim is not about the badness of wrongdoing in general, but about the badness of the specific wrongdoing of disobeying one’s superior; on this reading the Greek means ‘I know that it is bad to wrong one’s superior, whether god or man, by disobeying’, or ‘I know that it is bad to do wrong by disobeying one’s superior’. It is clear that in this passage Socrates claims some moral knowledge, but it is not clear that that knowledge is knowledge of SV. The same ground for unclarity affects a later passage,  –, which Vasiliou cites () as another case where Socrates claims knowledge of SV. In this passage Socrates says that he would be wronging himself by proposing another penalty as an alternative to the death penalty which Meletus has demanded, since he does not know whether death is good or bad, but he does know that the alternatives such as imprisonment or exile would be bad. Vasiliou takes it that he knows that they would be bad in so far as they would be instances of Socrates’ wronging himself, and so violations of SV. That may indeed be right, but it is equally possible that Socrates knows that they would be bad in so far as they would involve his abandoning his mission to philosophize, and thus disobeying his superior. In both these passages, it seems to me, the texts leave it unclear whether what Socrates is claiming knowledge of is SV as Vasiliou conceives it, or some more specific moral truth, not indeed unconnected with that principle, but not identical with it. A similar doubt arises about a well-known passage of the Gorgias ( – ), where what is at issue is not, indeed, knowledge, but assurance. In this passage Socrates says that the conclusion of the previous arguments, viz. that it is worse and more shameful to do wrong than to be wronged, has been ‘held firm and bound down by



C. C. W. Taylor

arguments of iron and adamant’, which Vasiliou takes as expressing Socrates’ ‘total, absolute confidence in SV’ (). But the claim that it is worse and more shameful to do wrong than to be wronged is a weaker claim than SV; the latter specifies an aim which should take precedence over all others, the former merely identifies the worse of two alternatives. The weaker claim follows from SV, but is not identical with it. The first of the four chapters on the Republic is devoted to book , chiefly the discussion with Thrasymachus. Though the distinction between aiming and determining principles does not disappear, it is less prominent in this section, which deals mainly with topics which have become standard in the literature, such as the kind of definition which is employed, the sense of Cleitophon’s intervention, and, centrally, the identification of Thrasymachus’ position. Vasiliou has interesting things to say on all of these issues. Against a recent suggestion that in asserting that justice is the advantage of the stronger/ ruler Thrasymachus is not offering a definition, but merely identifying a property which typically holds of justice, Vasiliou argues, rightly in my opinion, that Thrasymachus is at least approximating to a Socratic definition in specifying something which he thinks common to all cases of justice, but that in so doing he does not make a clear distinction between a specification of what justice is and a property which necessarily holds of it. Thrasymachus takes it for granted that ‘[F]ollowing the rule of the established government, which he takes to be the stronger, is what justice is’ (). What his slogan expresses is a ‘sophisticated sociological analysis [which] offers us insight into a feature of all actions proclaimed by the rulers to be just in society: they are to the rulers’ advantage’ (ibid.). Thrasymachus, then, starts out as a conventionalist about justice. When Socrates points out the inconsistency of asserting both that justice is conformity to law and that it is the advantage of the ruler, since rulers sometimes promulgate laws which are not to their advantage, Cleitophon, according to Vasiliou, comes to Thrasymachus’ aid by suggesting that what Thrasymachus has actually said was that the advantage of the superior is what the superior believes is to his advantage. If he accepts this suggestion Thrasymachus can restore consistency to his position by combining conventionalism about justice with subjectivism about advantage. Thrasymachus rejects this suggestion because he is firmly committed to the élitist view that the superiority of rulers consists in their

