This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Advances in Consciousness Research Advances in Consciousness Research provides a forum for scholars from different scientific disciplines and fields of knowledge who study consciousness in its multifaceted aspects. Thus the Series will include (but not be limited to) the various areas of cognitive science, including cognitive psychology, linguistics, brain science and philosophy. The orientation of the Series is toward developing new interdisciplinary and integrative approaches for the investigation, description and theory of consciousness, as well as the practical consequences of this research for the individual and society. Series A: Theory and Method. Contributions to the development of theory and method in the study of consciousness.
Editor Maxim I. Stamenov Bulgarian Academy of Sciences
Editorial Board David Chalmers
Earl Mac Cormac
University of Arizona
Duke University
Gordon G. Globus
George Mandler
University of California at Irvine
University of California at San Diego
Ray Jackendoff
John R. Searle
Brandeis University
University of California at Berkeley
Christof Koch
Petra Stoerig
California Institute of Technology
Universität Düsseldorf
Stephen Kosslyn
† Francisco Varela
Harvard University
C.R.E.A., Ecole Polytechnique, Paris
Volume 55 Mind and Causality Edited by Alberto Peruzzi
Mind and Causality Edited by
Alberto Peruzzi University of Florence
John Benjamins Publishing Company Amsterdam/Philadelphia
8
TM
The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.
Library of Congress Cataloging-in-Publication Data Mind and Causality / edited by Alberto Peruzzi. p. cm. (Advances in Consciousness Research, issn 1381–589X ; v. 55) Includes bibliographical references and indexes. 1. Philosophy of mind-Congresses. 2. Causation--Congresses. I. Peruzzi, Alberto. II. Series. BD418.3.M554 2004 128’.2-dc22 isbn 90 272 5189 4 (Eur.) / 1 58811 474 0 (US) (Hb; alk. paper) isbn 90 272 5190 8 (Eur.) / 1 58811 475 9 (US) (Pb; alk. paper)
Chapter 5 Mental causation and intentionality in a mind naturalising theory Sandro Nannini
69
Chapter 6 The envious frog Marco Salucci
97
Chapter 7 Knowing what it is like and knowing how Luca Malatesti
119
Table of contents
Chapter 8 Human cognition: An evolutionary perspective Ian Tattersall Chapter 9 Space, time and cognition: From the standpoint of mathematics and natural science Francis Bailly and Giuseppe Longo
131
149
Chapter 10 Causality in the texture of mind Alberto Peruzzi
199
Index
229
Preface
The notion of causality has received the longstanding attention of scientists and philosophers. Recent advances in the neurosciences and in the physics of “complex systems”, as well as in the philosophical perspective of “naturalisation” of knowledge, have produced subtle but relevant changes at the juncture connecting the notion of causality with that of the mind. Such changes affect in a possibly decisive way the traditional landscapes of determinism versus indeterminism, monism versus dualism, top-down versus bottom-up architecture, local versus holistic approach and linear versus non-linear system dynamics. As the discussion of causality in the theories of mind is currently undergoing rapid changes, it is an appropriate time to discuss ideas about the meaning of such changes in recent theories in physics, biology and cognitive science. How do they affect the received view about cause-effect relationships and in particular with regard to the structure, the genesis and the nature of “minds”? The notion of causality has also received various interpretations in connection with phenomena of different kinds and scales. However, the central position concerning causal links, in the world-view of either the researcher or the layman, is no prelude in acknowledging any domain-specific range of validity to different characterisations of such links. Thus one is led to assume that if there is room for a given causal thesis, such as determinism, or the principle of locality, or reduction of the whole to its parts, then it holds from top to bottom, i.e., the same thesis holds for the architecture of any complex system. There are many scientific domains in which such assumption is under discussion. One of them is particularly interesting since there the notion of causality is not only a theoretical tool, but also an object of investigation. The domain in question is that of cognitive sciences. Contemporary theories of mind provide almost the whole range of attitudes towards causality (viz., in the sense assigned to causal explanation, the structure of an ontology of causes and effects, and the behavioural evidence for them). Such attitudes previously showed up in the context of natural sciences, but now they concern the effectiveness of the mind’s patterns and, ultimately,
Preface
the formation of the very idea of causality (e.g., in the context of cognitive development). For a long time, the discussion on the “foundations” of the cognitive sciences has, almost exclusively, focused on the identity/analogy/difference between information processing models of progressively more complex subsystems of the mind. The philosophers who rejected any and all such models as inherently inadequate were not looking for other models, but rather seemed to be content with arguing that, in real organisms, software and hardware entertain more than a contingent coexistence (contrary to the “multiple realisations” thesis of functionalism) and that the semantics (perhaps, even the syntax) of the mind’s software cannot be disjoined from its hardware. Over the last decade or so, the situation has changed. The “dynamical systems approach”, despite its limitations, has established a bridge with both the “perception-action” and with the “ecological” approaches. Many efforts have been made to lay the ground for an intrinsically causal theory of mind, which means, first of all, that its complex, many-layered, architecture can only be understood by taking the (biological) history of the whole body into account and by looking at the specific way (and timing) it performs certain functions. Developmental psychology takes into account the causal sequences through which any cognitive capacity matures and is stably integrated with other capacities. Evolutionary biology considers other kinds of causal sequences as does any other science involved in trying to understand the preconditions for the existence of minds. At issue is more than a difference in scale. It seems that different patterns of causality have to be simultaneously considered. The classical and the connectionist model of mind mainly differ in terms of the sort of causal architecture they posit. In contrast to both models, the dynamical systems approach seems to support the rejection of any mind-body dualism. In both philosophy of mind and cognitive psychology, such models have been interfaced with ontological and methodological debates concerning the identity thesis, the nature of qualia, the role gestalts have to be assigned and the relationships between perception and language, to mention only a few topics. This interface is less articulated for the dynamical systems approach, but its naturalistic profile is more robust. In order to assess the exact reasons of reductionist versus emergentist viewpoints, it is useful that the different senses of “naturalism” can be clearly identified, as each of them is associated with an emphasis on specific data and methods. Here, the contribution of a precise philosophical analysis is helpful, even though its rigour is far from the experimental and simulative analysis of causal aspects of mind.
Preface
Within the study of language, much attention has been paid to determine the import of causal “chains” in fixing the reference for relevant classes of expressions. On a meta-theoretical level, a subject largely debated by analytic philosophers is that of which explanatory potential causal theories of reference can be safely assigned. Another, no less debated, issue concerns the interface between physical and mental causation. Depending on which solution is adopted, the task of determining the exact role of biological constraints, in both the phylogenesis and ontogenesis of mind, can be accomplished in different ways, which have consequences even for the nature/culture dichotomy. Rather than providing a historical reconstruction of the pathways that converge into the present, multi-faceted, debate on causality, this book faces some of the main issues in the recent literature concerning causality and mind. The set of papers covers a good part of the spectrum of present methodological perspectives and thus provides a collectively critical survey of the state of the art. Each paper suggests arguments that point at the need of taking simultaneously into account various approaches, in order to identify and evaluate their points of convergence and divergence. The paper by Brian Hopkins (“Causality and development: Past, present and future”, pp. 1–17) faces issues about the ontogenesis of mind and the mechanisms that create the changes observed. The task of explaining such changes is problematic, largely because of the narrow views that have been adopted about what constitutes causality in a developmental context. In fact, we have no satisfactory theory that accounts for the mechanisms of developmental change, which most probably occur under complex non-linear regimes. After a short overview of the historical background of what constitutes causality in development, Hopkins suggests that dynamical systems thinking can offer general guidelines for overcoming this problem. Illustrations are given from speculations about the link between prenatal and postnatal development and more particularly from those concerning the co-development of posture and action. The chances offered by a dynamical systems approach are also explored in Riccardo Luccio and Donata Milloni’s contribution (“Perception of causality: A dynamical analysis”, pp. 19–24). In order to vindicate Albert Michotte’s arguments in support of the thesis that causality can be directly perceived in many situations in that it presents the characters of the “encountered” (to use Wolgang Metzger’s terminology), Luccio and Milloni emphasize that experimental phenomenology, mainly in the vein of Gestalt psychology, has exhibited several instances of such direct perception of causality. They argue that many of them, from the so-called launch effect to the tunnel effect and Spizzo’s effect, manifest a characteristic run, with transitions from one perceptual pattern to
Preface
another, that could be explained at best in terms of non-linear dynamics. In Luccio and Milloni’s view, synergetics appears a particularly apt tool to build up a consistent interpretation of such dynamics. The paper by Andy Clark (“Embodiment and the philosophy of mind”, pp. 35–51) provides a broad philosophical perspective centred on the notion of “embodiment”, in contrast to both Cartesian dualism and contemporary reductionism. Against the persisting idea that the task of mind is that of constructing an inner model of the world, composed of representations that can then be manipulated in algorithmic way, Clark argues for a different picture of the mind as an interwoven system, incorporating elements of brain, body and world. However, this does not imply that human beings have to be seen as undifferentiated, holistic and representation-free structures. In particular, Clark appeals to the results of research on “animate vision” and robotics, in which body-centred strategies and environmental interactions have an essential role and reveal the importance of complex causal loops, driven by continuously-varying external parameters. After a comparison between radical interactionism with minimal Cartesianism, human thought is framed in terms of a dialectics between action-oriented resources, linguistic competence and cultural practices, where representations can be seen as dynamical patterns. The phenomenological approach developed by Maurice Merleau-Ponty and its connection with the methods of present-day psychology are the subject of Antonella Lucarelli’s paper (“Causes and motivations: Merleau-Ponty’s phenomenology confronts with psychological studies”, pp. 53–68). The recent Merleau-Ponty renaissance has credited the French philosopher as being one of the main forerunners of the “embodied mind” view and, in this vein, Lucarelli focuses on his objections to the assumption of an autonomous subject, independent from either biology or culture, as well as to the primacy of rational consciousness in the structure of mind. – In short, psychical causality cannot be investigated apart from bodily activity, which moulds the whole phenomenal field. As Lucarelli notes, Merleau-Ponty invites us to look at mental phenomena as the result of multiple interactions between subjective intentionality and cultural meanings as they are objectified in specific socio-historical contexts. This applies to language too: according to Merleau-Ponty, we speak as embodied minds, i.e., we do not speak about the body, but by and through the body. At stake here is the effacement of a purely representational, omnipotent, logical ego, in favour of an interactive practice registering the depth of the experiential fine-tuning to the other.
Preface
Rather than from an updated phenomenology of embodied meanings, the paper by Sandro Nannini addresses the question of naturalisation in the light of contemporary philosophy of mind (“Mental causation and intentionality in a mind naturalising theory”, pp. 69–96). Nannini discusses various kinds of criticism directed towards the naturalisation of mind. He does with reference to Karl Popper’s remark that materialism falls necessarily into epiphenomenalism, and to John McDowell’s claim that naturalists are wrong because they reduce the semantic potential of mental representations (their referencehooking in the external world) to the causal action of outer physical events on the brain through the senses. After replying to Popper’s and McDowell’s arguments, Nannini faces the problem of how brain processes can implement mental events without losing their intentionality. By comparing Jerry Fodor’s cognitive solution and Paul M. Churchland’s connectionist and eliminative solution, it is argued that the latter view is preferable to the former. No less debated topics, also related to issue of naturalisation, are considered by Marco Salucci (“The envious frog”, pp. 97–117), providing a detailed examination of both the arguments advanced by Thomas Nagel in 1974 and, ten years later, by Frank Jackson against the reduction of consciousness to brain states, and the several and articulated answers by identity theorists to these attacks. Although the reductionists’ counterarguments were not convincing (recently, also David Chalmers has claimed that the identity theory is wrong, by appealing to Nagel’s style arguments), Salucci presents seven points characterizing identity theory with the aim of showing why Nagel-style arguments are not conclusive. In Salucci’s view, any Nagel-style argument requests us to experience how our brain states are (or become) identical to mental states, but we cannot get such an experience because the brain has no special organ detecting its own states. These arguments simply point out a “natural” fact: the lack of such a special organ. So the antireductionist value of Nagel’s argument amounts to the antireductionist value of such a fact and it has no consequence other than that following from the lack of a couple of eyes fit to see infrared radiation. From a different perspective, Luca Malatesti (“Knowing what it is like and knowing how”, pp. 119–129) deals with strictly related issues of philosophy of mind, focusing on the so-called knowledge argument, by means of which Frank Jackson intended to reject physicalism. This argument exploits the intuition that by having colour experiences, we know what it is like to have these mental states, and in fact Jackson takes this knowledge to be about features of colour experiences that a complete scientific knowledge cannot accommodate.
Preface
In his paper, Malatesti considers two criticisms of the knowledge argument. Daniel Dennett and Patricia Churchland have maintained that the notion of complete scientific knowledge of colour (and colour vision is involved in the knowledge argument) is unintelligible. Although Malatesti endorses this criticism, he argues that a version of the knowledge argument weaker than Jackson’s can be formulated by using an intelligible notion of scientific knowledge of colour vision. Malatesti notes that David Lewis and Laurence Nemirow have formulated a second criticism of the knowledge argument. This criticism might also affect the weaker version of the knowledge argument that Malatesti is suggesting. But, one assumption of this argument is that knowing what it is like amounts to propositional knowledge, whereas Lewis and Nemirow have argued that knowing what it is like is a form of knowing how. Although there are important connections between knowing what it is like and the set of abilities they refer to, Malatesti argues that knowing what it is like to have an experience does not reduce to knowing how. The contribution by Ian Tattersall (“Human cognition: An evolutionary perspective”, pp. 131–148) allows the reader to integrate the approaches to naturalisation of mind so far discussed (and the difficulties they face in dealing with causality) with a phylogenetic analysis of the cognitive resources achieved by Homo sapiens. It is our symbolic cognition above all else that, as far as we know, distinguishes our species from every other organism that has ever lived. Thus, Tattersall deals with questions such as: When did our precursors acquire this unprecedented attribute? How did they move from a non-symbolic to a symbolic state of consciousness? Examination of our fossil and archaeological records suggests that this transition was not a matter of gradual honing by evolution over millions of years. Tattersall claims that it was a recent occurrence, and a rapid one, and argues that it was an emergent event, rather than one guided by the processes of natural selection. Tattersall’s suggestion is that this claim is an important one to consider before we conclude that modern human behaviours are fine-tuned for anything at all. What is the import of contemporary mathematics, as applied to the natural sciences, for the construction of more adequate models of the mind? By emphasising the variety of methods that turned out to be successful in physics, and the changes in such methods called for by biology, Francis Bailly and Giuseppe Longo (“Space, time and cognition: From the standpoint of mathematics and natural science”, pp. 149–197) relate different causal regimes to various frames for space and time in physics. After a clear and suggestive review of the basic theoretical notions of classical, relativistic and quantum approaches, they argue that current views in biology, and even more so in cognitive sciences, are
Preface
largely bound by underlying epistemological standpoints in physics. They provide a wide and conceptually deep discussion of key notions (e.g., critical state transition and asynchronicity) relating physics, biology and cognitive sciences to the foundations of mathematics. The resulting picture is also relevant to the present controversy on natural versus artificial intelligence. Finally, my contribution (“Causality in the texture of mind”, pp. 199– 228) deals with general philosophical issues about causality in connection with models of the mind. Rather than providing a survey of the literature, it identifies a few key-points in the debate opposing the computational model of mind, based on high-level information processing, to models of reduction and emergence. In this regard, special emphasis is, once again, centred on dynamical systems. While not synchronised with the results of experimental research, theories of mind are largely dependent on the role that causality is assigned in explaining mental properties. Since there is more than one idea of causality, I examine some differences among theories of mind arising from the appeal to one idea rather than to another. Such an examination leads to a comparison of different methods of explanation in the cognitive sciences. In particular, the relationships between semantic competence and sensory-motor systems are exploited as a source of relevant information. A properly naturalistic stance is suggested as being able to avoid the independence of the formal from the material as well as the need of appealing to global holism. As the non-linear character of strongly coupled dynamical systems does not support “physicalistic” reductionism, so the emergence of macro-patterns of perception and action does not imply dispensing with physics. The book originates from a Conference entitled “Mind and Causality”, organised by the Department of Psychology, University of Florence, in October 2001 in Florence, Italy. Together with the theoretical, experimental and philosophical interest of the talks given on that occasion, the warm and collaborative atmosphere among the participants lasted well beyond the span of the Conference and led to an enlargement of the discussion to a wider range of researchers on the same subject. I wish to thank all of them for their contribution to this volume. A particular acknowledgement goes to Brian Hopkins, whose encouragement and expert assistance have been decisive in realising the project. Alberto Peruzzi Florence, 2003
Chapter 1
Causality and development Past, present and future Brian Hopkins Lancaster University
.
Introduction
Ontogenetic development is both a complex and creative phenomenon – complex because of its high dimensionality and creative in terms of resulting in the formation of new structures and functions. At least, this very general depiction is shared by most contemporary developmental theorists even though their explanations of the processes involved can be radically different. Ultimately, such differences arise from contrasting and seemingly irreconcilable views about how to construe the mechanisms or causes of developmental change. Broadly speaking, such causes have been assigned to one or other form of ‘instructionism’ residing either in the products of genetic activity (i.e., maturationism) or in the shaping forces emanating from the external environment (i.e., environmentalism). Alternatively, and more recently, these two ‘instructionisms’ have been combined, resulting in something labelled ‘interactionism’ and in which internal and external prescriptions somehow meld together to generate developmental change. As rightly pointed by Oyama (1985), none of these ‘-isms’ can capture the true nature of development because they prescribe the mature state before it actually develops. In short, they do not enlighten us about the process of biological development (viz., how new forms with greater complexity emerge) nor about the changing mechanisms that underlie this process (viz., why a particular developmental change takes place). Such issues have long challenged developmentalists and as yet there are no satisfactory answers – at least not with the grand theories of development such as that elaborated by Piaget (1971) in his unique approach to structuralism. Perhaps then we require a different causal
Brian Hopkins
framework for dealing with the inherent complexity and creativity of ontogenetic development and one that abides by a set of principles germane to the behaviour of time-dependent complex systems? In attempting to address this question, and for which there are no readymade answers, we need first to consider historical changes in the meaning of causality so as to arrive at a scenario that promotes further insights into the outstanding problem of developmental change (Section 2). Inevitably, this exercise brings with it the distinction between determinism and indeterminism and what that implies for such a scenario (Section 3). This distinction is a central one for the various dynamical systems approaches to development, which in some cases seem to offer guidelines for the resolution of the change problem (Section 4). Finally, such approaches are evaluated in terms of their shortcomings and how they might be overcome (Section 5).
. Changes in the meaning of causality: A potted history The ultimate sources of all ideas about causality is, of course, Aristotle (384322 BP) with his delineation of material, efficient, formal and final causes (see Table 1). Aristotle’s causes can be classified with regard to necessary and sufficient conditions. A necessary (or sine qua non) condition is one without which development would not take place if it failed to occur. Material, effiTable 1. Aristotle’s four causes and the domains most associated with each one. For Aristotle, material and formal causes ensured a state of being while efficient and final causes were responsible for becoming (i.e., change). He further emphasized that: (a) developmental explanations require all four causes, (b) material, formal and efficient causes were necessary, but not sufficient, conditions for such explanations, and (c) sufficiency was imparted by final causes, which in turn were the product of formal and efficient causes. Cause
Meaning
Domain
Material
That from which something is produced That according to which something is done That by which something initiated Sake for which something is done
Neuropsychology
Formal Efficient Final
Cognitive psychology
BEING
Behaviourism Embryology
BECOMING
Causality and development
cient and formal causes all constitute necessary conditions in this respect. A sufficient condition is one with which development does occur by means of interactions between efficient and formal causes. The product of these interactions is Aristotle’s final cause that accounts for the apparent goal-directedness of development. It is important to recognize that sufficient conditions do not have to be strictly causal in nature. They may be causally connected with what it is they are a condition of, but without being the cause of it. For example, a particular behavioural or brain state may be a sufficient condition for selective attention, but it is not the cause of it. Following in the footsteps of Aristotle, Francis Bacon (1561–1626) established that causality could be open to empirical investigation. Thus, according to Bacon, if A causes B, then there is a logical relationship to be discovered by understanding the essences of A and B. Such an approach to causality that was part of his inductive methods for collecting and interpreting data resulted in Bacon’s major achievement: the separation of science from philosophy. Although Bacon was not interested in biological development, his legacy of what constituted causality for him was profoundly to affect those who did in the 17th century and who formed themselves into two competing factions: the epigeneticists and the preformationists (see Needham 1959). By the 18th century, opposition appeared to the Aristotle-Bacon interpretation of causality in the person of David Hume (1711–1776). In attempting to systematically develop the sort of empiricism advocated by John Locke (1632– 1704) and George Berkeley (1685–1753), Hume argued that causality is not a logical relationship waiting ‘out there’ to be discovered, but rather was based on inferences from experiencing a succession of events. In doing do, he shifted the study of causality from logic to psychology. Hume also strived to establish what were the defining characteristics of causality. Four of his delineations can be mentioned: cause and effect must overlap in space and time, a cause must always be prior to its effect, cause and effect must have a constant union, and the same cause always or invariably produces the same effect (the latter becoming known as the causality principle, the validity of which is asserted by the doctrine of causal determinism or causalism). Furthermore, Hume regarded a cause as a necessary condition or sufficient or both for the occurrence of an event. Labelled ’elementaristic causality’ by Valsiner (1997), and reflecting an adherence to laws of succession (see below), Hume’s theoretical edifice became challenged in the 19th century. The challenger was John Stuart Mill (1806–1873) with his notion of multiple causation, which he regarded as a principle of uniformity in nature and that stood in opposition to the simple, linear causality espoused by Hume and
Brian Hopkins
those before him. In brief, causation now was something that occurred through the grace of multiple intersections of interweaving causal lines and none of which on their own brought about an effect. As such, it was the first example of systems thinking and the first attempt to account for the issue of emergent properties in an apposite causal framework. Valsiner (1997) referred to such a framework as ‘systemic or circular causality’ and which is rooted in laws of co-existence (see below). Now, with Mill, we are beginning to encroach upon modern ideas about pertaining to causality and a framework more conducive to unravelling the complexities of developmental change. Systemic causality holds that the outcomes of a system are not due to one of its elements, but rest on functional relationships between them. As a consequence, if there is change in one of them, then the whole system changes. Moreover, these relationships are due to the actions of a catalysing agent, which as such neither produces effects nor is an immediate part of the system (Valsiner 1997). However, its action is a necessary condition for a new outcome to emerge. A further elaboration on multiple causation was provided by Mackie’s (1974) notion of an inus condition. By this is meant an insufficient, but nonredundant part of an unnecessary but sufficient condition. For example, providing support for an infant who has just started to walk by holding the hands is an insufficient (balance mechanisms for independent walking are needed), but non-redundant (infant uses support to produce forward locomotion) part of an unnecessary (other forms of support are possible such as holding onto a table edge) but sufficient condition (infant will produce forward locomotion). In the development of a particular ability (e.g., object permanence), there are probably many more than just one set of inus conditions that are sufficient to explain its appearance, but no one set is necessary. The task for the developmentalist then becomes one of specifying a particular conjunction of unnecessary but sufficient conditions required for the emergence of a new ability. Put another way, useful (causal) explanations of developmental change are arrived at by restricting the range of possibilities to clearly defined sufficient conditions, but without taking into account all the necessary ones. The notion of an inus condition relates directly to the vital task of experimentally verifying the agents of developmental change in terms of control parameters (see Section 4). In science in general, explanations of change have been typically based on one of two so-called laws (Nagel 1961): Laws of succession and laws of coexistence. The former, sometimes referred to as genetic explanations is an account (i.e., the explanans) of a sequence of events leading to the occurrence of a fact to be explained (i.e., the explanandum). Common in history, geology and biology, these accounts take the form “If x has property P at time t, then x has
Causality and development
property B at time t later than t” (Nagel 1961: 76). Characteristic of research on motor development until recently, such laws are not explanations, but descriptions of relatively invariant sequences of functional changes in that the explanans (e.g., maturation) is typically vacuous. In fact, theorizing on motor development has consisted of genetic explanations of sequences of explicanda in search of appropriate explanantia. In contrast, laws of co-existence attempt to account for spatial and temporal regularities in the behaviour of a system by relating the actions of its different components to one another. When one component changes, a new state of co-existence may emerge in which the system’s output consists of qualitatively different spatial and temporal properties. Thus, laws of succession might be explained by changes in the underlying patterns of co-existence between diverse components of a developing system. As we shall see, the making and remaking of successive states of co-existence is an important tenet of recent theorizing about developmental change. It should be recognized that laws of co-existence are not causal in the sense conveyed by elementaristic causality. A relevant example is Boyle’s law which holds that the pressure (p), volume (V) and absolute temperature (T) of a given mass of gas satisfy the equation pV = RT, where R is the gas constant. In this case, it is impossible to conclude that changes in pressure, volume or temperature of a gas cause a change in one of the others. No variable has priority in this respect as p, V and T determine one another in a mutual fashion. Thus, Boyle’s law, like Hooke’s law relating a mass supported by a spring to its elongation, is an example of systemic causality, but nevertheless strictly deterministic in its operation.
. Determinism versus indeterminism Determinism is the belief that all events are subject to causal laws and are themselves causes of other events [i.e., causes cause causes to cause causes to use Wilden’s (1972) alliteration]. Classical (Newtonian) mechanics, Maxwell’s electromagnetics as well as Einstein’s special and general theories of relativity were all founded on strict interpretations of determinism. Indeterminism amounts to a denial of this belief and in its mildest form contends that there is at least one event for which no preceding events are necessary antecedents. With the rise of quantum mechanics in the first half of the 20th century, the view gained ground that the laws of nature were not deterministic, but at best assertions about the statistical probability of occurrences at the sub-atomic level.
Brian Hopkins
Despite opposition to such assertions (e.g., by Einstein with his famous statement that “Der Herr Gott würfelt nicht” or “God doesn’t play dice”), quantum mechanics wrought major changes in the formal structure of scientific explanations, which were gradually assimilated into the biological sciences, largely through the writings of Schrödinger (1944). Consequently, the polarities of determinism and indeterminism (or rather stochasticism) became welded into theoretical approaches that attempted to account for the dynamical behaviour of time-evolving systems (see Section 4). In Freudian psychoanalysis as well as in both biology [e.g., Waddington’s (1975) concept of canalisation] and psychology [e.g., Werner’s (1948) orthogenetic principle], development has been portrayed in strongly deterministic terms. Still a persistent tendency, especially in developmental psychology, such a portrayal could be a legacy of classical epigenesis, a basic principle of which is that: Development itself, besides being a period of gene action, is a period of complex stepwise (epigenetic) reactions in which the conditions of one step cause the next. (Needham 1959: 107)
Ever since the pioneering histological studies of Ramôn y Cajal (1852–1934), however, incontrovertible evidence has accrued that the achievement of topological specificity between effector and receptor cells during neural development emerges from incredibly complex processes. These processes involve stochastic competition for a limited number of target sites resulting in irreversible neural events such as selective cell death and axonal retraction. In turn, the degree of stochasticity is constrained by morphogenetic fields that in effect transform randomness at the microlevel into a new biological order at the macrolevel (Stent 1981). If, according to the argument being forwarded here, development involves an interplay between determinism (necessity) and stochasticity (constrained chance), what sorts of theoretical approaches would resonate with this claim? This question is addressed in the next section dealing with dynamical systems approaches to development, but it requires a preparatory step. Such a step can be taken with a basic equation from quantum statistical mechanics designed to capture the increasing and irregular evolution of a Brownian particle (also know a Wiener process) by using a stochastic term to represent the random forces of molecules at the macrolevel. This is Langevin’s (1872–1946) stochastic differential equation, which can be symbolised as follows: dq/dt = q˙ = N(q, p) + F(t)
Causality and development
Table 2. Dynamical systems approaches in terms of their originators, the initial field of application and relevant developmental studies. To date, such studies have not been based on irreversible theormodynamics, perhaps because its application requires a definite and stringent mathematical model. Approach
Main figure
Field
Developmental application
1. Catastrophe theory 2. Chaos theory 3. Irreversible non-equilibrium thermodynamics 4. Synergetics 5. Topological dynamics
Biology v/d Maas & Molenaar (1992) Meteorology Robertson et al. (1993) Chemistry – Physics Fitzpatrick et al. (1996) Mathematics Clark et al. (1993)
where q = a vector with many dimensions (q1 , q2 , qn . . .) specifying the state of the system and N = a deterministic non-linear function of the state vector that depends on a (set of) time-evolving parameter(s) (p) and random (or stochastic) forces (F) acting on the system due to external or internal noise. The point to be made is that theoretical approaches addressing the behaviour of complex and open systems differ in terms of whether the nonlinear function is only deterministic (i.e., it uniquely specifies past and future states whose initial conditions are known precisely) and if it includes a stochastic term. This difference gives rise to dynamical systems approaches that are fully deterministic (e.g., chaos theory; topological dynamics), mainly stochastic (e.g., irreversible thermodynamics) or both deterministic and stochastic (e.g., synergetics). Catastrophe theory is a special case in that it was originally conceived as being fully deterministic (Thom 1975), but has since been reformulated to include stochastic processes (Cobb & Zacks 1985). Table 2 summarizes these approaches in giving their origins and references to instances when they have been applied to the study of development.
. Dynamical systems approaches: Determinism and indeterminism Langevin’s equation leads to a distinction that lies at the heart of the sort of determinism espoused by dynamical systems approaches. This is between order parameters (synergetics), behavioural variables (catastrophe theory) or collective variables (the state vector q) on the one hand and control parameters (the p term) on the other (see Hopkins 2001, for further details).
Brian Hopkins
An order parameter is a single macroscopic entity that captures the behaviour of a complex system in terms of low dimensional descriptors (e.g., relative phase between moving limbs). It operates to create the collective state (or ordering) of a system and as such is a topographical variable and not a control entity. The latter is the function of a control parameter and which is analogous to an inus condition. In short, it functions to move systems between its collective states. There are four important features about a control parameter. Firstly, when it is linearly scaled beyond some critical value, it may induce stochastic behaviour in the order parameter followed by a discontinuous change to another qualitatively different state. Secondly, it is completely aspecific with regard to the change it induces. Thus, a control parameter does not contain prescriptions as to how a system should change unlike its symbolic counterparts such as genetic programmes or schemata. Thirdly, different mechanisms can serve as control parameters at different ages (e.g., hormonal at one age, neural at another and cognitive at yet another). Fourthly, some form of experimental manipulation is required to identify such age-specific control parameters with any confidence. The last two points are brought into relief by a study of testosterone effects in seagulls (Groothuis & Meeuwissen 1992). When administered to young chicks, it induced precociously complete agonistic displays. With adult birds, high testosterone levels were not obligatory to trigger this display as it also occurred when this hormone was at a low level. All told, these findings identify testosterone as an age-specific control parameter in that when it is scaled-up beyond some critical value a later developmental transition was simulated. The non-specific catalysing effect of control parameters is only applicable to non-linear systems that operate in far-from-equilibrium conditions. A hallmark of such open systems is that they respond by self-organizing when faced by external perturbations in these conditions. Self-organization, which complies with the laws of co-existence, refers to a process by which new structures and functions spontaneously emerge without specification from the external environment. Its meaning can be illustrated with reference to formal logic, the starting point for which is logical calculus, and to the humble snowflake. A logical calculus involves, among other things, rules for the formation of complex expressions. In a sequence of such formation rules, the extremum (or last rule) states that no expression is valid unless its existence follows from the preceding rules. The type of logic required for understanding the meaning of ‘follows from’ in the extremum rule led to new types of logic known as formal systems. Such systems are depicted as being self-contained in that they make
Causality and development
no appeal to an external logical system. In a similar way, self-organizing systems exhibit an internal logic arising from the system’s own dynamics and not through an external, independently existing mode of organization. For example, the lattice structure of a snowflake arises from quantum exchange interactions in which exchange particles or quanta of thermal energy are manifested by self-propagating excitatory loops (Nittman & Stanley 1986). These loops create and maintain a cooperative effect responsible for the macroscopic order of the system. Thus, given certain boundary conditions based on a field of temperature sensitive gradients, thermodynamic (or extremum) processes are triggered resulting in the non-linear transformation of structureless water droplets into the lattice structure of snowflakes. Self-organization then is a temporal and spatial process in which an essentially structureless system becomes organized by non-linear, partly random mechanisms involving the self-assembly of interacting constituents – none of which contain prescriptions for the new state of organization. So far, we have emphasised how dynamical systems approaches cater for determinism (viz., that control parameters deterministically, but unspecifically, induce changes in the order parameter of self-organizing systems). What then is to be gained by a deterministic system having some degree of indeterminism (i.e., stochastic properties)? One potential benefit is that it amplifies the emergence of new forms generated by dynamical, but deterministic, regimes associated with periodic, quasi-periodic and chaotic attractors. In doing so, it enables new behaviours to be assembled creatively so that novel and specific task demands can be addressed throughout development. Another is that stochastic fluctuations superimposed on deterministic processes facilitate exploration of the local environment and thereby maximize the extraction of relevant sources of information. The ability to capitalise on such fluctuations in this way may constitute a source of individual differences in development. This last point raises the possibility that deviant development could be a consequence of either a lack of deterministic constraints on stochastic processes (e.g., as might be the case in hyperactive children with attentional deficits) or of a rigidly determined system (e.g., in those children with cerebral palsy). Further speculations along these lines can be found in Hopkins (2002). Dynamical systems approaches such as synergetics that allow for stochastic behaviour in an order parameter are especially germane to the study of developmental transitions (i.e., a change from one attractor state to another in a well-defined developing system during a restricted period of development). Defined in terms of non-equilibrium phase shifts, they constitute the simplest manifestation of self-organization in high dimensional complex systems.
Brian Hopkins
Such shifts in an order parameter are referred to as bifurcations. As a bifurcation point is approached, an order parameter begins to manifest stochastic behaviour or what are termed critical fluctuations or anomalous variance (Gilmore 1981). In addition, its number of degrees of freedom is reduced and its behaviour can be captured in low dimensional terms. According to the theory of noise-induced transitions (Horsthemke & Lefever 1984), such random fluctuations in the behaviour of an order parameter provide the material upon which deterministic processes can act to create new spatiotemporal patterns defined in terms of one or other attractor state. Thus, in complex systems determinism and stochasticity are complementary. Having said that, the problem remains of being able to distinguish stochasticity whose dimension is infinite from deterministic chaos that has a broad-based spectral component like noise as standard waveform analyses such as the Fast Fourier Transform cannot reliably make this distinction. Fortunately, solutions are available to overcome this problem (e.g., see Pincus 1991).
. Whither developmental causality? As we have seen, dynamical systems approaches emphasise that ontogenetic development has a number of characteristic features – all of which mitigate against traditional notions of causality being applied to time-evolving systems. Firstly, development cannot be ascribed to monocausal agents. Rather, it is a process of self-organization resulting from interactions (i.e., circular causality) between co-existing sub-systems that are marshalled to deal with the demands of particular tasks. Secondly, developmental change is a consequence of an interplay between deterministic and stochastic processes, with the former ensuring the achievement of species-characteristic end states and the latter facilitating the creative assembly of new states. Thirdly, new states emerge from discontinuous phase transitions in which age-specific, but non-prescriptive, control parameters are scaled up beyond some critical value. Just prior to, and during, such transitions the order parameter can display an array of behaviours associated with self-organizing systems such as critical fluctuations and sudden jumps without any intermediate states (see Hopkins 2001, for further details) What is important to bear in mind is that these and other features of dynamical systems approaches do not provide a full account of development. This is because they are concerned with capturing the organizational principles of change rather than the specific mechanisms that generate transformations across ontogenetic time. To do so, requires building a theoretical bridge
Causality and development
between dynamics (i.e., action) and information (i.e., perception) – something that has yet to be achieved in a satisfactory manner and which continues to divide dynamical from cognitive or information-processing approaches to development. Such a bridge-building enterprise is faced with a number of problems (see Table 3). Perhaps the greatest challenges to a theory of perception and action commensurate with dynamical systems thinking are contained in the independence and missing dimension problems. The first is a deeply philosophical issue stretching back at least to George Berkeley and the publication of his An essay towards a new theory of vision (1709) and De motu (1721). The philosophical complexities that have attached themselves to this problem are beyond the scope of this contribution. The missing or lost dimension problem, not touched upon by philosophers of perception, needs to be resolved if information and dynamics are to be aligned in a single unified theory of development. In a nutshell, what is at issue is how kinematics (information) that do not contain mass and thus force can specify kinetics (dynamics). Without resolution, this problem continues to undermine attempts at providing models of circular causality in which information residing in low-energy kinematic flow fields map onto high-energy force fields during the development of action. Can the differing causal frameworks associated with the dynamical systems and cognitive approaches to development ever be theoretically reconciled? In other words, can continuous, rate-dependent information be theoretically aligned with symbolic, rate-independent information? Inspired by Bohr’s resolution of the wave-particle problem in quantum mechanics, Pattee (1987) contended that the two forms of information are complementary, but irreducible to each other. Thus, resolution boils down to attaining a correspondence between indicational and specificational information. Information in the dynamical sense specifies potentials for action and is thus not arbitrary. From the cognitive perspective, however, it is an arbitrary convention indicating what has to be done, but not how. Accordingly, symbols are somehow transformed from a stage of response selection to that of programming as meaningful (i.e., indicational) information. Put this way, the two approaches seem to be irreconcilable. Resolution and reconciliation for some are to be found in connectionist models – a generic term for a whole raft of dynamically inspired simulation techniques that include neural network models. There are reasons, however, for doubting whether connectionist models do indeed provide a tertium quid between the two approaches. One such reason concerns the distinction between laws and rules.