Aiming and Determining



correct grasp of what is advantageous to them, which assumes objectivism about advantage. Since I think that subjectivism about one’s own advantage is an extremely silly position (as Socrates implies at Theaet.   – ), I prefer a reading which does not saddle Cleitophon with it. We have it if we take Cleitophon’s words τὸ τοῦ κρείττονος συµϕέρον ἔλεγεν ὃ ἥγοιτο ὁ κρείττων αὑτῷ συµϕέρειν (  –) not in the sense ‘he said that the advantage of the superior is what the superior thinks to be advantageous to him’ (Vasiliou’s reading), but as ‘by “the advantage of the superior” he meant [ἔλεγεν] “what the superior thinks to be advantageous to him”’ (to which Polemarchus responds ‘But that’s not what he said’ (ἀλλ᾿ οὐχ οὕτως . . . ἐλέγετο). On this reading Cleitophon is not saddling Thrasymachus with a subjectivist account of advantage, but suggesting that Thrasymachus had not formulated his view correctly; whereas what he said was ‘justice is the advantage of the superior’ (that is, rulers lay down laws which actually promote their advantage), what he really meant, and therefore should have said, was ‘justice is what the superior thinks advantageous to him’ (that is, rulers lay down laws with the aim, not necessarily achieved, of promoting their own advantage). Polemarchus’ objection to this eminently sensible (and charitable) suggestion amounts to the insistence that Thrasymachus should be held to the letter of what he actually said, instead of being allowed to reformulate it. Conventionalism about justice is not, in Vasiliou’s view, the whole of Thrasymachus’s position. He makes the interesting suggestion () that Thrasymachus’ praise of injustice in his ‘Great Speech’ shows, not that he has shifted from his original conventionalism to some form of realism (as suggested by various commentators) but that, while he is a conventionalist about justice, he is a realist about injustice. ‘[F]or Thrasymachus injustice is a real, objective feature of the world independent of human agreements, conventions or laws, but . . . justice is not. . . . For Thrasymachus injustice is a matter of exercising the natural impulse to pleonexia . . . Justice, by contrast, involves some human intervention to establish laws’ (, author’s italics). On this view, while there is real injustice in a state of nature, there is no justice: for example, lions really commit injustice against the antelopes they prey on (the inversion of Callicles’ claim at Gorg.   , that the behaviour of beasts of prey is a paradigm of natural justice), but there is no justice in the state of nature, since there are no laws or agree-



C. C. W. Taylor

ments. Vasiliou finds the same view in the myth of the Protagoras: before Zeus conferred shame and justice on mankind (i.e. before the development of political institutions, including law) humans threatened the survival of humanity by acting unjustly towards one another ( ). I am not convinced that these texts require (though they certainly allow) the attribution to Thrasymachus, or to Protagoras, of the complex position that whereas justice exists only by nomos, injustice exists by phusis. The phenomena can be equally accommodated by the more conventional view that, while there is strictly speaking neither justice nor injustice in the state of nature, since both presuppose a framework of agreements and other institutions regulating human interaction, humans in a state of nature have tendencies to aggression and self-aggrandizement which require to be kept in check by the development of those institutions. Once those institutions exist, they generate a vocabulary which allows those natural tendencies to be characterized in specifically moral and legal terms. Thus, rather than saying that injustice is something which exists by nature we can better say that what exists by nature is aggression, self-seeking, etc., and that specific forms of aggression and self-seeking are appropriately characterized by our moral vocabulary as violations of rights, as dishonesty, or whatever. As said above, the most controversial thesis in the discussion of the Republic, and indeed in the entire book, is the claim that what is standardly reckoned as a revisionary account of justice, namely psychic harmony, is not an account of what justice is, but an account of why we should accept SV. Justice is the property that all just acts have in common, in virtue of which they are just, but what that property is we are not told, except schematically, viz. that all just acts are just in virtue of instantiating the Form of Justice. What instantiating the Form of Justice consists in can be known only to the philosopher who knows the Form of the Good, and thereby the Form of Justice. All just acts indeed have this property in common, that it is by doing them that one comes to possess psychic harmony, but that is not what makes them just; it is instead a necessary concomitant of their being just (see above). It is not the case that acts are just in virtue of being such as to promote psychic harmony; rather it is by doing acts which are, independently, just that individuals achieve psychic harmony. And it is because doing just acts is the