Brian Hopkins
Table 3. Problems associated with applying the ecological approach (Gibson 1979) and the perception-action approach (Kugler 1986) to dynamical systems theory. The problems are not strictly separate in that those confronting the ecological approach also concern the perception-action approach. A2 and C1–C2 are problematic for dynamical systems approaches in general, particularly in their application to development. Rather, it is a problem for cognitive approaches – the problem of the homunculus (see Hopkins 2001). A3 and B present the most serious theoretical (A3) and modelling (B) challenges to a direct realist theory of perception and action built on natural laws. Approach
Problem
A. Ecological 1. Multiple Affordance specification1
2. Intentionality1
3. Independence3
Tau
4. Alternatives2
5. Correlational2
B. Perception- Missing action dimension1
C. Both
1. Empty vessel3 2. Metaphor1
Description
References
There is only 1:1 bidirectional mapping between informational invariants and object properties: for each action only 1 affordance, for every affordance only 1 action. This is considered to be implausible (a) and its rejection would: i. mark a return to the proximal-distal distinction (b) of indirect perception (1: many mapping) and ii. falsify claim that information is specificational. An alternative is directed perception (c): many-to-1 mapping with animal selecting between sources of information. But this form of perception is rule- rather than law-governed. Intention in perception defined as ability to apprehend objects (a). Too general as it would include physical systems (e.g., thermostat). Intention in action can be seen as intentions harnessing affordances and their associated effectivities in realising a goal-directed action. Intentions are then an extraordinary class of constraints (b). This says nothing about their ontological status. According to direct realism, affordances exist independently of perceiver who may or may not perceive them. Excludes animal from determination of affordances and thus contrary to principle of mutualism (a). Also contrary to idea that objects change their affordances during development. Leads to conclusion mutuality and realism should be given up (b). TTC* could be obtained in other ways than relative rate of expansion of approaching surface on retina [e.g., by means of binocular divergence and convergence (a), and distance/velocity ratio (b)].
a. Neisser (1977) b. Brunswick (1956) c. Cutting (1991)
a. Tresilian (1993) b. McLeod & Ross (1983)
Evidence in support of a tau-based control of prospective actions is largely correlational [i.e., it provides qualitative not quantitative predictions (a)]. An exception is (b) which demonstrated tau-based control of vertical impulse in running over an irregular surface. An alternative is that it is situational dependent (c).
a. Tresilian (1994) b. Warren et al. (1986) c. Schiff & Oldak (1990)
Information is a kinematic flow field with dimensions of length and time, dynamics a kinetic force field with dimensions mass, length and time. In inversely mapping from kinematics to kinetics, mass is a missing dimension. How then does information (e.g., TTC) specify dynamics (e.g., force) in producing a successful action? No satisfactory (mathematical) solution as yet. Stress environmental structures at expense of psychological structures. The environment is just a substitute for a schema. Need to replace metaphors with formal theory and logical deduction of its consequences (a). Metaphors disintegrate when parallels drawn with reality (b). Sometimes not clear if metaphors (figures of speech) or analogies (similarity along some dimension) involved.
Warren & Shaw (1985)
a. Reed (1987) b. Kugler et al. (1990)
a. Ben-Ze’ev (1981) b. Costall (1986)
Bremner (1993) a. Bogartz (1994) b. Black (1962)
1. Theoretically resolvable through quantitative models. 2. Empirically resolvable through appropriate perturbation experiments. 3. May require some reconciliation with more cognitive approaches (A3) and/or neurophysiological evidence commensurate with approach (C1). *TTC = Time-To-Contact
Causality and development
Dynamical systems adherents are mainly concerned with the pursuit of natural laws rather than rules. Laws are retrospective in seeking causal explanations for the origins of processes (“How or why did this happen?”) while rules are prospective accounts of how goals and intentions constrain action (“If you do that, then you can expect this”). In fact, laws and rules are subject to different sorts of constraints, the former being holonomic and the latter non-holonomic (Pattee 1987). Holonomic constraints are dynamical, ratedependent and informational boundary conditions that do not have a material (i.e., symbolic) representation in a system. As laws are not physically embodied, these constraints are law governed. When a system’s behaviour is governed by rules independently of laws, then non-holonomic constraints such as goals or intentions are operative. Such constraints (e.g., in the form of symbols) are materially embodied within the system (e.g., the output of a computer programme is constrained non-holonomically by a set of prescriptive codes). Treating perception as being law governed means that it has primacy over action, which may be more under the governance of non-holonomic constraints and thus rules (Kugler et al. 1990). If so, then during development, we must learn via rules how to act lawfully if are to realise what is intended. Thus, we are engaged in the learning laws by means of testing rules rather than subjected to the laws of learning. Stated another way, this implies that the developing child discovers affordances by exploring a delicate balance between laws (of perception) and rules (for actions), Theoretically, we can re-state the challenge as one of reconciling holonomic with non-holonomic constraints in arriving at a unified theory of development. Connectionist modelling has still to achieve such a reconciliation. If connectionist models, while promising, do not represent the royal road to reconciliation between dynamical systems and cognitive approaches, what does? One step in the right direction could be through a re-evaluation of what Gibson (1979) termed mediated perception. What is often not recognized is that he did not maintain an outright dismissal of mediated or indirect perception. Gibson (1966) made the distinction between perception based on information and perception mediated by instruments, pictures, language and other symbols. It is a distinction between knowledge of the environment (direct perception) and knowledge about the environment (indirect perception) – between law-based, specificational information and rule governed, indicational information. As Gibson frequently stressed, mediated perception is the primary means of transmitting culturally specified knowledge to others, of transforming tacit into explicit knowledge: what we know tacitly about the world comes from direct perception while explicit knowledge is garnered from de-
Brian Hopkins
scriptions, depictions and the like. For Gibson though, direct perception constitutes the fundamental form of cognitive contact with the environment while indirect perception may extend and re-organize it, but not alter it. It is this strict demarcation between perceptual processes that separates Gibson from mainstream psychology. The next step follows from the fact that Gibson’s demarcation is echoed in Vygotsky’s (1978) sociocultural interactional approach to development. And it is here that we may find fertile ground conducive to a reconciliation. Vygotsky used the term natural perception to denote a form of direct perception in which there is a dependence of action on the visual field and a circular fusing of perception with action (a standpoint also shared by other Russian theorists of action such as Bernstein 1967). Like Gibson, Vygotsky treated mediated perception as an extension of direct perception. For Vygotsky, however, the relationship between perception and action could be de-fused by auxiliary signs (rather than symbols) and in particular by intentions. Thus, when an individual encounters a novel or ambiguous situation, cognitive processes may ‘kick in’ to break the moment-to-moment guidance of action by direct perception so as to find a solution. Vygotsky went another step further than Gibson in proposing that as psychological processes become internalized, they then serve to embed perception into persistent functional systems which for Vygotsky were the causal agents of development. It is with this step that the seemingly unbridgeable gap between the dynamical and cognitive approaches re-appears. Despite this difference, the affinities between the theoretical standpoints of Gibson and Vygotsky hint at ways in which the opposing paradigms may be reconciled in a single theory on the development of action. On the one hand, cognitive approaches need to theoretically acknowledge a role for direct perception, perhaps along the lines of Cutting’s (1991) directed perception. On the other hand, dynamical systems approaches should account for the socially-specified forms of representation inherent to mediated perception and how they may promote change in cognitive processes. Having said that, the highest priority for dynamical systems theorists should be the establishment of mathematically formulated natural laws capable of revealing the processes and mechanisms of developmental change. As for the cognitive theorists, they will have to demonstrate that changes in its rule-based structures can be both modelled and analyzed as time-dependent, dynamical processes susceptible to holonomic constraints. Without such attempts, the two approaches remain irreconcilable ways of doing science (Turvey & Carello 1981), with the conse-
Causality and development
quence that theoretical advances in understanding developmental causality will be stymied.
. A concluding remark We started with Aristotle and we finish with him. Modern interpretations of causality originated with his four causes and led to material, efficient, formal and final causes becoming allied to genetic, elementaristic causal, systemic causal and teleological explanations, respectively. Only the first two managed to sustain scientific respectability in post-Newtonian physics. The demise of formal causes as natural phenomenona retarded the development of biology as an independent science. Attempts to revive it in post-Darwinian biology in the guise of prescriptive genetic programmes have been justly criticised (see Hopkins & Butterworth 1990). Nevertheless, formal causality lives on under the auspices of self-organization in which spontaneously occurring relationships between elements or processes of a system are capable of affecting it some time later. It is in this respect a variant of the general law of co-existence. If nothing else, the present contribution amounts to a plea for incorporating this law and the principles of self-organization more fully into further theorizing about causes in development.
References Ben-Ze’ev, A. (1981). J. J. Gibson and the ecological approach to perception. Studies in the History of the Philosophy of Science, 12, 107–139. Bernstein, N. A. (1967). The Coordination and Regulation of Movements. Oxford: Pergamon Press. Black, M. (1962). Models and Metaphors. Ithaca, NY: Cornell University Press. Bogartz, R. S. (1994). The future of dynamic systems models in developmental psychology in the light of the past. Journal of Experimental Child Psychology, 58, 289–319. Bremner, J. G. (1993). Motor abilities as causal agents in infant cognitive development. In G. J. P. Savelsbergh (Ed.), The Development of Coordination in Infancy (pp. 47–77). Amsterdam: North-Holland. Brunswick, E. (1956). Perception and the Representative Design of Psychological Experiments (2nd. ed.). Berkeley, CA: University of California Press. Clark, J. E., Truly, T. L., & Phillips, S. J. (1993). On the development of walking as a limit-cycle system. In L. B. Smith & E. Thelen (Eds.), A Dynamic Systems Approach to Development: Applications (pp. 71–93). Cambridge, MA: MIT Press.
Brian Hopkins
Cobb, L. & Zacks, S. (1985). Applications of catastrophe theory for statistical modeling in the biosciences. Journal of the American Statistical Association, 80, 793–802. Costall, A. (1986). The ‘psychologist’s fallacy’ in ecological realism. Teorie & Modelli, 3, 37– 46. Cutting, J. E. (1991). Four ways to reject directed perception. Ecological Psychology, 3, 25–34. Fitzpatrick, P., Schmidt, R. C., & Lockman, J. J. (1996). Development of clapping behavior. Child Development, 67, 2691–2708. Gibson, J. J. (1966). The Senses Considered as Perceptual Systems. Boston: Houghton-Mifflin. Gibson, J. J. (1979). The Ecological Approach to Visual Perception. Boston: Houghton-Mifflin. Gilmore, R. (1981). Catastrophe Theory for Scientists and Engineers. New York: Wiley. Groothuis, T. & Meeuwissen, G. (1992). The influence of testosterone on the development and fixation of the form of displays in two age classes of young black-headed gulls. Animal Behaviour, 43, 189–208. Haken, H. (1977). Synergetics: an Introduction. Berlin: Springer. Hopkins B. (2001). Understanding motor development: insights from dynamical systems perspectives. In A. F. Kalverboer & A. Gramsbergen (Eds.), Handbook on Brain and Behaviour in Human Development (pp. 591–620). Dordrecht: Kluwer. Hopkins B. (2002). Developmental disorders: an action-based account. In K. J. Connolly & J. Valsiner (Eds.), Handbook of Developmental Psychology (pp. 292–329). London: Sage. Hopkins, B. & Butterworth, G. (1990). Concepts of causality in explanations of development. In G. Butterworth & P. E. Bryant (Eds.), Causes of Development: Interdisciplinary Perspectives (pp. 3–32). Brighton: Harvester Press. Horsthemke, W. & Lefever, R. (1984). Noise-induced Transitions: Theory and Applications in Physics, Chemistry and Biology. Berlin: Springer. Kugler, P. N. (1986). A morphological perspective on the origin and evolution of movement patterns. In M. G. Wade & H. T. A. Whiting (Eds.), Motor Development in Children: Aspects of Coordination and Control (pp. 77–106). Dordrecht: Martinus Nijhoff. Kugler, P. N., Shaw, R. E, Vincente, K. J., & Kinsella-Shaw, J. (1990). Inquiry into intentional systems I: Issues in ecological physics. Psychological Research, 52, 98–121. Lorenz, E. M. (1963). Deterministic nonperiodic flow. Journal of Atmospheric Science, 20, 282–291. Maas van der, H. & Molenaar, P. C. M. (1992). Cognitive development: An application of catastrophe theory. Psychological Review, 99, 395–417. Mackie, J. L. (1974). The Cement of the Universe: A Study of Causation. Oxford: Oxford University Press. McLeod, R. W. & Ross, H. E. (1983). Optic flow and cognitive factors in time-to-collision estimates. Perception, 12, 417–423. Nagel, E. (1961). The Structure of Science: Problems in the Logic of Scientific Explanation. London: Routledge & Kegan Paul. Needham, J. (1959). A History of Embryology. Cambridge: Cambridge University Press. Neisser, U. (1977). Gibson’s ecological optics: consequences of a different stimulus description. Journal for the Theory of Social Behaviour, 7, 17–28. Nittman, J. & Stanley, H. E. (1986). Tip splitting without interfacial tension and dendritic growth patterns arising from molecular anisotropy. Nature, 321, 663–668. Oyama, S. (1985). The Ontogeny of Information. Cambridge: Cambridge University Press.
Causality and development
Pattee, H. H. (1987). Instabilities and information in biological self-organization. In F. E. Yates (Ed.), Self-organizing Systems: The Emergence of Order (pp. 325–338). New York: Plenum Press. Piaget, J. (1971). Biology and Knowledge: An Essay on the Relations between Organic Regulations and Cognitive Processes. Edinburgh: Edinburgh University Press. Pincus, S. M. (1991). Approximate entropy as a measure of system complexity. Proceedings of the National Academy of Science USA, 88, 2297–2301. Prigogine, I. (1980). From Being to Becoming: Time and Complexity in the Physical Sciences. San Francisco: Freeman. Reed, E. (1987). Why do things look as they do? The implications of James Gibson’s The Ecological Approach to Visual Perception. In A. Costall & A. Still (Eds.), Cognitive Psychology in Question (pp. 90–114). Brighton: Harvester Press. Robertson, S. S., Cohen, A. H., & Mayer-Kress, G. (1993). Behavioral chaos: beyond the metaphor. In L. B. Smith & E. Thelen (Eds.), A Dynamic Systems Theory Approach to Development: Applications (pp. 119–150). Cambridge, MA: MIT Press. Schiff, W. & Oldak, R. (1990). Accuracy of judging time to arrival: Effects of modality, trajectory and gender. Journal of Experimental Psychology: Human Perception and Performance, 16, 303–316. Schrödinger, E. (1944). What is Life? Cambridge: Cambridge University Press. Stent, G. S. (1981). Strength and weaknesses of the genetic approach to development of the nervous system. Annual Review of Neurosciences, 4, 163–194. Thom, R. (1975). Structural stability and morphogenesis: An outline of a general theory of models (French 1st ed., (1972)). Reading, MA: Benjamin. Tresilian, J. R. (1993). Four questions of time to contact: A critical examination of research on interceptive timing. Perception, 22, 653–680. Tresilian, J. R. (1994). Perceptual and motor processes in interceptive timing. Human Movement Science, 13, 335–373. Turvey, M. T. & Carello, C. (1981). Cognition: The view from ecological realism. Cognition, 10, 313–321. Valsiner, J. (1997). Culture and the Development of Children’s Action: a Theory of Human Development. New York: Wiley. Vygotsky, L. (1978). Mind in Society: the Development of Higher Psychological Processes. Cambridge, MA: Harvard University Press. Waddington, C. H. (1975). Evolution of an Evolutionist. Edinburgh: Edinburgh University Press. Warren, W. H., Young, D. S., & Lee, D. N. (1986). Visual control of step length during running over irregular terrain. Journal of Experimental Psychology: Human Perception and Performance, 12, 259–266. Warren, W. & Shaw, R. E. (1985). Events and encounters as unit of analysis for ecological psychology. In W. Warren & R. E. Shaw (Eds.), Persistence and Change (pp. 1–23). Hillsdale, NJ: Erlbaum. Werner, H. (1948). Comparative Psychology of Mental Development. New York: International University Press. Wilden, A. (1972). System and Structure: Essays in Communication and Exchange. London: Tavistock Publications.
Chapter 2
Perception of causality A dynamical analysis* Riccardo Luccio and Donata Milloni University of Florence
.
Phenomenology of causality
It is interesting to note that in the last few years there has been a renewed interest in a field of research with seemingly little more to say to psychology: perception of causality. This renewal of interest is mainly due to researchers who work on man-machine interfaces, like Bartram (1997a, 1997b), Ware et al. (2000), Kerzel and Hecht (2000). A striking fact is that the main reference for such research is the work of Albert Michotte, the Belgian experimental phenomenologist, who operated so close to Gestalt psychology some fifty years ago. Unquestionably, the psychological study of causality began with Michotte (1954, 1962) with his first experiments being carried out before the Second World War (Michotte 1941; see also Nuttin 1966, on the importance of Michotte in contemporary psychology). Similar work, done independently by Heider and Simmel (1944), demonstrated before Michotte (1950) that we can see intentions and emotions in animate objects. His series of experiments on this topic constitute one of the landmarks in experimental psychology of the twentieth century, and can be compared with only a few other instances of research that really changed our way of conceiving psychology such as Wertheimer’s work on the phi phenomenon and Pavlov’s classical conditioning. Using rotating disks and then moving beams of light on a screen, Michotte discovered two very important effects that are still of paramount importance in studying the perception of causality: the “launching” and “entraining” effects (lancement and entraînement). With the launching effect, we have two objects,
Riccardo Luccio and Donata Milloni
A and B (e.g., two squares, one red, the other black) on a screen. The subject fixates B that is at rest, but at a certain moment A begins to move towards B until it just touches B, then stops. At about this time (precise timing is of primary importance), B is made to move, and the subject has the inescapable perception of a direct causal relationship between the movement of the first and the movement of the second object such that B appears to be launched by A. In the entraining effect, we have substantially the same situation as for the launching. The difference is that when A touches B, it doesn’t stop but continues its run together with B. What the subject perceives is that the movement of B is caused by A in that A entrains B. As the reader has probably noticed, we don’t say that the subject “has the impression that. . .”, or similar expressions. What we do say is that the subject directly perceives the causal relations in both cases, as a primary component of its phenomenal experience. In Metzger (1954), causality has the character of the “encountered” in which the events are “unified for causation”. With minimal variations of the experimental conditions for both launching and entraining, it is possible to find a number of other highly interesting perceptual phenomena. It is informative to illustrate more completely the apparatus and the method used by Michotte in his experimental work. In this respect, we will cite in this respect many Italian authors as, in the 50s and 60s, Italian psychology made some very important contributions.
. The conditions for the perception of causality Michotte’s first apparatus constituted of grey disks mounted on a rotating support (later on, he also used projected lights). On the disks are two thick lines of different colour, both of which are positioned relative to the ground. The lines are curved, with varying degrees of curvature. In front of the subject, there is a screen with a horizontal slit that serves as porthole through which one can see the lines. Looking through the porthole, the lines are seen as objects that can stay still (when the line is concentric to the circumference of the rotating disk) or move at different speeds, according to the ray of curvature and the relative position of the line on the disk (see Fig. 1). First, we consider the launching effect and then the entraining effect. To perceive causality, it is necessary that the experimental conditions are such that the subject can simultaneously perceive two distinct objects, animated by a proper movement (and which as a first approximation we refer to as active and passive), according to a definite functional relationship that
Perception of causality
A
B
C
Figure 1. [from Michotte, 1954]
manifests itself as an unit-event of causal connection. One primary concept in this relationship is the “ray of action”. The ray of action of A on B (or of B on A) is the ensemble of the relative values (in terms of length, however expressed) of the path of A relative to B, such that it makes possible the perception of the phenomenal causality. So, for instance, if B runs along a very long path, then beyond a given length it will appear to be animated by a proper motion, without any functional relationship with the motion of A, and consequently without any phenomenal causality. Similarly, if we vary either the relative speed of the two objects, or the relative direction of their motion, we can obtain evidence that allows us to delimitate the amplitude of the stimulus field in which the perception of phenomenal causality appears. For instance, by keeping constant the speed of A (a typical speed being 40 cm/sec) and varying the speed of B up to a tenth of the previous one, the rays of action of A and B progressively decrease from 6 to about 1 cm. There are many other phenomena that can emerge. For example, if the approaching motion of A is very slow, the impression is that A adheres to or becomes unified with B. On the contrary, if it is very fast, we receive the impression of an impact, an impression that is increased if B moves fast towards A. Thus, one can see a sort of co-penetration of the two objects that appear to be endowed with a “living force”. The launching effect is independent of the colour of the objects, while the size and the shape play a more definite role. For the size, Brown’s (1931) law holds. For the shape, if its physiognomic qualities are consistent with the direction of the motion (e.g., A has the shape of an arrow), the effect (i.e., the ray of action) increases.
Riccardo Luccio and Donata Milloni
It is particularly interesting to note that the launching effect appears in the ray of action, even if A and B don’t possess the character of “things”. They could be fuzzy images or have a different consistency such that A appears as a wood-like ball, and B as a patch of light. We said before that timing is very important. This can be seen if we introduce a certain interval between the arrival of A and the beginning of the motion of B. Keeping constant a speed of 40 cm/sec, an interval lower than 100 msec doesn’t interfere; if the interval is higher than 100 msec, but lower than 130, the impression is of a delayed launching, as if B must overcome a certain inertia. With higher intervals, the effect disappears. We can also introduce a spatial interval, that is, stop the object A before it reaches B. The impression can be that of “impact at distance”, “launching without impact”, or the like. Anyway, the faster A is, the easier it is to still have the launching effect with a spatial interval. Also of interest is the observation of the effect of the relative direction and relative speed of motions. When, for instance, the motion of A is not collinear with that of B, the impression of the impact fails, but perception of causality could be maintained in that B appears as “freed” from a tie to A. If the motion of B is not a continuation of the motion of A, but there is an angle between the two directions, the effect weakens, and disappears with an angle of 90◦ . Regarding the relative speed, one can observe that the effect is greater when A is faster than B, and optimality is gained when the ratio is 3/4 to 1. When B is faster than A we have the “disjunction effect”, as then B would be made free from a tie, and its motion appears to be no longer passive, but active, even more so than the one of A. As we said above, the entraining effect occurs as soon as A reaches B and then they both move together. The clearest effect is obtained when A passes over B, and this objects begins its motion after a while. When the speed of the two objects together is slower than the speed of A alone, we have the impression of an impact at the moment when A reaches B. This impression is stronger the greater the difference between the two speeds. If A is slower than the two together, one perceives A approaching B “furtively”, and then as carrying it away. Related to the entraining effect is the “carrier effect” that is seen when a mobile that moves actively appears to carry another one, that moves passively. One obvious case is to have a moving object (e.g., a rectangle on which there is a coloured patch) stationary in relation to the rectangle, but moving with it. In this case one sees only one moving object with a coloured patch belonging to it.
Perception of causality
But, if during the motion, we move the coloured patch relative to the rectangle, the carrier effect appears in full force.
. The importance of Michotte’s work Why do we consider the work by Michotte and his associates to, a true “epistemological breakthrough”, a sort of scientific revolution? It is well known in the history of Western thinking that the prevailing interpretation of the psychological meaning of causality derives from Hume. According to Hume (1739–1740), the notion of causality (or necessary connection) between different events derives from the regularity of the succession of the phenomena, and it is grounded on the provision and expectation of the appearance of one of the events. Hume states that any two successive events in our experience are mutually independent, and that we can never perceive a genuine causal link. It is the temporary sequence of the two events that is sufficient to evoke the impression that the second event is caused by the first one (i.e., the first event is the causa efficiens of the second one). The process is associative in nature, and in psychology it has been firmly grounded in the theory of the unbewußter Schluss (unconscious inference) by Helmholtz (1878), which continues to be influential among many researchers on perception. Michotte’s demonstrations changed abruptly the way of considering the psychological notion of causality. As in the case of Wertheimer’s phi phenomenon (1912), the specificity of the situations studied by Michotte derived from the fact that the perceptual situation which is seen is not the result of the summing up of the events that compose it. Rather, it presents phenomenal features that are specific and original in their own right, and that cannot be reduced to those of the composing events singularly considered or to the “physical conditions” of presentation. Notice that the effect is still compelling if the structure of the experimental situation is well known to the subject (see Bozzi 1969, for a thorough discussion of this point, who has given in our opinion the best analysis of the problem, unfortunately available only in Italian). The “active” motion of A, and the “passive” one of B does not originate from some kind of evaluation, inference, attribution of meaning, or attribution of some peculiar phenomenal feature to the perceptual situation. This one is, instead, picked up immediately with its mere connotation of causality – phenomenal causality, to be precise. Such causality, in other words, is not thought, imagined, added, or “projected on. . .”. It is simply seen as such. Furthermore,
Riccardo Luccio and Donata Milloni
when the subject reflects critically upon the perceptual situation that he observes, he cannot avoid perceiving it with such characteristic features. To eliminate, or at least to weaken the phenomenon, the experimenter must change the experimental conditions (e.g., by introducing an interval between the events). Many researchers after Michotte have made his findings more precise, most in the spirit of Gestalt psychology. (We confine ourselves to refer to Crabbé 1967; Kanizsa & Metelli 1956, 1961; Ono 1960; Metelli & Passi-Tognazzo 1959– 1960; Minguzzi 1968.) The main findings of Michotte were always confirmed, and if anything further specified in some important respects (e.g., regarding the role of the physiognomic or tertiary qualities of the stimuli; see Kanizsa & Vicario 1968). The demonstration that an analytical setting by the observer can weaken or destroy the effect (Gemelli & Cappellini 1958; Beasley 1968) cannot be considered a counterdemonstration. An analytical setting often destroys most genuine perceptual effects, beginning with the celebrated “Wertheimer’s laws” (Wertheimer 1922). Another often-quoted counterexample is the fact that subjects often reject terms like launching or triggering to describe their experience (Boyle 1960; Beasley 1968). However, this demonstrates only a lack of apt terms in the everyday experience to describe what they perceive. More important in our opinion is that also very young infants also perceive causality (Leslie 1982; Leslie & Keeble 1987). Researchers like Spelke (1998) have advocated a top-down mechanism for such perception, stating that the infants reason or create expectancies about causal events. To do so, anyway, they must appeal to some folk-like physical principle, and so we simply displace the problem. It is interesting to note that recently Schlesinger and Barto (1999) have realized a mechanism that simulates causal perception in infants, which operates on the basis of a bottom-up mechanism, thus supporting an interpretation of the phenomenon in Michotte’s terms. According to Riedl (1992), our ability to perceive causality directly plays an important evolutionary adaptive role. Moreover, our perceptual systems are so structured to detect the presence of causes between events, even when they do not exist. Postulation of non-existent causes has no dangerous consequences, whereas missing them when they exist could lead to very serious outcomes. As Kerzel and Hecht (2000) point out, this tendency to attribute causal relationships to events is particularly clear in the evidence produced by research on naïve physics. If this all is true, it is then important to look at what contemporary cognitive psychology can say about this topic.
Perception of causality
. Causality and Gestalt problems in cognitive psychology So far, we have seen that the most important contributions to the psychological problem of causality in Michotte’s vein were given by Gestalt psychologists, or at least by researchers close to this tradition, But we all know pretty well that as a school, after the death of its main figures (Wertheimer, Köhler, Koffka, and more recently Metzger and Kanizsa), Gestalt psychology no longer exists. Nevertheless, and this is the lesson to be taken from the account of causality, its ideas are such that still today they cannot be ignored, at least not by students of perception and thinking. At the same time, the ideas of Gestalt psychologists were very often misunderstood, and this has given room to several mistakes and wrong interpretations that were late in being clarified. However, one has to emphasise that often such misunderstandings derive just from some theoretical incongruities, or from the ambiguities of some concepts by Gestalt psychologists themselves (Kanizsa & Luccio 1986; Kanizsa 1994). In our opinion (see Luccio 1983), one of the most promising developments in cognitive science that could prove itself able to deal with most Gestalt problems is the so-called non-linear dynamical systems theory. As Stadler and Kruse (1990) point out, there is a continuity between gestalt theorising on autonomous order formation (above all in Köhler’s formulation) and this currently fast developing theory in terms of self-organising systems operating far from equilibrium. In this respect, a prominent role was played in the last few decades by Hermann Haken (1983a, 1983b) with his theory of synergetics. Let us briefly consider this approach. According to synergetics, pattern formation can be described in terms of the evolution of state vectors of the following form: q(x, t) = (q1 , q2 , q3 , . . . , qn ) The evolution is described in terms of q˙ , the time derivative. Haken’s analysis leads to identifying a nonlinear function N according to which the temporal changes occur; this function depends on a control parameter. Internal and external fluctuations are instead described by a function F(t). However, the dynamics of the whole system is governed by order parameters alone. This means that if q˙ describes the system at the micro-level, the high-dimensional equation could be reduced at the macro-level to equations for the order parameter ξu alone: ˙ = N(ξ ˆ u ) + Fˆ ξ(t)
Riccardo Luccio and Donata Milloni
where Nˆ is a nonlinear function, Fˆ represents internal or external fluctuations, the dot • indicates the time derivative. The reduction of the behaviour of this complex system is described by Haken as an example of the slaving principle: near instability, the macroscopic behaviour of the system is dominated by few modes, which suffice for its description. When the control parameter changes, the old status is replaced by a new one, of the form q = q0 + veλt Notice that λ is an eigenvalue, which can assume positive or negative values. So, the solution for q from the starting point q0 can be decomposed into two parts: the first one for positive eigenvalues, the amplitude of which is the order parameter ξu and the second one for negative eigenvalues, the stable mode, the amplitude of which is the order parameter ξs : q (x, t) = q0 + ξu (t) vu (x) + ξs (t) vs (x) u
s
Figure 2 shows what happens in such a system. At the beginning, when the control parameter is under a critical value, there are fluctuations in the system that cause a mild increase in the order parameter, which tends to relax towards a stable or resting state (Fig. 2a): ξu tends to 0, and for the slaving principle also ξs tends to 0, so that q = q0 . When the control parameter exceeds the critical value, the first state is replaced by two possible ones. There is a breaking of the symmetry and a bifurcation in the two possible states and only one is chosen (Fig. 2b). A very important consequence is the phenomenon of hysteresis. When the system is in one of two possible states, as in Figure 2b, we notice a tendency to stay in the present state, that lasts more than when the system “relaxes” for the first time. This kind of evolution can be seen very easily in many kinds of physical, chemical, biological, and psychological systems. A classical example (indeed, the very starting point of Haken’s theorising) is the laser paradigm. When a laser-active material is excited (when the lasing begins), for instance by being irradiated with light, if the degree of excitation (the control parameter) exceeds a critical value we can ascertain that the atoms cooperate by emitting a coherent wave without any noise. With a greater excitation, this wave firstly breaks into ultra short laser pulses and a chaotic motion occurs later on. A change in the control parameter beyond a critical value determines a qualitative change of the system.
Perception of causality F(x)
F(x)
x–
0
x
x
(a) F(x)
x+
0
(b) 1.5 F(x)
1.5 F(x)
0.5 –1 0 1 –3 x –0.5 Control parameter under critical value
0
x 3 –3 –2 –1 Control parameter = critical value
–1.5
0 1
2 x 3 Control parameter > critical value
–1.5
(c) Figure 2.
The application of this model of a non-equilibrium phase transition is straightforward in behavioural problems, and mainly in perceptual problems. The situation of multistability, as Kruse and Stadler (1990) point out (see also Kruse & Stadler 1995; Kanizsa & Luccio 1995), is obviously a privileged field of research. In our group, we have made many experimental investigations in this area (see Chiorri 2002, for a review and a detailed theoretical discussion; as well as Kanizsa et al. 1994). More generally, however, this approach offers a very powerful tool for interpreting the self-organisation processes of the phenomenal field, in Köhler’s (1920) sense. To understand how it can be applied to the perception of causality, we need to discuss Spizzo’s effect.
. Spizzo’s effect We turn to a very interesting effect on the perception of causality, discovered in 1983 by Giovanni Spizzo, a former student of Paolo Bozzi, and which passed
Riccardo Luccio and Donata Milloni
unnoticed until now. Spizzo’s starting point was a set of observations made in various circumstances of everyday life such as opening and closing the hand in connection with the rhythmical appearance and disappearance of traffic lights. He noticed that in such circumstances we have the strong impression that the hand’s motion is responsible for the appearance of the signal, although we are aware of the absurdity of such a feeling. Spizzo decided to control experimentally the impressions he had encountered in such everyday situations. He presented rhythmically to his subjects luminous stimuli that lasted 500 msec with an inter-stimulus interval (ISI) of 500 msec. They had to touch the tip of the thumb with the tip of the index each time that the stimulus lit up, and to separate it from the finger during the ISI at an angle of about 45◦ . There were three situations in succession: (A) the subjects were at 50 cm from the source of light; (B) at 10 m, and (C) they touched the source of light with the hand. After some training, they had the following instructions: “Synchronize your action with the rhythmical lighting; when the light appears, your fingers must be just in touch, when it disappears they must be open. When you feel that you have reached a good synchronisation, describe what is the relationship you feel between the appearance of the light and the closing of your fingers.” The subjects gave their answers, which could be positive (P) or negative (N). The Ps were answers in which the subjects explicitly stated that they had perceived a causal relationship between their action and the rhythmical light, giving answers such as: “It looks like I’m making the light go on and off ”. They gave P answers in: 83% of the cases in situation A, 79% of those in situation B, and 91% in situation C. If the subjects provided an N answer, they were faced with more suggestive questions, like: “Do you feel that there is only a temporal coincidence, or rather there is a link, as if they were unified? That there exists a dependence of one event on the other? Describe the way you feel the connection”. There were some minor variations in the situation, but the most important one has been described above. Notice that the subjects realised that there was no relationship between their actions and the rhythmical lighting. According to Spizzo, one is forced to admit a “causal structure of the perceptual data” (p. 225). Another interesting point is that, contrary to Michotte’s evidence, the range of the temporal interval between action and lighting in Spizzo’s situations is from – 50 to 150 msec. In other words, the effect is still present when the light precedes the action.