Aiming and Determining



way to psychic harmony that we should value doing just acts above all, i.e. that we should be committed to SV. I am sufficiently wedded to the current orthodoxy to find this ingenious suggestion, well-argued though it certainly is in Vasiliou’s presentation (the details of which have, for reasons of space, to be passed over), ultimately unconvincing. At the end of book  Socrates says that in the preceding discussion they went wrong in trying to determine the value of justice before they determined what it is, and he, Glaucon, and Adeimantus agree to remedy that defect by first investigating what justice is and then what the good of it is. To investigate what it is they will first identify it in the larger letters of the polis and then in the smaller letters of the individual: that is, they will identify it first as a property of a polis and then as a property of an individual. The justification of this procedure is that it must be the same nature (eidos,   ) which is instantiated in the larger and in the smaller entity. Justice in the polis proves to be that state in which each of the three classes makes its own specific contribution to the good of the whole—that is, social justice is social harmony. But that is merely an image of true justice, which is concerned not with one’s external behaviour or activity, but with one’s internal (sc. behaviour or activity) ( ); true justice is that state of the individual soul in which each of the parts of the soul makes its appropriate contribution to the good of the whole person: that is, justice as an attribute of the individual is psychic harmony. Justice is then treated primarily as a property of individual persons, and secondarily as a property of social organizations, though the nature of justice is the same in both. Moreover, this passage continues by telling us what property of actions justice is; it is the property of preserving and promoting justice as a property of the individual, i.e. psychic harmony (  –). This, it seems to me, is precisely what the analogy of justice with health (  – ) would lead us to expect: health is promoted and preserved by healthy diet, a healthy lifestyle etc., precisely because what it is for diet or lifestyle to be healthy is just for them to be such as to promote and preserve health. It would be quite bizarre to suggest that healthiness in diet etc. might be some independent property P, such that the health of the person happens to be preserved and promoted by whatever possesses P. Vasiliou argues that because the education of the Guardians requires that they come to have harmonious souls by performing virtuous actions, those actions must be independently



C. C. W. Taylor

virtuous in order that they have the requisite causal power. But that is as fallacious as it would be to argue that in order to become healthy by eating healthy food, that food must be independently healthy in order to have its causal power. He objects to the view that justice is psychic harmony because on that view ‘it becomes a mystery how anyone could determine whether an action is just or not. A person would need to see whether it contributes to psychic harmony. But how would one do this?’ (, author’s italics). But how does one see whether a certain kind of food is healthy? By experience of its effects. Similarly, it is a matter of experience that certain kinds of action tend to promote a well-integrated personality and others to hinder its development. Of course it must be granted that by Platonic standards the accounts of justice and of the other virtues in the Republic fall short of providing the reader with knowledge, since they are not grounded in knowledge of the Forms. All that is said is at the level of belief; knowledge would require that we take the ‘longer way’. But at that level, it seems to me that the texts make it tolerably clear that Socrates is presenting his belief as to what the property of justice is. I believe that that account is revisionary, as the orthodox interpretation maintains and Vasiliou denies. It is quite another question whether it is true. I am conscious that in thus concentrating on certain points of disagreement I have failed to do justice to the generally high quality of this book. The level of detailed knowledge of the texts and the care and clarity with which Vasiliou presents his case is most impressive. I should like to indulge myself by concluding with a tiny, incidental point, which relates to a personal hobby-horse of mine. Concerning several dialogues which begin from practical issues, including the Protagoras, Vasiliou says that we never find out how these practical issues are resolved, giving the question ‘Does Hippocrates enroll with Protagoras?’ as an instance of such an unresolved question (). The final word of the dialogue is ἀπῇµεν, ‘we came away’, where ‘we’ refers to Socrates and Hippocrates. I submit that this strongly suggests that Plato intended a negative answer to Vasiliou’s question. Corpus Christi College, Oxford

INDE X L O C O R U M Aetius . . :  n.  Alexander of Aphrodisias De anima, ed. Bruns :  Aquinas In Aristotelis libros De anima expositio, ed. Pirotta :  n.  Aristotle Categories a–: – n.  a ff.:  n.  De anima a–:  n.  b–:  b–:  a–:  n.  b–:  n.  b–:  n.  a–:  n.  b–: ,  b–:  b–:  n.  b–:  b–:  n. ,  a:  n.  a ff.:  n.  a–:  n.  a–:  n.  a–b:  a–:  a–:  a:  a–:  n. ,  a–:  a–b:  b–:  n.  b–:  n. ,  b:  b–a:  n.  b–:  b–: 

b–:  b–a:  n.  b:  n.  b:  n.  b–a: – b–:  n.  b ff.:  b–:  b–:  b–:  b:  b–:  b–: – n. , ,  b:  b:  n.  b:  n. . ,  b–a:  b:  n. ,  b–a:  a:  n.  a–:  n.  a–:  n.  a–:  n.  a–b:  n.  a:  b:  n.  b:  b–a:  b–a:  a–:  n. ,  a:  a–:  a–:  a–:  a–:  a–:  b–:  n.  b–:  n.  b–:  b–:  b–:  n.  a: – n.  a–b:  n.  a–:  a:  b: 

 b

 –: ,  b–:  n.  b:  b:  b:  a:  n.  b–:  b–:  a–:  a–:  a–:  a–:  a ff.:  b–:  a–:  a–:  a:  a–:  n.  a:  n.  a–:  a:  n.  a–:  n.  b–:  a–:  n.  a–:  n.  a–:  n.  b–:  a–:  a–b:  a–:  a ff.:  a–: ,  n.  a–b:  a–b:  n.  b–:  b:  b–:  n.  a–:  n.  De generatione animalium b–:  n.  a–:  n. ,  n.  a–:  n.  b–:  n.  b–:  n.  a–:  n.  a–:  n.  a–:  a–:  n.  b–:  n.  b–:  n.  b–:  n.  a–:  a–: 

Index Locorum b–: – n.  b–:  n.  b–:  n.  a–:  n.  b–:  n.  a–:  n.  a–:  n.  b ff.:  n.  b–a:  n.  b–:  n.  b–:  n.  b–:  n.  a–:  n.  a–:  a:  b–a:  n.  b–:  a–:  a–:  a–:  a–:  a–:  a–:  a:  a–b:  n.  b–:  n.  b–:  n.  b–:  n.  a–b:  a–b:  n.  b ff.:  b–:  b–: – n.  b–:  b:  n.  a–:  a–b:  n.  a–b:  a–b:  n.  De generatione et corruptione a–b:  n.  b:  De incessu animalium a–b:  n.  De iuventute et senectute, de vita et morte b– a–b:  n.  b–:  n.  De longitudine et brevitate vitae a–:  n.  De memoria et reminiscentia a–: 

Index Locorum De partibus animalium a–b:  a–:  n.  a–:  n.  b–:  n.  b–:  n.  De respiratione a–:  n.  De sensu a–:  De somno a–:  n. ,  a–: ,  Eudemian Ethics a–: – n.  b:  n.  Historia animalium b–a:  n.  Metaphysics a–: – n.  b:  b–: – n.  a–:  b–:  n.  a–:  n.  b–a:  n.  a–b:  a–b:  n.  b–a:  n.  a–:  n.  a–:  a–:  b–: – n.  b–:  b ff.:  b–: ,  n. , – n.  a–:  n.  a–:  n.  a–:  n.  b–:  b ff.:  b–: – b–:  n. ,  b:  n. ,  n. ,  b ff.:  b: – n. , – n.  b–: , , , ,  b:  n.  b–: – n.  b–:  b–: , , , ,  b:  n.  b–:  n. ,  ff.

b



 –: , ,  b: – n. ,  b–: , – n. ,  b ff.: , – n.  b–a:  b–a: – b–: – n.  b: , ,  b–: – n.  b–: – n.  b: – n.  b–:  b–: , , ,  n.  b:  n.  b–:  n.  b:  n. ,  n.  b: ,  b–a: – n.  b–:  n.  b: – n.  b–:  n.  b–:  b–:  b:  n.  a–:  a–: ,  a–:  a–: , ,  n.  a–:  a–: , , – n.  a: ,  n. , – n. , – n. ,  a–: – n.  a–: ,  n. ,  a:  a–:  a–: ,  n. , , ,  a: ,  n.  a–:  a:  a–:  a–: , , ,  n. , , – n.  a–:  a–: – n. , ,  a–: , ,  ff.,  n. , , – n.  a:  n.  a–:  a:  n.  a ff.:  a–: – n.  a: 