Perception of causality
. Rhythmical patterns, dynamic systems and causality Bearing in mind Spizzo’s effect, let us turn to a field of research that was extensively investigated with the tools of the non-linear dynamical systems theory: the generation of rhythmical patterns and the recognition of rhythmical sequences (Kelso 1995). As we will see, this approach appears particularly promising in dealing with situations like the ones investigated by Spizzo. First, notice that almost every motor task requires harmonization of a manifold of actions. A classical example is walking, which requires the activation of such a great number of muscle groups that almost the whole body is involved. However, both the walkers and the observers do not notice such complexity. The cooperation of so many muscle groups leads to a coherent, steady, fluent and overt motion, which suggests that this situation is similar to the one considered above. Second, notice that, for bodily movements, the tendencies toward simplicity, regularity, and so forth, are particularly apparent. Some elegant experiments by Vogt (1988) on the reproduction of movements clearly indicate this tendency toward Prägnanz (as singularity), characterizing secondary cognitive processes (Kanizsa & Luccio 1986, 1990). Let us examine some examples. One can ask a subject to perform a rhythmic task with fingers (Kelso, Delcolle, & Schöner 1990), the pace being given by the rhythm of a metronome (see also the setting mentioned in Kelso & Pandya 1989). Another task is to perform a rhythmical task with the two arms, as the tempo is increasingly fast, with a sudden shift from out of phase to in phase. Yet another is to perform a bimanual rhythmical task in such a way that there is a phase difference between the two hands, the pace being given again by a metronome (Haken, Kelso, & Bunz 1985). In all of these cases, at a certain point we observe a sharp and spontaneous phase transition from a motor behaviour to another. To confine ourselves to the last example, which is the prototype for this kind of research (see Schöner & Kelso 1988a, 1988b, for a more complete theoretical treatment), subjects had to perform a coordinated rhythmical bimanual task at increasingly faster speeds. At a certain a critical value one can observe that a spontaneous change in the movement pattern: previously, the task could be performed either in phase or counterphase, but now only in phase.
Riccardo Luccio and Donata Milloni
The problem is to determine the order parameter, as the only variable of which the phase transition is a function. In the present case, the order parameter is the phase φ. The time derivative of the phase is dV + Qζ φ˙ = – dφ with
V φ = –a cos φ – b cos 2φ ,
with a and b as constants. ζ is a white noise with strength Q that accounts for random fluctuations. When a <1 4b the system has two attractors, with a relative phase equal to either 0 or π. Whereas, if a >1 4b there is only one stable phase with φ equal to 0. Kelso and Pandya (1989) have developed this model, by shifting from production to recognition of a rhythmical pattern. From this perspective, the two processes are one and the same thing. In our opinion, this model is very apt for investigating Spizzo’s effect. For this reason, we have planned a series of experiments investigations to collect data that will be analysed according to this theoretical framework. The data we have collected so far, concerning tasks of perceiving and producing rhythmical events, have shown a marked hysteresis effect with an area of perception ambiguity. In our experiment, the rhythm was represented by a flash effect on the display of a laptop with which the subjects had to get synchronized by a tapping behaviour. A good – or even excellent – synchronization proved to be, in a preliminary investigation, essential for the perception of a causal chain between the two events. The experiment consisted in rhythmical series, which differ for the frequency increase or decrease of flash tempo as well as for the in-phase or out-of-phase rhythmical start. Data collected for each observer have outlined a specific perception tendency: when the frequency of flash increases, the rhythm tends to be quickly lost, whereas when the frequency decreases the rhythm is more resistant to change. The area between the sigmoid curves of the 4 series compared manifests the hysteresis effect. A transition from a kind of perception from another has been showed with either in-phase or out-of-phase start. The perception system of causality/synchronization may be described through
Perception of causality
a synergetic model such as that presented by Haken, Kelso and Bunz (1985), by means of the potential equation V(φ) = –a cos(φ) – b cos(2φ). The system’s attractors would be related at two modes of the percept, where φ is the difference in phase between flashes and rhytmical tapping of the subjects.
. Conclusion The research on perception of causality by Albert Michotte was a true landmark in the history of experimental psychology, and its importance goes far beyond the domain of this discipline. In all handbooks of psychology, the results of such research are quoted as in a very positive light. Of course, the interest for this subject ebbed away with a decline of interest in experimental phenomenology and more generally for the Gestalt school. In the last few years, however, there is a renewal of interest in this topic, mainly from three different areas: man-machine interface, developmental psychology and naïve physics. In general, students in these areas consider the results so far obtained as conclusive, and simply exploit them for their own interests. In our opinion, this is a reductive way to consider the problem. In the field of perception of causality, there are still a lot of problems that deserve more accurate investigations. From another point of view, contemporary cognitive psychology offers many new conceptual tools that look very apt to deal with this kind of problem, and which simply did not exist at Michotte’s time. In our opinion, the best example is given by non-linear dynamical systems theory. One area in the study of causality that has been insufficiently investigated concerns the causal relationships between stimulus and action. Spizzo’s effect is a good instance of the problems one can meet in that area. Recent research, especially by Kelso, on rhythmical patterns within the framework of synergetics offers powerful conceptual tools to investigate these phenomena. It is now time to use them accordingly.
Note * The first part of this paper is largely inspired by an old scientific exchange, till now unpublished, that the senior author had in the ’70s with the late Professor Angelo Beretta (1938–1979). This paper is devoted to his memory.
Riccardo Luccio and Donata Milloni
References Bartram, L. (1997a). Can motion increase user interface bandwidth in complex systems? In IEEE Conference on Systems, Man and Cybernetics (1997, Orlando, 1686–1692. Los Alamitos, CA: IEEE Inc. Bartram, L. (1997b). Perceptual and interpretative properties of motion for information visualization. In Proceedings of New Paradigms in Information, Visualization and Manipulation, Las Vegas, November 13–14, 1997 (pp. 3–7). Las Vegas: ACM Press. Beasley, N. A. (1968). The extent of individual differences in the perception of causality. Canadian Journal of Psychology, 22, 399–407. Boyle, G. G. (1960). A contribution to the study of phenomenal causality. Quarterly Journal of Experimental Psychology, 12, 171–179. Bozzi, P. (1969). Unità, Identità, Causalità. Bologna: Cappelli. Brown, J. F. (1931). The visual perception of velocity. Psychologische Forschung, 14, 199–232. Chiorri, C. (2002). Metodologia per la Ricerca Sinergetica in Psicologia. Firenze: CUSL. Crabbé, G. (1967). Les Conditions d’une Perception de la Causalité. Monographies Françaises de Psychologie vol. 12. Paris. Gemelli, A. & Cappellini, A. (1958). The influence of subject’s attitude in perception. Acta Psychologica, 14, 12–23. Haken, H. (1983a). Synergetics: An Introduction. Berlin: Springer. Haken, H. (1983b). Advanced Synergetics: Instability Hierarchies of Self-Organizing Systems and Device. Berlin: Springer. Haken, H., Kelso, J. A. S., & Bunz, H. (1985). A theoretical model of phase transitions in human hand movements. Biological Cybernetics, 51, 347–356. Heider, F. & Simmel, M. (1944). An experimental study of apparent behavior. American Journal of Psychology, 57, 243–259. Helmholtz, H., von (1878). Die Tatsachen in der Wahrnehmung. Berlin: Hirschwald. Houssiadas, L. (1964). An exploratory study of the perception of causality. Journal of Experimental Psychology, Monographic Supplement, 36. Hume, D. (1739–1740). A Treatise on Human Nature. London: Noon. Kanizsa, G. (1994). Gestalt theory has been misinterpreted, but also has had some real conceptual difficulties. Philosophical Psychology, 7, 149–162. Kanizsa, G., Kruse, P., Luccio, R., & Stadler, M. (1994). Conditions of visibility of actual paths. Japanese Psychological Research, 36, 113–120. Kanizsa, G. & Luccio, R. (1986). Die Doppeldeutigkeiten der Prägnanz. Gestalt Theory, 8, 99–135. Kanizsa, G. & Luccio, R. (1990). The phenomenology of autonomous order formation in perception. In H. Haken & M. Stadler (Eds.), Synergetics of Cognition (pp. 186–200). Frankfurt: Springer. Kanizsa, G. & Luccio, R. (1995). Multistability as a research tool in experimental phenomenology. In M. Stadler & P. Kruse (Eds.), Ambiguity in Mind and Nature (pp. 47–68). Berlin: Springer. Kanizsa, G. & Metelli, F. (1956). Connessioni di tipo causale tra eventi percettivi: L’effetto attrazione e l’effetto lancio inverso. In Atti dell’XI Congresso degli Psicologi Italiani (pp. 1–10). Milano: Vita e Pensiero.
Perception of causality
Kanizsa, G. & Metelli, F. (1961). Recherches expérimentales sur la perception visuelle d’attraction. Journal de Psychologie Normale et Pathologique, 4, 385–420. Kanizsa, G. & Vicario, G. (1967). Caratteri espressivi ed intenzionali dei movimenti: la percezione della ‘reazione’. Rivista di Psicologia, 61, 319–321. Kanizsa, G. & Vicario, G. (1968). La percezione della reazione intenzionale. In G. Kanizsa & G. Vicario (Eds.), Ricerche sperimentali sulla percezione (pp. 69–126). Trieste: Edizioni Università degli Studi di Trieste. Kelso, J. A. S. (1995). Dynamic Patterns. The Self-Organization of Brain and Behavior. Cambridge, MA: MIT Press. Kelso, J. A. S., Delcolle, J., & Schöner, G. (1990). Action-perception as a pattern formation process. In M. Jeannerod (Ed.), Attention and Performance XIII (pp. 139–169). Hillsdale, NJ: Erbaum. Kelso, J. A. S. & Pandya, A. S. (1989). Dynamic pattern generation and recognition. In N. I. Badler, B. A. Barsky, & D. Zelter (Eds.), Making Them Move (pp. 171–190). San Mateo, CA: Morgan Kaufmann. Kerzel, D. & Hecht, H. (2000). Visual causality. In M. May & U. Oestermeier (Eds.), Interdisciplinary Perspectives on Causation (pp. 119–139). Bern: Bern Studies in the History and Philosophy of Science. Köhler, W. (1920). Die physischen Gestalten in Ruhe und im stationären Zustand. Braunschweig: Vieweg. Kruse, P. & Stadler, M. (1990). Stability and instability in cognive systems: Multistability, suggestion and psychosomatic interaction. In H. Haken & M. Stadler (Eds.), Synergetics of Cognition (pp. 201–217). Frankfurt: Springer. Kruse, P. & Stadler, M. (Eds.). (1995). Ambiguity in Mind and Nature. Berlin: Springer. Leslie, A. M. (1982). The perception of causality in infants. Perception, 11, 173–186. Leslie, A. M. & Keeble, S. (1987). Do six-month-old infants perceive causality? Cognition, 25, 265–288. Luccio, R. (1993). Gestalt problems in cognitive science. In V. Roberto (Ed.), Intelligent Perceptual Systems. New Directions in Computational Perception (pp. 2–19). Berlin: Springer. Luccio, R. (1994). Visual thinking. Stability and self-organization. In V. Cantoni (Ed.), Human and Machine Vision. Analogies and Divergencies (pp. 277–287). New York: Plenum Press. Metelli, F. & Passi Tognazzo, D. (1959–1960). Risultati di una serie di esperimenti sull’effetto di attrazione fenomenica. Atti dell’Istituto Veneto di Scienze, Lettere ed Arti, 118, 219– 230. Metzger, W. (1954). Psychologie. Darmstadt: Steinkopf. Michotte, A. (1941). La causalité physique est-elle un donné phénoménale? Tijdschriften van Philosophie, 3, 290–328. Michotte, A. (1950). The emotions considered as functional connections. In M. L. Reymert (Ed.), Feelings and Emotions (pp. 114–126). New York: McGraw-Hill. Michotte, A. (1954). La Perception de la Causalité. Louvain: Publications Universitaires. Michotte, A., Thinès, G., & Crabbé, G. (1962). Causalité, Permanence et Realité Phénoménales. Paris: Nauwelaerts.
Riccardo Luccio and Donata Milloni
Minguzzi, G. (1968). Sulla validità della distinzione tra percezione di nessi causali e percezione di dipendenze funzionali. In G. Kanizsa & G. Vicario (Eds.), Ricerche Sperimentali sulla Percezione (pp. 161–196). Trieste: Edizioni Università degli Studi di Trieste. Nuttin, J. (1966). La vie et l’oeuvre du Professeur Albert Michotte. Psychologica Belgica, 5, 87–115. Ono, A. (1960). An investigation on perception of causal relation. Tohoku Journal of Psychology, 2, 123–132. Schlesinger, M. & Barto, A. (1999). Optimal control methods for simulating the perception of causality in young infants. In M. Hahn & S. C. Stoness (Eds.), Proceedings of the 21st Annual Conference of the Cognitive Science Society (pp. 625–630). Mahwah, NJ: Lawrence Erlbaum Associates. Schöner, G. & Kelso, J. A. S. (1988a). A synergetic theory of environmentally-specified and learned patterns of movement coordination. I. Relative phase dynamics. Biological Cybernetics, 58, 71–80. Schöner, G. & Kelso, J. A. S. (1988b). A synergetic theory of environmentally-specified and learned patterns of movement coordination. II. Component oscillator dynamics. Biological Cybernetics, 58, 81–89. Spelke, E. S. (1998). Nativism, empiricism, and the origins of knowledge. Infant Behavior and Development, 21, 181–200. Spizzo, G. (1983). La regolare associazione come fonte di connessioni causali fenomeniche. Ricerche di Psicologia, 26, 221–230. Vogt, S. (1988). Einige gestaltpsychologische Aspekte der zeitlichen Organisation zyklischer Bewegungsabläufe. Bremer Beiträge zur Psychologie, 77. Ware, C., Neufeld, E., & Bartram, L. (2000). Visualizing causal relations. In D. Wills & D. Keim (Eds.), Proceedings of IEEE Information Visualization, Late Breaking Hot Topics (pp. 39–42). San Francisco, CA: IEEE Inc. Wertheimer, M. (1912). Experimentelle Studien über das Sehen von Bewegung. Zeitschrift für Psychologie, 62, 371–394. Wertheimer, M. (1922). Untersuchungen zur Lehre von der Gestalt I. Psychologische Forschung, 1, 47–58.
Chapter 3
Embodiment and the philosophy of mind* Andy Clark Washington University
.
Introduction: The rediscovery of the body and of the world
Cognitive Science is in some sense the science of the mind. But an increasingly influential theme, in recent years, has been the role of the physical body, and of the local environment, in promoting adaptive success. No right-minded Cognitive Scientist, to be sure, ever claimed that body and world were completely irrelevant to the understanding of mind. But there was, nonetheless, an unmistakable tendency to marginalize such factors: to dwell on inner complexity whilst simplifying or ignoring the complex inner-outer interplays that characterize the bulk of basic biological problem solving.1 This tendency was expressed in, for example, the development of planning algorithms that treated real-world action as merely a way of implementing solutions arrived at by pure cognition (more recent work, by contrast, allows such actions to play important computational and problem solving roles2 ). It was also expressed in David Marr’s3 depiction of the task of vision as the construction of a detailed threedimensional image of the visual scene. For possession of such a rich inner model effectively allows the system to “throw away” the world and to focus current computational activity in the inner model alone.4 More generally, the whole vision of cognition as inner operations on internal world models reflects an explanatory strategy which might reasonably be dubbed ‘isolationism’:5 Isolationism: The world is (just) a source of inputs and an arena for outputs; the body is just an organ for receiving inputs and effecting outputs (actions); the task of early processing is to render the inputs as an inner world-model of sufficient thickness to allow the bulk of problem solving activity to be defined over the inner model alone.
Andy Clark
Isolationism, it is fair to say, is in increasing disrepute. But the precise shape of an alternative paradigm remains unclear. Anti-isolationist assertions range from the relatively innocent insistence that we won’t achieve a balanced vision of what the brain does until we pay more heed to the complex roles of body and world, to the self-consciously revolutionary accusation that mind itself is not, after all, a special realm populated by internal models and representations so much as an inextricably interwoven system, incorporating elements of brain, body and world – a system which resists informative analysis in terms of the old notions of model, representation and computation.6 The most radical antiisolationist vision thus depicts human beings as a species of (so-called) postCartesian agents.7 The post-Cartesian agent is a locus of knowledge, acts for reasons and has beliefs and desires. Yet she harbors no internal representations and resists analysis in terms of any cognitively important distinctions between “inner” and “outer” processes, between perception, cognition and action, or between mind, body, and world. I shall argue that the full post-Cartesian vision is unconvincing and that a key move in the argument (a move I dub the “Cognitive-to-Coping Shift”) is both dialectically suspect and empirically unsound. More positively, I shall suggest that a weaker anti-isolationist stance still requires some deep revisions in our understanding of the inner vehicles, their contents and the adaptive roles of internal representation and inner world models. The foundational and conceptual challenges are thus real enough, even when stripped of their radical post-Cartesian trimmings.