Index Locorum a

 :  a:  a:  a–:  a:  a:  b–:  b–:  b:  b–:  b–:  b–: – n.  b–: – n.  b: – n.  b–:  b–:  n.  a–: – n.  a: – n.  a–: – n.  a: – n.  a: – n.  a–: – n.  a: – n.  a: – n.  a–a:  n.  b–:  n. ,  a–:  n.  b–:  n.  a–:  n.  b–:  n.  b:  n.  b–a:  n.  a–:  n.  b–:  n.  b–:  n.  a:  n.  a–:  n.  b–:  n.  b–a:  b–:  a–:  b:  n.  b:  n.  b–: , ,  b:  b–a:  n.  b:  b–:  n.  b–: – n.  b–: – n.  a–:  a–:  a–: 

a–:  a–:  a–:  b–:  b:  n.  b:  n. ,  b–a:  a:  n.  a–: – n.  b–:  n.  b–:  n.  b–:  Nicomachean Ethics a–:  n.  a–:  n.  a–:  n.  a–:  a–b:  n.  a–:  a–:  n.  b–a:  b–:  b–:  b–:  n.  b–:  b:  n.  a–:  a:  n.  a–:  a–: ,  a–:  a:  n.  a:  n.  a:  a:  n.  a:  n.  a–:  n.  b–:  n.  b–:  b–:  b–:  b:  n.  b–:  b–: ,  n.  b:  b–:  b–:  b–:  a–:  n.  a:  n.  a–: ,  a–:  b–: 

Index Locorum b

 –:  n.  a–:  n. ,  a:  n.  b:  n.  b–:  a–: ,  b–:  n.  a–:  n.  a–:  a ff.:  n.  a–:  n. ,  a–:  a–: – n.  a–:  b–:  b–:  n.  a–:  a–:  b–:  a ff.:  n.  a–:  a–:  a–b:  b–a:  a–:  b–:  b–:  b–:  a–:  n. , – n.  a–: – n.  a–:  a–: – n.  b–:  n. ,  b: – n.  a–b:  a:  b:  b–:  a–:  b–: , – n.  b–: – n.  b:  b–:  n.  Physics b–a: ,  a–:  n.  a–:  a–:  n.  b–:  n.  b–: ,  b–:  b–:  a–: 

b

 –:  b–:  b–:  b–:  b–: – n.  b–:  b–:  b–a:  n.  a–:  n. ,  a–:  n.  a–:  a–:  n.  a:  n.  a–:  n.  a:  n.  a:  n.  a–:  n. ,  a–:  n.  a–b:  a–: ,  a:  a:  a–:  a–:  a:  a–b:  n. , ,  n.  b–:  n.  b–: ,  n. , ,  b–:  b:  b–:  b–: ,  b: ,  b–:  b:  n.  b–: ,  b–: ,  b:  n.  b:  b–:  b–:  b–:  b–:  b–:  b: – n.  b–:  n.  Posterior Analytics a–b:  n.  a–: – n. , – n.  a–: – n.  a–: , – n.  a: – n.  a–: – n. 



 a

Index Locorum b

 – :  n. , – n.  a:  a:  b:  b–:  b–: , , – n. , ,  b–:  b–: ,  b:  b–: – n. ,  b:  b–: – n.  b–: – n.  b: , – n. ,  b–: – n.  b–:  n.  b–:  n.  b–:  n.  b:  n.  a–:  n.  a–:  n.  a–:  n.  a–:  n.  a–:  n.  a–:  n.  a:  n.  a–:  a–:  a–: – n. , ,  n.  a:  a–:  n.  a:  a:  a:  n.  a–: – n. ,  n.  a–:  n. ,  a–:  n. ,  a–:  n. ,  a:  a–:  n.  b–:  n.  a–: – n.  a–b:  n.  a–:  n.  Rhetoric b–:  n.  Sophistici elenchi a–: – n.  a–: – n.  Topics b–: – n.  b–: – n. 

b–: – n.  b–: – n.  Cicero De natura deorum . :  n.  Democritus,  DK B /: , , , ,  n.  B :  Diogenes Laertius . –:  n.  . :  n.  . :  . :  n.  . :  n.  . :  n.  . :  n.  . :  n.  Euripides Iphigenia in Tauris –:  :  Eusebius Praeparatio Evangelica . . –:  Galen De elementis ex Hippocrate libri ii . :  n.  De experientia medica . :  n.  De sectis ad eos qui introducuntur . :  n.  . :  n.  Hellenistic Philosophers, ed. Long and Sedley :  n.  :  :  n.  :  :  :  :  :  :  :  n. 