. Inner symbol flight The outright rejection of the notion of internal representation is usefully seen as the extreme limiting case of a (generally admirable) process of inner symbol flight. This process involves the progressive rejection of more and more of the apparatus and assumptions associated with the vision of cognition as the manipulation of chunky inner symbols. According to this simple (and historically important) vision, semantically sensible transitions between mental states are explained in terms of syntactically constrained transitions between inner symbol strings. These symbol strings contained discrete elements corresponding rather closely to the semantic elements identified in sentential descriptions of the relevant mental states. Thus, the thought that John loves Mary is realized as a complex inner symbol string that incorporates distinct and independently manipulable elements standing for <John> and <Mary>.8
Embodiment and the philosophy of mind
This vision of simple inner symbolic atoms (unstructured base items corresponding rather closely to familiar concepts and relations enshrined in daily discourse) was challenged by the development of distributed connectionist9 models. The “sentential paradigm”10 was replaced, in this research, by a vision of internal representations as distributed patterns of activity across a whole array of simple processing units. Such distributed patterns were allowed to overlap in semantically significant ways, giving rise to a variety of computationally significant side-effects including free generalization, damage-resistance, etc.11 More recently still, we have witnessed increased attention to the temporal dynamics of the inner representational vehicles. The use of e.g., single recurrent neural networks12 allows information to be encoded not just in instantaneous patterns of activity but in temporally extended processing trajectories. In these networks, much of the information-processing power resides in the way a current state allows or restricts future change and evolution. The progression has thus been from a view of simple, atomistic inner symbols, to a notion of spatially distributed patterns, to a notion of spatially and temporally distrusted patterns. The inner vehicles of content, courtesy of this progression, have come to look less like simple inner states and more like complex inner processes. The metamorphosis, moreover, is probably still incomplete. Some rather plausible next steps include seeing the inner vehicles as multiply functional and seeing the inner architecture as dynamically reconfigurable. Multiple functionality would mean that one and the same inner resource may play a variety of content-bearing roles13 (perhaps varying in accordance with local context). Dynamic reconfigurability would mean that the inner structures are themselves subject to rapid change and reorganization, as when the release of a chemical neuromodulator causes two neural networks to temporarily “fuse” and behave as one. The moral, then, is that our understanding of the nature of the (putative) inner vehicles of content is in a state of extreme flux, characterized by a rapid flight from the initial image of static, chunky unstructured inner symbols. This flight has, in addition a more content-related aspect. For at the same time as the inner vehicles become more complex, so the characteristic contents have become more partial and fragmentary. This is because the emphasis has shifted from isolationist forms of problem solving towards iterated series of agent-environment interactions. This shift lies at the very heart of the agenda of a more embodied and environmentally embedded approach to cognitive science and is nicely exemplified by recent work in the field known as “Animate Vision”.14
Andy Clark
Recall Marr’s depiction15 of the task of vision. The task, according to Marr is to construct a rich inner model of the three dimensional visual scene on the basis of the available (two dimensional) input information. Work in “Animate Vision”, by contrast, depicts the task as, simply, the use of visual strategies to control behavior, in real-world contexts, at as low a computational cost as possible. To this end, animate vision avails itself of three central ploys. 1. The use of task-specific cues and shortcuts. 2. The use of body-centered (egocentric) strategies. 3. The use of repeated environmental interactions. Task-specific cues and shortcuts include, for example, the use of “personalized” idiosyncratic strategies such as searching for bright yellow (a cheap, easy visual cue) when searching for my coffee cup (which is canary yellow). Egocentric strategies include the use of so-called “deictic pointers” (see below). And repeated environmental interactions allow us, for example, to visit and re-visit different aspects of a visual scene retrieving specific information only as and when required. The case of deictic pointers merits a longer look. A pointer in classical Artificial Intelligence, is an inner state that can function in self-contained computational routines but which can also point to other data structures.16 This pointing allows the retrieval, when required, of more detailed information, and the effective binding of certain items of information to others. Such binding can be temporary, as when we bind certain features (e.g., bright yellow) to certain current visual locations (top left of visual field). Deictic pointers, as Ballard et al. describe them, are physical actions – such as foveating a certain location – which play a similar kind of functional role. The very act of foveation, it is suggested, may be used to temporarily bind color to location, or to direct a reaching motion to a target. A further example is the binding of a reaching-and-grasping routine to a target object using what is informally called a “do-it-where-I’m-looking” strategy. Here, the system is set up so that the grasping motion is directed to the fixated visual location. In all these cases: The external world is analogous to computer memory. When fixating a location, the neurons that are linked to the fovea refer to information computed from that location. Changing gaze is analogous to changing the memory reference in a silicon computer. (Ballard, Hayhoe, Pook, & Rao 1997: 725)
The general thrust of the animate vision research, then, is that bodily actions (such as saccadic eye motions) play vital computational roles, and that re-
Embodiment and the philosophy of mind
peated agent-environment interactions obviate much of the need to create allpurpose, detailed internal world models. Instead, we visit and re-visit different aspects of the scene as and when required, allowing the world to function as its own best model. The research program is thus staunchly anti-isolationist. But it is not by any means “post-Cartesian” – it does not reject the ideas of internal models and representations, so much as reconfigure them in a sparser and more interactive image. We thus read of “inner databases” that associate objects (e.g., my car keys) and locations (on the kitchen table), of “internal featured representations”, “indexical representations”, and so on. What is being rejected is not the notion of inner content-bearing states per se, but rather the much stronger notion of rich, memory-intensive, all-purpose forms of internal representation. A similar conclusion is suggested by work17 in real-world robotic navigation in which knowledge of location is encoded as a sensory-motor routine that will actually move the robot from its present position to the desired spot. In this way, the inner map is itself the controller of the appropriate action. There is no need for a further system to access the “map” and to plan a route. Instead, the knowledge is at once both descriptive and prescriptive18 – a dual nature that affords great economies both in terms of response-time and computational effort. The crucial distinction, it seems to me, is thus not between representational and non-representational solutions so much as between action-neutral forms of internal representation (which may increase flexibility but require additional computational work to specify a behavioral response) and action-oriented forms (which build the response into the representation itself). The best work in animate vision and real-world robotics, I claim, suggests at most that the use of truly action-neutral internal representations may be rather rare in biological cognition (my own view, discussed briefly in Section 5 below, is that the use of such representations coincides, rather exactly, with the possession of a rich public language). Such conclusions are radical and challenging. But they fall well short of a full post-Cartesian rejection of inner models and representations. What considerations might drive us to question the idea of inner content-bearers tout court?
. Radical interactionism One leading anti-representationalist argument19 turns on the presence of dense, reciprocal causal exchanges uniting agent and environment in a com-
Andy Clark
plex web of mutual influence. Under such conditions, it is argued, the kind of de-composition and analysis that works so well in the case of e.g., a contemporary computer program simply gets no foothold. The problem (it is suggested) is that the notion of x representing y is too one-way and too simplistic to do justice to cases in which x is continuously affecting and being affected by y and vice versa. Yet typical agent-environment interactions often present just such a complex and circular causal profile. Consider ballroom dancing. As you dance, your motions (if you are a good dancer!) are both continuously influenced by and are an influenced upon, those of your partner: the two sets of motions co-evolve in a highly interdetermined way. Nor is the presence of two human agents essential to the phenomenon. The same holds true as you windsurf: you constantly affect and are affected by the set of your rig. Van Gelder20 makes the same point using the example of the Watt (or centrifugal) governor – a device which maintains a steam engine at a steady speed by both affecting and being affected by the engine speed. Such episodes of mutual influence were much discussed both in early cybernetics21 and in the work of the French phenomenologist Maurice Merleau-Ponty.22 Where such continuous, dense, circular causal influence obtains, it is argued, the tools of representational (and computational) analysis run aground. The idea of explaining the shape of the on-going agent-environment interaction by depicting an inner state as representing an outer one is coarse and distortive. Instead, the elements co-evolve in a mathematically precise way that is best captured (so the argument goes) by the use of coupled differential equations in which the current value of certain internal variables appear as parameter setting in the evolution equation for the external system and vice versa.23 Fortunately, the details of such a dynamical systems model are unimportant for present purposes.24 What matters is rather the general shape of the argument. Van Gelder25 puts it well: The internal operation of a system interacting with an external world can be so subtle and complex as to defy description in representational terms. (ibid.: 381)
Before responding to this argument, it is worth pausing to clarify the challenge. For what is at issue is not the status of certain systems (ourselves, for example) as representers. That is a given. We surely do represent our world, our past, our possible futures, our absent friends and so on. We think of these things and states of affairs and thus represent them to ourselves. What is not a given (and what is at issue here) is that we use internal representations to do so. The point
Embodiment and the philosophy of mind
(and I think it is a good one) is that the notion that cognition involves internal representations (and computations defined over them) is meant to be not a simple rehearsal of the fact that we are thinkers, but a substantial and explanatorily potent empirical hypothesis: the kind of thing that could indeed turn out to be false. The claim, to a first approximation, is that there are distinct, identifiable inner resources whose systemic or functional role is to stand in for specific features or states of affairs. This notion of internal stand-ins is, however, itself ambiguous. It is ambiguous26 between a weak notion, in which x “stands” for y if x is an inner resource that (a) carries information about y and (b) is used to control behavior, and a strong notion in which the inner resource must be capable of functioning as a genuine surrogate, i.e., be capable of systematically controlling appropriate behavior even if y is absent or non-existent. A neural population27 closely keyed to bodily orientation and used to control skilled action may thus be counted as a kind of weak stand-in. And even this weakly representational gloss tells us something useful about the purpose of the neuronal population and may shed light on larger scale systemic organization (we may see which other neuronal populations access that specific body of information and hence gain insights into their roles). But such a population, though it engages in informationbased control of action, need not be capable of driving appropriate actions in the absence of the (weakly represented) state of affairs. It is this capacity to act as an inner surrogate in the absence of the target environment feature that, I suggest, characterizes the strongest and most conceptually unequivocal cases of internal representation.28 The problem that I wish to highlight should now be apparent. The entire argument concerning the circular causal complexity of agent-environment interactions is vitiated, I believe, by its failure to engage the issue of strong representation. All the examples share (and must share) a certain problematic feature, viz., they are all cases in which the target behavior is continuously driven by the relevant environmental parameter. Yet a major motivation for the positing of internal representations in the first place is to explain our puzzling capacity to go beyond tightly coupled agent-world interactions and to coordinate our activities and choices with the distal, the possible and the non-existent. The notion of internal representation is thus grounded in the notion of real inner surrogates and is merely extended (perhaps problematically) to the case of (merely) information-bearing inner states. This helps to explain why the best cases for the argument-from-continuous-reciprocal-causation may strike us as rather poor example of traditionally cognitive phenomena. For they depend crucially on the constant presence of the relevant environmental factors and
Andy Clark
thus do not strike us as especially “representation-hungry”29 scenarios in the first place. Properly representation-hungry scenarios would be planning next year’s vacation, using mental imagery to count the number of windows in your old house, doing mental arithmetic, dreaming, etc. The dialectical situation is, however, rather delicate. For the anti-representationalist may now reply that the point of her argument, in part, is to suggest that these traditional cases (of what might be termed “environmentally de-coupled reason”) are in fact empirically marginal and that the bulk of intelligent response displays precisely the richly interactive profile the argument highlights. Environmentally de-coupled reason, it is claimed, is at best a tip of the iceberg phenomenon. What is being promoted is thus a shift of emphasis away from off-line cogitation and onto real-time interactive engagement30 – a kind of “cognitive-to-coping” shift. This shift in emphasis is welcome. From both an evolutionary and a developmental31 point of view, real-world real-time responsiveness is clearly in some sense primary. But as part of any general anti-representationalist argument, the move is both dialectically suspect and empirically flawed. The problem is that the recognition that the richly interactive case is biologically basic is, as we shall see, perfectly comparable with the claim that “off-line” environmentally de-coupled reason is not the mere tip of the adaptive iceberg. The way to forge a genuinely cognitive science of embodied, environmentally embedded agency is, I believe, to look harder for the bridges between densely coupled and strongly representationally mediated forms of adaptive success. Such bridges are at the heart of the conciliatory position I dub “minimal Cartesianism”, and to which I now turn.
. Minimal Cartesianism Minimal Cartesianism seeks to locate the roots of pure contemplative reason in the kinds of richly interactive settings emphasized in recent work on embodied cognition (Sections 2 and 3 above). Thus consider the phenomenon of skilled reaching.32 Smooth, skilled reaching involves the use of proprioceptive feedback – signals that tell the brain how the arm is oriented in space. But the timing of these signals poses a problem. The minimal delay between the onset and the use of such information looks to be between 200 and 500 milliseconds.33 Yet we make essential trajectory corrections, that look to be governed by such feedback, within the first 70 milliseconds34 of reaching. How does nature turn the trick?
Embodiment and the philosophy of mind
This problem of requiring feedback before it is practically available crops up in industry too: in chemical plants, bioreactors and so forth.35 One common solution, in these cases, is to add a “forward model” or “emulator” into the systems. This is a circuit that takes as input a specification of both the previous state of the system and the commands just issued, and that gives as output a prediction of the feedback that should later arrive. The emulator thus generates a kind of mock feedback signal available substantially in advance of the real thing. Nature, it now seems, may deploy much the same strategy. There is a growing body of neuroscientific evidence36 that suggests that neural circuitry spanning the cortico-spinal tract, the red nucleus, the inferior olive, and the cerebellar cortex may be playing just such a role. Such circuitry looks to take a copy of the afferent motor command and to output a fast prediction of the feedback later due to arrive by the slow 200–500 millisecond route. The same trick has been replicated in a variety of neural network37 models. What matters for our purposes, however, is an additional conjecture. It is the conjecture38 that the biological emulator circuit plays a dual role. This dual role involves first the fine-tuning of on-line reaching (the normal case, in which the emulator circuit acts as an aid to smooth real-time reaching), and second, the production of visual-motor imagery allowing the off-line mental rehearsal of motor routines. In the latter case, the emulator circuit is running alone, de-coupled from the real-world action system. Such an additional role for the very same emulator circuitry implicated in real-time reaching is independently plausible and helps explain some otherwise puzzling results. These include the robust finding that mental rehearsal can actually improve sports skills and the activity of the cerebellum (generally thought of as a motor area) during mental imagery.39 Motor emulation circuitry, if this is correct, is both an aid to fluent, realworld action and a support for independent, environmentally-decoupled mental rehearsal. It is thus a minimally Cartesian mental tool, but one that is parasitic upon adaptations closely geared to the promotion of real-time agentenvironment interactions. As a result, the kinds of content that get represented are closely tied to the bio-mechanics and action-taking profile of the agent. And the form of the inner vehicles of content is left quite open – such vehicles may involved complex temporally extended processes, as indicated in Section 2. Given such a profile, we can see why isolationist methodologies and assumptions (see Section 1) are inadequate even in the case of certain kinds of environmentally decoupled cognitive skills. For such decoupled skills may nonetheless remain action-oriented in the sense just described, with both con-
Andy Clark
tents and mechanisms being profoundly informed by the agents real-time interactive purposes. Once again, however, the failure of isolationism should not be seen as an invitation to scepticism about representation and inner models. In the emulator case, it is clearly apparent that we are dealing with identifiable circuitry whose functional role is to model specific aspects of extra-neural (in this case bodily) reality. Yet this inner modeling is of a type that is perfectly continuous with the various morals and emphases suggested by the actionoriented research discussed earlier. The conciliatory position that I favor thus involves combining the embodied stress on real-world, real-time action with a search for the biologically basic roots of more decoupled forms of thought and problem solving. It is only by confronting this latter class of cases that representationalism can be given a fair trial.
. Scaling, rationality and complexity Minimal Cartesianism aims to build bridges between the recent emphasis on richly interactive tasks and the more traditionally cognitive focus on decoupled reason. To that end it stresses the use of multiple, partial, action-oriented inner models and of deictic, idiosyncratic and action-oriented internal representations. The compelling question, at this point becomes whether we must posit, in addition inner resources that have much of the character of classical symbolic models. That is to say, can we hope to explain the full gamut of human cognition without at some point reinventing the original image of context-neutral, rich, action-independent, highly manipulable inner symbolic structures? It is, I think, worth a try! More accurately, what may be worth a try is an approach which does not eschew altogether the use of such richly structured, action-neutral encodings but which ties them very closely to our experiences with public language and other externalizable and interpersonally shareable symbol systems. Complex human cognition would then be depicted as the fecund interface between a variety of action-oriented internal resources and a larger web of linguistic competence and cultural tools and practices. This larger web (or scaffolding) acts so as to substantially alter the computational spaces that can be explored by our form of basic, on-board biological reason. A classic example40 is the use of pen and paper to expand our basic mathematical horizons, allowing us to use an iterated sequence of simple inner computations (7 × 7, 4 × 4) to solve more complex problems (such as 777 × 444). Public language, I elsewhere argue, plays a wide variety of similar roles.41 The mere act of labeling, as Dan Dennett42 points out, affords great
Embodiment and the philosophy of mind
economies of search and classification. While the capacity for linguistic rehearsal may, according to Ray Jackendoff,43 enable us to attend to the details of our own thoughts and thus open up new possibilities of reflection and analysis.44 External artifacts and social organizations likewise alter and transform the tasks that individual brains need to perform. In this vein, Ed Hutchins45 offers a persuasive account of ship navigation in which it is the overall system comprising multiple brains, bodies and instruments that “solves” the problem. Each crew member, acting within this larger nexus, merely monitors and responds to certain conditions, and alters a few aspects of the shared work space so as to support the activity of the others. The whole process constitutes an environmentally extended computational flow in which props and artifacts (such as nautical slide rules) also play a major role. The Minimal Cartesian treatment of basic biological reason may (just may) thus scale-up so as to illuminate the full panoply of human thought and reason. But it will only do so if we take the issue of external scaffolding very seriously and recognize the special virtues of public language: the one actionneutral symbolic code we know ourselves to actually possess. An implication of this approach to the scaling problem is that we will need, at times, to study these larger systems (of multiple communicating brains and artifacts) as organized wholes and to recognize extended computational processes spanning the boundaries between brain, body and world. Such assertions can easily be mistaken for antipathy towards the study of the inner resources and processes. But the real challenge, once again, is to interlock the two approaches and thus to relocate individual human reason in its proper ecological niche. All this raises questions about the notion of human rationality itself. Isolationist cognitive science tended to depict rationality in terms of semantically apt transitions between inner mental states. Turing’s achievement, as repeatedly stressed by Jerry Fodor46 , was to show how such transitions could be supported by a mechanical process. The environmentally extended approach just mooted does not reject that account. It may (and should) incorporate Turing’s central idea of inner processes whose syntactic47 properties preserve semantic relations. But this will be just part of a much larger theory that allows rational behavior to supervene on wider webs of structure involving other agents and aspects of the local environment. Finally, a worry about complexity. Even if the general project sketched in this paper proves attractive (the project of bridging between interactionbased models and environmentally decoupled, representation-involving, reason), there remains a worry concerning the potential complexity of the inner vehicles of content. The worry, touched on in Section 2, is that the inner ve-
Andy Clark
hicles may be too spatially and temporally complex to effectively isolate. The worry gains force from recent demonstrations of the role of recurrent connections48 in modulating the information-processing profile of neuronal populations49 and from the sheer difficulty of assigning specific content bearing roles to tracts of neural machinery. These complexities and difficulties can lead to a subtly different kind of scepticism in which it is the complexity of the inner story itself (rather than the inner-outer interactions) that is supposed to make trouble for the representational approach. The issues here are more straightforwardly empirical and it is impossible, given the current state of research, to make any firm predictions. But one clear possibility is that new analytic tools may yet provide the means to identify functionally important patterns of activity. Dynamical systems analyses, of the kind sometimes promoted as an alternative to the representational approach, may in fact serve the representationalist’s cause against the backdrop of burgeoning spatial and temporal complexity. This possibility is clearly noted by van Gelder50 himself, who allows that “an exciting feature of the dynamical approach is that it offers opportunities for dramatically re-conceiving the nature of representation in cognitive systems”. Internal representations, then, may be realized not as simple inner states but as dynamical patterns of just about any conceivable kind. Such patterns may, in addition, be transient entities that form only in response to the details of current context. We thus better appreciate the limits of the inner vehicle metaphor itself. Such vehicles need be neither simple nor persisting51 in order to play a representational role. Van Gelder’s concession is important. He does not take himself to have shown that there are no internal representations. Just that there might not be any, and that even if there are they may take a very different form to the one we once expected. I have tried to show that the specific skeptical considerations he advances (concerning the potential complexity of agent-environment interactions) fail (and must fail) to make contact with the core representationalist hypothesis, which is grounded in our capacities for environmentally decoupled reason. The revisionary representationalist option, however, is both appealing and increasingly in evidence in actual cognitive scientific applications. In sum, our vision of basic biological reason is rapidly changing. There is a growing emphasis on the computational economies afforded by real-world action and a growing appreciation of the way larger structures (of agent and artifacts) both scaffold and transform the shape of individual reason. These twin forces converge on a rather more minimalist account of individual cognitive processing – an account that tends to eschew rich, all-purpose internal models and sentential forms of internal representations. Such minimalism, however,
Embodiment and the philosophy of mind
has its limits. Despite some ambitious arguments, there is currently no reason to doubt the guiding vision of individual agents as loci of internal representations and users of a variety of inner models. Rather than opposing representationalism against interactive dynamics, we should be embracing a broader vision of the inner representational vehicles and seeking the deeper continuities between rich interactive strategies and off-line, environmentally de-coupled, reason. The reward may be a better vision of reflective agency itself.
Notes * This paper first appeared in O’Hear, A. (Ed.). (1998). Current Issues in Philosophy of Mind, Royal Institute of Philosophy Supplement 43 (pp. 35–52). Cambridge: Cambridge University Press. The present version contains minor changes. . Some exceptions to this trend include the work of Gibson (1979) and Merleau-Ponty (1942). Work in Animate Vision and ecological optics (see Section 2 below) is clearly influenced by Gibsonian ideas, while more philosophical treatments, such as Varela, Thompson and Rosch (1991), explicitly acknowledge Merleau-Ponty. There is a brief discussion of these historical roots in Clark (1997: Ch. 8). . See e.g., Agre, Rosenschein (1996), Kirsh and Maglio (1995), Hutchins (1995). . See Marr (1982). . See Ballard (1991), Churchland, Ramachandran and Sejnowski (1994). . Roboticists refer to this isolationist vision as the (increasingly discredited) idea of a simple Sense-Think-Act Cycle. See e.g., Malcolm, Smithers and Hallam (1989). . See Haugeland (1995), van Gelder (1995), Port and Gelder (1995), Thelen and Smith, (1994) Varela, Thompson, and Rosch (1991). . This vision is clearly contemplated in e.g., Haugeland, (1995) and van Gelder (1995), though both authors recognize the large space of intermediate possibilities. The term “PostCartesian Agent” is from van Gelder (1995: 381). See also Haugeland (1995: 36), Thelen and Smith (1994: 338), Port and van Gelder (1995: IX). . See Fodor and Pylyshyn (1988: 13). . See Rumelhart, McClelland (1986). . Churchland (1986). . See Clark (1989) for discussion. . Elman (1991). . For some hints of such content-sensitive complexity, see Knierim and VanEssen (1992). . Ballard (1991). . Marr (1982). . See e.g., Pylyshyn (1987).
Andy Clark . Mataric (1991). This work is further discussed in Clark (1997: Ch. 2). . For more on this theme, see Millikan (1996). . This argument is the centerpiece of van Gelder (1995) and is also visible in van Gelder and Port (1995a), Thelen and Smith (1994), Varela, Thompson and Rosch (1991). Other anti-representationalist arguments are considered in Clark (1997). . van Gelder (1995). This example is treated in detail in Clark and Toribio (1994). . For example, Ashby (1956). . Merleau-Ponty (1942). . For an accessible introduction to these dynamical approaches, see Kelso (1995). A classic text is Abraham and Shaw (1992). . For a fuller discussion, see Clark (1997: Ch. 5, 6, and 8). . van Gelder (1995). . See Clark and Grush (1997). . For example, the posterior parietal neuronal population in the rat which encodes information about which way the rat’s head is facing and which is exploited in radial maze running – see McNaughton and Nadel (1990). . Bogdan (1988) makes the same point. See also Smith (1996). . See Clark and Toribio (1994). . This move is explicitly made in Haugeland (1995) and is also clearly in evidence in van Gelder and Port (1995a). . See Thelen and Smith (1994). . I borrow this case from Grush (1995). An extended treatment is available in Clark and Grush (1997). . This figure is established by, for example, using artificial vibrators strapped to the tendons to disrupt proprioceptive signals from the muscle spindles, and timing the gap between such disruptive input and alterations to the arm motion itself, see Kelso (1995). . See Grush (1995). . See Grush (1995) for a review. . See Ito (1984), Kawato, Furukawa and Suzuki (1987). . Kawato (1990). . Grush (1995). . Decety and Grezes (1999). . Rumelhart and McClelland (1986). . Clark (in press). . Dennett (1995). . Jackendoff (1996). . Ibidem: 19–22. . Hutchins (1995).
Embodiment and the philosophy of mind . For example, see the comments on pp. 277–278 of Fodor (1991). . Syntactic properties are any non-semantic properties that can be directly exploited by a physical system. Temporally extended processes, as described in Section 2, are in this sense syntactic too. . Knierim and Van Essen (1992). . van Gelder (1995: 376). . Of course there must be something that persists or else memory-based action would be impossible. The point is just that the space of internal representational vehicles may be much larger than the space of persisting inner states. . See Port and van Gelder (1995).
References Abraham, R. & Shaw, C. (1992). Dynamics. The Geometry of Behavior. Redwood, CA: Addision-Wesley. Agre, P. & Rosenschein, S. (Eds.). (1996). Computational Theories of Interaction and Agency. Cambridge, MA: MIT Press. Ashby, W. Ross (1956). Introduction to Cybernetics. New York: Wiley. Ballard, D. (1991). Animate vision. Artificial Intelligence, 48, 57–86. Ballard, D., Hayhoe, M., Pook, P., & Rao, R. (1997). Deictic codes for the embodiment of cognition. Behavioral and Brain Sciences, 20(4), 723–767. Bogdan, D. I. (1988). On information. Mind and Language, 3(2), 123–140. Churchland, P. M. (1989). A Neurocomputational Perspective. Cambridge, MA: MIT Press. Churchland, P., Ramachandran, V., & Sejnowski, T. (1994). A critique of pure vision. In C. Koch & J. Davis (Eds.), Large-Scale Neuronal Theories of the Brain (pp. 23–61). Cambridge, MA: MIT Press. Clark, A. (1989). Microcognition: Philosophy, Cognitive Science and Parallel Distributed Processing. Cambridge, MA: MIT Press. Clark, A. (1997). Being There: Putting Brain, Body and World Together. Cambridge, MA: MIT Press. Clark, A. (1998). Magical words: how language augments human cognition. In S. Boucher & P. Carruthers (Eds.), Thought and Language. Cambridge: Cambridge University Press. Clark, A. & Grush, R. (1997). Towards a Cognitive Robotics. St. Louis: Washington University at St. Louis. Clark, A. & Toribio, J. (1994). Doing without representing? Synthese, 101, 401–431. Decety, J. & Grezes, J. (1999). Neural mechanisms subserving the perception of human actions. Trends in Cognitive Sciences, 3(5), 172–178. Dennett, D. (1995). Darwin’s Dangerous Idea. New York: Simon & Schuster. Elman, J. (1991). Representation and structure in connectionist models. In G. Altman (Ed.), Cognitive Models of Speech Processing (pp. 342–382). Cambridge, MA: MIT Press. Fodor, J. (1991). Replies to critics. In B. Loewer & G. Rey (Eds.), Meaning in Mind: Fodor and his Critics (pp. 255–319). Blackwell: Oxford.
Andy Clark
Fodor, J., & Pylyshyn, Z. (1988). Connectionism and cognitive architecture: a critical analysis. Cognition, 28, 3–71. Gibson, J. J. (1979). The Ecological Approach to Visual Perception. Boston: Houghton-Mifflin. Grush, R. (1995). Emulation and Cognition. Ph.D. Dissertation, Berkeley: University of California. Haugeland, J. (1995). Mind embodied and embedded. In Y.-H. Houng & J.-C. Ho (Eds.), Mind and Cognition (pp. 3–38). Taipei, Taiwan: Academia Sinica. Hutchins, E. (1995). Cognition in the Wild. Cambridge, MA: MIT Press. Ito, M. (1984). The Cerebellum and Neural Control. New York: Raven Press. Jackendoff, R. (1996). How language helps us think. Pragmatics and Cognition, 4(1), 1–34. Kawato, M. (1990). Computational schemes and neural network models for formation and control of multi-joint arm trajectory. In W. T. Miller III, R. Sutton, & P. Werbos (Eds.), Neural Networks for Control (pp. 197–228). Cambridge, MA: MIT Press, Kawato, M., Furukawa, K., & Suzuki, R. (1987). A hierarchical neural network model for the control and learning of voluntary movement. Biological Cybernetics, 57, 69–195. Kelso, J. A. Scott (1995). Dynamic Patterns. Cambridge, MA: MIT Press. Kirsh, D. & Maglio, P. (1995). On distinguishing epistemic from pragmatic action. Cognitive Science, 19, 513–549. Knierim, J. & Van Essen, D. (1992). Visual cortex: cartography, connectivity and concurrent processing. Current Opinion in Neurobiology, 2, 150–155. Malcolm, C., Smithers, T., & Hallam, J. (1989). An Emerging Paradigm in Robot Architecture. Edinburgh: Edinburgh University, Department of Artificial Intelligence. Marr, D. (1982). Vision. San Francisco, CA: W. H. Freeman. Mataric, M. (1991). Navigating with a rat brain: A neurobiologically inspired model for robot spatial representation. In J. A. Meyer & S. Wilson (Eds.), From Animals to Animats. Cambridge, MA: MIT Press. McNaughton, B. L. & Nadel, L. (1990). Hebb-Marr networks and the neurobiological representation of action in space. In L. Nadel (Ed.), Neural Connections, Mental Computations (pp. 431–472). Cambridge, MA: MIT Press. Merleau-Ponty, M. (1942). La Structure du Comportment. Paris: PUF. Millikan, R. (1996). Pushmepullyou representations. In L. May, M. Friedman, & A. Clark (Eds.), Mind and Morals (pp. 145–162). Cambridge, MA: MIT Press. Port, R. & van Gelder, T. (Eds.). (1995). Mind as Motion: Dynamics, Behavior and Cognition. Cambridge, MA: MIT Press. Pylyshyn, Z. (Ed.). (1987). The Robot’s Dilemma: The Frame Problem in Artificial Intelligence. Norwood: Ablex. Rumelhart, D. & McClelland, J. (1986). On learning the past tenses of English verbs. In D. Rumelhart et al. (Eds.), Parallel Distributed Processing: Explorations in the Microstructure of Cognition, Vol. 2 (pp. 216–271). Cambridge, MA: MIT Press. Smith, B. (1996). The Origin of Objects. Cambridge, MA: MIT Press. Thelen, E. & Smith, L. (1994). A Dynamic Systems Approach to the Development of Cognition and Action. Cambridge, MA: MIT Press. van Gelder, T. (1995). What might cognition be, if not computation? Journal of Philosophy, 92(7), 345–381.
Embodiment and the philosophy of mind
van Gelder, T. & Port, R. (1995a). It’s about time: an overview of the dynamical approach to cognition. In Port & van Gelder (Eds. 1995), 1–44. Varela, F., Thompson, E., & Rosch, E. (1991). The Embodied Mind. Cambridge, MA: MIT Press.
Premise My reconsideration of Merleau-Ponty’s insightful views on language, mind and culture in relation to actual discussions on mental causality owes much to the fact that he set himself the goal of bringing phenomenology into closer contact with the realities of life in order to reach a philosophical account of the “common-sense world”.1 His views can also offer some arguments against the standpoint that there is a psychological thought process ready to reflect on a mind that often presents itself as free of all the stuff of every-day life. They also serve to question the multiple motives that may lead a human being not only to a more exact knowledge of the world-facts, but also to a manifold of choices, more or less conscious, in structuring his personal life. Neuroscientists claim that they “. . . have, at last, made the idea of a universal human nature legitimate once again” (Marconi 2001: 138) in contrast to the received view of culture and human beings, which holds to “the independence of language from biology” (ib.: 128). The followers of neural network connectionism investigate brain dynamics as a complex system capable of selfregulation, thanks to its strongly interconnected structure. Through simulated brain architecture, they try to reproduce the human capacity for construing different and not predictable balances interacting with virtual contexts, with such contexts covering both natural and social aspects of the human world. This way of investigation, besides giving a good understanding of what a brain can do and how it effectively works, gives a place for individual beliefs
Antonella Lucarelli
and motivations, and provides a model of the mind enriched by its manifold of specific competences. Expert systems, on the other hand, have encountered remarkable difficulties in giving an account of what each one of us makes of daily life (Boden 1989). To support the advantages of this kind of investigation, it has been claimed that research on expert systems has proved that psychotherapeutic words have better results than psycho-medicines, as they can produce “specific and much more enduring changes, totally relieving psychological disease [whereas] psychopharmacological quick, temporarily, a-specific results do not reach the basic causes of disease” (Parisi 2001: 46–48). The evermore-increasing complexity of these physical systems seems to progressively degrade subjective processes of any interest. Faced with ever more rapid cultural changes, and of an historical context that requires increasingly refined analyses to be understood, human mind presents itself awkwardly and insufficiently equipped in its search for immediate solutions. Will the angels’ language (De Monticelli & Di Francesco 1989) be able to stimulate in “brutes” a practical way of life that is attuned to the conflicts and contradictions of interest to psychologists in a more-than-formal analysis? Much more than many other philosophers, Merleau-Ponty considered a main task for phenomenology to be that of achieving a better understanding of human phenomena as spontaneously experienced in every-day life. In his view, the accomplishment of this task could not escape a reconsideration of psychological theory and practice as well as the investigation of a possibly consistent convergence of results coming from psychology with an enlarged phenomenological approach. His accurate analysis, criticism and suggestions addressed to psychology of the time are still an important source of reflection and a spur towards new conceptualisations, especially in the field of development and learning.
.