Index Locorum Matthew, Gospel of : –:  n.  Metrodorus of Chios,  DK A :  n.  Philoponus In Aristotelis De anima libros commentaria, ed. Hayduck . –. :  n.  . –:  n.  Plato Apology  :    –:   –:  Crito  –:  Euthyphro   – :  Gorgias   :   –:    – :   – :  Lysis   –: – n.  Meno   –  :  n.    – :  n.    – :  n.   – :  Parmenides  –:  n.   :  n.  Hypothesis VII:  Phaedo  :  n.   :  n.    : – n.    –  :  Phaedrus   ff.:  n.  Philebus  – :    –:    :   – :    –  :  n.    –:   –:   : 

  :    :    :   –:   –:   –:   –:   – :    –  :   –:    –:    –  :    :  n.    –:   –: – n.    –: – n.    –: – n.  Protagoras  :  Republic   –:    :    –: – n. ,  n.   :    –:    – :    : – n.    –: – n.    : – n.    : – n.    : – n.   –:  n.    : – n.    : – n.    ff.:    – :   :   –:    –:   :   – :  n.    –  :   –:  n.    :    – : – n.    :  n.    –:    :  n.    :  n.    –  :    – :   – :    :  n. 



   :  n.    –  :  n.   –:   – :   – :  n.   – :  n.    –:  n.    :    :    – : ,    :    :    –:    –:   :    – :    ff.:   :    –:    –:    –:    – :    – :    :    –:    –:  n.    :    :   :  n.    –: ,    :     ff.:    :    –:   :  n.    –:    :    –:    – :    –:     ff.:    – :    –  : ,    :    –:   – :   :  n.   –:  n.   –:  n.   –:  n.  Sophist  – :  n.    –:  n. 

Index Locorum  :    –:  n.    :    –: ,  n.    –: – n.    –: ,  n.    – :  n.    –:  n.    –:  n.  Symposium  –: – n.   – : –,    – : – n.  Theaetetus  – : – n.    – :    ff.:  n.    :    :   – :  Timaeus  – :  n.    – :  n.    – : ,  n.    – :  n.    :  n.    ff.:   – : , ,    –  :  n.    –:    :   :   :  n.    –  :    – :  n.    –:  n.    –  :    :    :    – :    –:   –:    –: ,    :  n.     ff.:    :   – :  n.    ff.: ,   :    –  : ,  n.    –:    – :    : 

Index Locorum  – :  n.    – :    –:  n.  Plotinus Enneads . . . –: ,  . . . –. :  n.  . . . –:  n.  . . . –:  . . . –:  . . –:  n. ,  . . . –:  n.  . . . –:  n.  . . . –:  . . . –:  n.  . . : ,  . . . :  . . . :  . . –:  . . . –:  . . –: ,  n.  . . : , , ,  n. ,  n. ,  . . . –:  . . . –:  . . . –:  n.  . . :  . . . –: – . . . –:  . . :  n.  . . . –:  n.  . . . –:  n. ,  n.  . . : ,  n. ,  . . . –: – . . . –:  n.  . . . –: ,  . . . –:  . . . –:  . . . : – n.  . . . –:  . . . –:  n.  . . . –:  n.  . . . :  n.  . . . :  n.  . . . –:  n.  . . . :  . . . –:  . . . :  . . . –: –,  . . . –:  n.  . . . –:  . . : 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

: ,  n. ,  . –: ,  . –:  . –:  . –:  . –: – n.  . –:  . –:  –:  :  . –:  . –:  :  . –:  n. , – : , ,  . –:  –: , ,  . –:  :  n.  : ,  –:  n.  . –:  . –:  . –:  n.  . –:  n.  . –:  n.  :  n.  . –: – n.  : ,  n. 

Plutarch Adversus Colotem  –:  n.  De communibus notitiis  : – n.   – :  n.  Quaestiones convivales . : – n.  Porphyry Fragments, ed. Smith :  Seneca the Younger De ira . . :  n.  Sextus Empiricus Adversus mathematicos . :  n.  . –:  . :  n.  . –: 





Index Locorum

. : ,  n. ,  Pyrrhoneae hypotyposes . :  n.  . :  . –:  n.  . –:  n.  . :  n.  . –:  . :  n.  Simplicius In Aristotelis Physica commentaria, ed. Diels .  ff.:  n.  . :  n.  . –:  n.  Sophonias In libros Aristotelis De anima paraphrasis, ed. Hayduck . –:  n. 

Stobaeus . . a:  Stoicorum veterum fragmenta, ed. von Arnim i. :  i. :  ii. :  ii. :  n.  ii. :  ii. :  Themistius In libros Aristotelis De anima paraphrasis, ed. Heinze . –:  n. 

Notes for Contributors to Oxford Studies in Ancient Philosophy . Articles should be submitted with double or ½ line-spacing throughout. At the stage of initial (but not final) submission footnotes may be given in small type at the foot of the page. Page dimensions should be A or standard American quarto (½× ″), and ample margins should be left. . Submissions should be made as a file in PDF format attached to an e-mail sent to the Editor. Authors are asked to supply an accurate word-count (a) for the main text, and (b) for the notes. The e-mail which serves as a covering letter should come from the address to be used for correspondence on the submission. A postal address should also be provided. If necessary, arrangements for alternative means of submission may be made with the Editor. Authors should note that the version first submitted will be the one adjudicated; unsolicited revised versions cannot be accepted during the adjudication process. The remaining instructions apply to the final version sent for publication, and need not be rigidly adhered to in a first submission. . In the finalized version, the text should be double-spaced and in the same typesize throughout, including displayed quotations and notes. Notes should be numbered consecutively, and may be supplied as either footnotes or endnotes. Any acknowledgements should be placed in an unnumbered first note. Wherever possible, references to primary sources should be built into the text. . Use of Greek and Latin. Relatively familiar Greek terms such as psychē and polis (but not whole phrases and sentences) may be used in transliteration. Wherever possible, Greek and Latin should not be used in the main text of an article in ways which would impede comprehension by those without knowledge of the languages; for example, where appropriate, the original texts should be accompanied by a translation. This constraint does not apply to footnotes. Greek must be supplied in an accurate form, with all diacritics in place. A note of the system employed for achieving Greek (e.g. GreekKeys, Linguist’s Software) should be supplied to facilitate file conversion. . For citations of Greek and Latin authors, house style should be followed. This can be checked in any recent issue of OSAP with the help of the Index Locorum. . In references to books, the first time the book is referred to give the initial(s) and surname of the author (first names are not usually required), and the place and date of publication; where you are abbreviating the title in subsequent citations, give the abbreviation in square brackets, thus:



Notes for Contributors T. Brickhouse and N. Smith, Socrates on Trial [Trial] (Princeton, ), –. Give the volume-number and date of periodicals, and include the full page-extent of articles (including chapters of books): D. W. Graham, ‘Symmetry in the Empedoclean Cycle’ [‘Symmetry’], Classical Quarterly,   (), – at –. G. Vlastos, ‘The Unity of the Virtues in the Protagoras’ [‘Unity’], in id., Platonic Studies, nd edn. (Princeton, ), – at . Where the same book or article is referred to on subsequent occasions, usually the most convenient style will be an abbreviated reference, thus: Brickhouse and Smith, Trial, –. Do not use the author-and-date style of reference: Brickhouse and Smith : –.

. Authors are asked to supply in addition, at the end of the article, a full list of the bibliographical entries cited, alphabetically ordered by (first) author’s surname. Except that the author’s surname should come first, these entries should be identical in form to the first occurrence of each in the article, including where appropriate the indication of abbreviated title: Graham, D. W., ‘Symmetry in the Empedoclean Cycle’ [‘Symmetry’], Classical Quarterly,   (), –. . If there are any unusual conventions contributors are encouraged to include a covering note for the copy-editor and/or printer. Please say whether you are using single and double quotation marks for different purposes (otherwise the Press will employ its standard single quotation marks throughout, using double only for quotations within quotations). . Authors should send a copy of the final version of their paper in electronic form by attachment to an e-mail. The final version should be in a standard word-processing format, accompanied by a note of the word-processing program used and of the system (not just the font) used for producing Greek characters (see point  above). This file must be accompanied by a second file, a copy in PDF format of the submitted word-processor file; the PDF file must correspond exactly to the word-processor file. If necessary, arrangements for alternative means of submission may be made with the Editor.