The phenomenal field
The courses held by Merleau-Ponty at the Sorbonne in 1949–1951 were devoted to various pedagogical and psychological aspects of mental development. His line of argumentation surprised his students at first (Damish 1964), who probably did not expect any answer to phenomenological questions from psychology, and even less from the psychology of the infant. The choice was not so amazing because Merleau-Ponty considered the investigation of behavioural development to be the key method of what he called a genetic approach to hu-
Causes and motivations
man phenomena. Developmental psychology seemed to be the best starting point for a possible convergence with phenomenology in the understanding of mental phenomena. The central essay of the course, namely, Les sciences de l’homme et la phénomenologie, enlarges the dialogue to sociology, anthropology and history, but the final essay, Les relations avec autrui chez l’enfant, summarizes and confirms the major aim of the courses. The great difficulty with psychology, Merleau-Ponty argues, is that, from its foundation in Leipzig by Wilhelm Wundt (1879) – it has assumed the approach and successful results of the natural sciences as ideals. The introspectionists broke the complex phenomena of consciousness into their presumably irreducible elements, so that at last the whole system of mind turned out to be constructed by composition of elementary units (i.e., the ‘atoms of mind’). Behaviourism was no less atomistic, except that it tried to establish causal relationships between elements of environment and behavioural responses. Gestalt psychologists and Albert Michotte strongly put into question the deterministic model of cause-effect chains considered to be proper focus of study in the natural sciences. Their inquiries on visual perception struck the heart of the matter: the discrepancy between the natural scientific framework adopted by psychology and their results demanded at least a shift towards a probabilistic model of psychical learning in which the relationship between internal and external components had to be reconsidered (Conti 1974). Gestalt psychologists did speak of experience as a structure, and of the ‘phenomenal character’ of behaviour; Koffka (1935), for example, distinguished the subjective from the physical environment, the former being phenomenal, and the latter objective. What he meant had already been clarified by Köhler (1925) in his investigation on the intelligence of monkeys. In fact, Köhler had stressed the importance of meaning in identifying the phenomenal field, as distinct from the physical environment: the same chair a monkey uses to sit on becomes a staircase when another monkey climbs up on it to reach some hanging bananas. Thus, the separation of mental contents and concrete things was about to be surpassed by focussing on the mediation bestowed by animal behaviour to the physical environment. It was the animal’s gesture or behaviour that, while giving a meaning to present physical objects, established the psychological, phenomenal field. This field replaces the priority previously ascribed to an ego-logical subject by registering the meaning that a specific situation assumes in the bodily gesture (Levin 1999). The priority of the phenomenal field points to the importance of the pre-personal dimension of experience, constituted by bodily actions – a sort of 0-degree (Nullpunkt) of consciousness. Nonetheless, it is intentionally related to the external situation
Antonella Lucarelli
and imbues it with a meaning. It is the predominance of the body that helps us to understand the way in which meaning enters and rules the phenomenal field.2 Köhler later observed that if physics itself can use structural and field terms, psychology must not hesitate in doing the same. With the Principle of Isomorphism (i.e., a structured correspondence between brain dynamics and mental phenomena) any previous criticism of behaviourism and statements about meaning and the phenomenal field seemed to collapse. Moreover, isomorphism seemed to dispose of any role ascribed to socio-historical factors. If psychic phenomena are just epiphenomena wholly explained by some chemical interactions – ultimately, physical motions – in the brain, how can we consider subjective development in relation to the historical/cultural world? Only Lewin (1931) understood the phenomenal problem and considered subjective context as a field of forms, of cultural constructs, exercising forces and tensions on the subject and influencing his meaning-oriented behaviour (Merleau-Ponty 1964b: 160). Thus, Merleau-Ponty decided to adopt the notion of Gestalt, or structural whole (in his own way), as a methodological principle, a sort of warning against the tendency to concentrate on only one aspect of human development. In particular, he tried to show that individual development is a process of growing appropriation and differentiation of subjective intentionality in relation to the meanings of the specific historical/cultural context. In this perspective we cannot speak of the intercourse between the two separately established poles of nature and culture, but of an essential mesh of both, which only results in intelligible meanings construing the phenomenal experience. The following words express Merleau-Ponty’s conviction that intentional meaning is the focal feature of human experience: We cannot distinguish between a first level of behaviour to be designated as “natural” and a spiritually construed, “cultural” level. Each level is both natural and construed, i.e., all words and behaviours owe something to the basically biological being and at the same time they escape the simplicity of animal life: vital behaviours are raised up from their animal sense, through a sort of evasion and a genius for ambiguity, which could jointly serve to define the human being. (Merleau-Ponty 1945: 221)
The ‘genius for ambiguity’ is essentially supplied by subjective intentions aimed at meaning. In recent times both Bruner (1986, 1990) and Harré (1987, 1998) have emphasized the tension towards meaning as the main characteristic feature of
Causes and motivations
human individuation. Bruner investigated the spontaneous unfolding of autobiographical narrative while Harré pointed to a discursive conception of the mind. Merleau-Ponty would certainly have subscribed to their effort to grasp subjective phenomena as they spontaneously unfold in every-day life. The problem with scientific psychology, Merleau-Ponty noted, is that it considered subjective phenomena as escaping any empirically objective grasp and as representing an inaccessible, inner world. But there is special sense in which psychic phenomena do represent the external world. Perception, for instance, is reflected in gestures and any other bodily behaviour by the way in which the subject assigns meaning to actual situations. Psychologists must consider (I) that the perceived object is neither the thing-in-itself nor the act of perceiving it, and (II) that the meaning bestowed upon the object does not exhaust the multiplicity of meanings of the object itself. Perception grasps the phenomenal essence of the object in a partial, unfinished way – as an outline, or a profile, of the object. Psychology is concerned exactly with the structure of these outlines. Therefore as a science, it has to deal not so much with the physically real, but rather with phenomenal praxis. The possibility of a rigorous account of experience must be based, in Merleau-Ponty’s view, on intentional analysis, which is an analysis of meaning – of the perceived objects as they are subjectively meant (e.g., the chair as comfortable or as uncomfortable). Psychologists had thus to distinguish between the external reference, a thing or a person in physical space-time, and the meaning or the internalised reference as the subjective activity apprehends it. Subjective phenomena are not private impressions, momentary sensations occurring so quickly that they cannot be retrieved. Rather, they have temporal breadth and spatial extension, open to behavioural recognition and linguistic description. Because of intentionality, phenomena are neither locked in a mysterious, unreachable interior deprived of access to others,3 nor can they be obtained by a prior, ego-logical and purely conceptual thought. From first perceptions to spoken language, subjectivity is essentially open to the world and, in particular, to other people. It is consequently endowed of structure in many specific ways and available to psychological researchers through various manifestations. From the simplest practical behaviour to culture-laden discourse, human behaviour is open to being appropriated and understood by others. Consciousness is not an object but it does objectify itself in behaviour. In Merleau-Ponty’s approach, mental development may be considered as a process of never ending mediation between subjective meanings and those afforded by the historical/cultural context. To support the emphasis on the
Antonella Lucarelli
way this larger context, reflecting a system of already construed meanings, can influence behaviours traditionally ascribed to the most private dimension, Merleau-Ponty referred to Margaret Mead’s Sex and temperament in three primitive societies (1935). He particularly appreciated her cultural enlargement of psychoanalytic theory from what concerns the Oedipus complex. The difference among various ways of children developing is connected with the difference in sexual relationships between males and females, not considered as a cause, but as a cultural medium. In Mead’s description every specific cultural context is a balanced system of multiple elements: each one of them makes the whole system explicit and cannot be isolated from other elements. The way of life as a proper structure of each specific context passes to the newborn through his multiple interactions with objects and persons. The way in which every social group builds up in time its sexual stereotypes is a key factor in the whole cultural structure, and it affects any relationships of that particular social group. In particular, it affects the relationships between mother and infant. Examples afforded from other anthropological studies confirm the importance of ‘cultural meanings’ in shaping patterns of behaviour. The role ascribed to cultural context emphasises the importance of individual performances as in some recent perspectives on human development. Starting from an accurate re-reading of Vygotsky, and taking account of the ecological approach by Bronfenbrenner (1976), the results obtained by Cole (1996), Wertsch (1991) and Valsiner (1992) suggest new methods for investigating human development through the accurate analysis of the differentiated influence of specific contexts on individuals. Merleau-Ponty’s phenomenal-cultural approach owes much to his assiduous reading of Wallon’s works.4 Wallon’s perspective, even if less renowned, was consistent with many of Vygotsky’s tenets (see Van der Veer 1996). In Les origins du caractère chez l’enfant (1935) for instance, after distinguishing the ‘expressive gesture’ from the ‘pseudo-gesture’ in the newborn, Wallon emphasised the need for a psychology concerned with the analysis of personal meaning. This concern led him to investigate the importance of the biological features of the individual as having an influence on the differential attunment with others starting with the very first interactions. Such a perspective becomes more radical in studies by Trevarthen (1991, 1998), who stated that the way the new-born exercises biological competences, mostly concerned with the communication of feelings, heralds the dawning of one’s subjective consciousness. Merleau-Ponty would subscribe this view, stating that phenomenal consciousness starts from praxis, intended as a ‘way of
Causes and motivations
living in the world’ (see Merleau-Ponty 1990: 136). Here, one might recall that psychology was for him precisely concerned with the structures of this praxis. In fact, Merleau-Ponty argues that the phenomenal world is accessible in a way that psychology, as a science, should be founded on such accessibility. There should be no more searching for causal sequences as psychology is concerned with the contingent and provisional, albeit incomplete, subject-relative construction of meanings that depends on concrete situations in the commonsense world. In such perspective, psychology has not to deal with externally real entities, objectively given to the subject, but rather with the (different) meanings that any experience can acquire in the construction of (different) subjects. Despite all the difficulties inherent to the analysis of these phenomena, Merleau-Ponty claimed that they can be grasped as they are really lived. Moreover, it is only by grasping them as they are really lived that a proper scientific, investigation of human experience can start. Science ought not to reify phenomena that are not things, since it is committed to grasping them exactly as they manifest in live contexts. Human behaviour is always mediated by a socio-cultural context, through which only the construction of meaning becomes something objective. This mediation covers the whole range of psychological structures and makes it impossible to connect individual behaviours in linearly ordered causal sequences. In contrast, it helps us to understand the intentional morphogenesis of meanings. Psychology is an ‘indirect knowledge’ that can reach the internal structure of phenomena by investigating subjective meaning (Merleau-Ponty 1964b: 163). Thus the problem of causality must be differently posed. This can be achieved by introducing the notion of motivation: in contrast to a third person description of relationships between external objects and persons, motivation calls for the first person irreducible intertwining of meanings that result in the subjective construal of behaviour. We cannot give an adequate account of this intertwining by means of linearly ordered causal chains. In this approach, Merleau-Ponty claims, the number of people observed is not particularly relevant. The monographic analysis of single cases can be sufficient to investigate the internal structure of intentionality, as it progressively emerges in a subject coping with a specific context. Here again Merleau-Ponty anticipates some recent views, perhaps not popular but accurately supported (Smith 1995a, b), which stress the importance of qualitative analyses (of single cases) as the proper method for testing individual meaning about some key points of existential practice. Even though insufficient by itself, phenomenological psychology could be helpful in enlarging the horizon of psychological science. For a science seeking
Antonella Lucarelli
to a methodically rigorous and non-dogmatic picture, a general understanding of entities that possess consciousness must rely on different methods than a science dealing with entities not possessing it. If the rational criteria are the same, the manner of using them is different. One may apply a linear causeeffect analysis to inorganic bodies and come up with universal laws about their behaviour, according to the composition of physical forces. But one cannot do the same with the intertwined unfolding of motivations over time, with human consciousness in its complexity and with the constitutive role of intentional meaning in shaping human life and its value. External causes and internal motivations cannot be confused, nor can physical forces and meanings. Even though all of them can be subsumed under the general criteria of scientific knowledge, each domain demands a specific use of such criteria. The structure of perception exhibits features mutually interwoven with developmental timing (Merleau-Ponty 2001b: 28). In the course of time, some meaning-structures disappear and others replace them, but the coexistence of competing structures also occurs: conceptual economy may well be a useless strategy in child-life but it ends with having first importance in adultlife. Meaning undergoes a corresponding shift through the unfolding dialectics of the individual and the world. In the increasing variety of expressions from child’s to adult’s experience, we have to identify the specific meanings involving the subject’s phenomenal field as a whole. Exactly as in front of a painting one can say that “while lending his body to the world, [the painter] can change the world in paintings” (Merleau-Ponty 1964a: 16). It was by means of the notion of chiasm that Merleau-Ponty tried to capture the intertwining between subjective intentionality, expressed through the body, and the cultural world. This notion brings to mind some recent accounts of development according to which the external (i.e., cultural) elements are considered no longer as separated from the subject, but as directly interwoven with the internal activity of the organism (Marchetti & Liverta Sempio 1995: XXXII). Merleau-Ponty had already grasped the priority of this original and irreducible intertwining, as is clear from his claim that rational consciousness is always a “reflection on what lacks reflection in order to confer meaning” (Merleau-Ponty 1964b: 148). The resulting enigma is described in the following words: It has to do with the fact that my body is something looking and at the same time it is something that can be looked upon. [. . .] My body looks at myself while I am looking at it, touches myself while I am touching it, and possesses
Causes and motivations
this reversibility as an intrinsic resource. It is myself by a sort of slip. [. . .] My visible and moving body pertains, as a thing, to the material world; it is one thing among others. But as it looks and feels and moves, it takes things all around as a prolongation of itself, and the stuff of the world is assimilated to that of my body. (Merleau-Ponty 1964a: 19)
The body is both the principal constraint and the basis of phenomenal experience. Its coordination goes along with the process of individuation while reducing human consciousness to a 0-degree in the sleep-dream state. A sleeping human being exemplifies the 0-degree of consciousness (Gauchet 1971), or of “what dreams in us” (Merleau-Ponty 1989: 14). Yet an emotional feeling links the sleeping person to his images. Following the criticism by Politzer (1929) of psychoanalysis, MerleauPonty stated that the interpretation of dreams has no need to postulate the presence of the unconscious as far as it is assimilated to a sort of box of removed experiences that only the expert could interpret. Dreams pertain to phenomenal experience and identify a condition of consciousness resembling adult attachment to others in particular affective relationships, where the ‘others’ involved are themselves nearest to dreams or myths. Hence, Merleau-Ponty argues, the screen between reality and imagination is never definitely completed in human life.
. Language and the phenomenal field Language learning is acquiring a new way of being-in-the-world, which affects the whole subjectivity including intellectual, affective and practical intentions. Through language the child acquires a sense of identity and progressively takes distance from what Merleau-Ponty, following Wallon, defines as “syncretic sociability” (Merleau-Ponty 1997: 222). This transition is marked by both a growing consciousness of the historical/cultural context and an increasing appropriation of collective meanings. But this expanded existential condition involves a risk, since language is the medium of a self-referential system, which exercises a great power on personal feelings. If it enlarges the boundaries of expression, it may also contribute to losing the bodily, polymorphic relationship with the phenomenal world. Thus it involves the risk of escaping into some ‘conceptual’, formally defined system of meanings.
Antonella Lucarelli
Merleau-Ponty anticipated what Stern (1985) has described more accurately: on one hand, language enormously enriches our grasp on reality, on the other hand it offers a device of mystification, which may distort, or even lose, existential experience. Human beings may alienate their existential meaning by fleeing into the use of ‘conceptual’ language. Symbolic forms, as crystallized formulae, might invade even the inner subjective feelings and result in utterances mystifying the subject’s own thoughts.5 Merleau-Ponty strongly engaged in demonstrating that linguistic expression is connected with the phenomenal field. According to a psychogenetic approach the first words are the prolongation of gestures6 and are directed by the same kind of intentionality.7 Merleau-Ponty, in fact, appeals on Trubetzkoy and Jacobson’s phonological investigations, which identify a previous level underneath language that reveals the organisation of sounds as a continuous movement towards meaning.8 The coincidence with Vygotsky’s (1935) description of the progressive relationships between subjective intentions and language is astonishing. This is not only the case with regard to phonology, but, as we shall see, in the very definition of the word as the unit of linguistic analysis. Why? Because the word, as the basic expression of meaning, embodies the inner subjective language and allows for communication with others. In order to stress that language pertains to the range of phenomenal behaviour, Merleau-Ponty pointed out that the first words emerge from the wild, pre-cultural body – or as Bonomi (1967: 77) sensibly noted, “. . . symbolic communication starts with man’s first cry”. But language also indefinitely enlarges the dialogue with other people9 and marks the entrance and the intertwining of this dialogue with collective history. The possibility of symbolic abstractions can lead to an idealistic view of language quite indifferent to the complexity of the living. In science, as well as in logic and philosophy, the ideal of a wholly transparent language in which any proposition could be expressed in standard form, thanks to pure semantic competence, paves the way to an idea of knowledge as independent from any actual constraints of space and time. Merleau-Ponty admits that this interpretation relies on a possibility inherent to the intentional horizon of linguistic expression, one that also belongs to phenomenal field and must be located in the historical context of collective knowledge to be rightly understood. Historically, the first expressions were probably a kind of singing in which human beings sang their feelings before communicating their thoughts. In the same way they represented their lives in paintings, before the invention of writing (Merleau-Ponty 1945: 252).
Causes and motivations
To avoid a possible misunderstanding of this remark, Merleau-Ponty claimed that he was not proposing a naturalistic interpretation of the origins of language. Rather, his aim was rather to tighten the link between thought and language and to suggest that this link pertains to phenomenal experience. In his later works on language, Merleau-Ponty (2001c, d) further investigated the way language leads the unfolding of subjective meanings and thus the relationship between the speaking subject and the language - system. This relationship is not at all transparent in a language autonomously transforming itself through what seems a constant and transversal langagière work. MerleauPonty adopted and partly renewed the structuralist ideas introduced into linguistics by Ferdinand De Saussure (see Saussure 1980). Saussure had sharply divided the level of the langue, as the system proper, from the level of subjective expression, parole. In doing so, he ascribed the latter to the accidental, psychological register, which only acquires relevance a posteriori, in the diachronic dimension of speech. In contrast, Merleau-Ponty held that langue and parole are contemporary and inseparable: the subjective expression and the linguistic system constitute a whole structure and each subjectladen use of words in speech contributes to transforming the whole system because new meaning of words can thus emerge. Through the importance of the subjective dimension he dovetailed structuralism with phenomenology and suggested an original solution to the problem of synchrony and diachrony in language (Paci 1963: 225). What is mostly important for Merleau-Ponty is the argument according to which the word/sentence, as actually used in speech, never expresses a previous concept or thought. The unit of linguistic analysis is “the meaning of the word”, and that confused, shortened, and fragmented inner sort of language, aiming to be embodied in language, cannot be defined as thought. If thought is something meaningful, it only exists when expressed in words. According to this approach, psychology is particularly concerned with the individual speaker’s effort to understand and produce meanings. When the subject is unable to transcend in words the particularity of his perspective, and strains in search of adequate expression, then a specifically psychological dimension becomes focused, because such effort pertains to phenomenal experience. When subjects surpass or remove their personal perspectives, psychology is removed too. In the analysis of personal narratives, for example, we have to frame every expression within the dynamics of each individual’s behavioural patterns. Such needs to be done if we intend to trace back the origin of that specific se-
Antonella Lucarelli
mantic structure and to understand the specific history of that single person in phenomenal context. Merleau-Ponty would probably subscribe to the statement made by Wittgenstein (1953): imagining a language is imagining a form of life. It is this specific way of considering language as pertaining to phenomenal experience which, as Braddock (2001) notes, is removed by general AI models and it is also removed by Dennett’s multiple drafts model of consciousness according to which, set aside from any contingent disposition lacking its exact identification, we can but attain to the verbal account of subjective experience. Verbal account, as we have already seen, is not enough in Merleau-Ponty’s approach to grant, by itself, faithful intentional expression. In fact, in the discussion of the thesis that words stand for things just in virtue of arbitrary conventions, Merleau-Ponty observed that nonetheless the subjective articulation of words keeps something of their original, existential meaning10 with which psychology is chiefly concerned.
. Conclusions The distinction between two kinds of psychology, corresponding to the first person and the third person perspective (Lostia 1994), covers the whole history of psychological research and suggests the co-existence of two methodological options: psychology as a human science, or psychology as a natural science, respectively. This distinction should not be confused with the contraposition between a pure, disembodied psychology and a biology-oriented one that reduces mental processes to neural activity. Both kinds of psychology belong to one and the same field and do not necessarily contradict each other. The investigation of the many-layered hierarchy of levels, through which mental processes depend on brain’s activity keeps its fundamental importance. We can surely say that Merleau-Ponty would have appreciated present research in the neurosciences, as he did with Pavlov’s investigations (Livet 1990). Nevertheless, the distinction between the two perspectives hides an epistemological problem. The problem concerns both the legitimacy of historical/clinical methods, with their specific traits with respect to the experimental method of natural sciences, and the theoretical options regarding the relevance to be assigned to linguistic expression in describing psychic phenomena (Brigati 2001).
Causes and motivations
There is no doubt that Merleau-Ponty fought for a first person psychology, as instantiated in the humanities by researchers interested in the mental life of people as living beings in a historical/cultural context. It is a kind of approach that neither forgets the naive psychology of common sense nor relies on abstract models. No advancement in the biological foundations of psychiatry, in the neurosciences and in information-processing psychology could ever divide the individual from his phenomenal experience of life. We can use Merleau-Ponty’s radical appraisal of first person psychology, together with some related concepts, such as phenomenal field and meaningoriented intentionality, as a stimulus to better understand present-day investigations based on a contextual view (Liverta Sempio & Marchetti 1995), which stresses the dialectics of personal and cultural history. In our opinion, the embodied mind approach, which can be traced back to Merleau-Ponty is actually helpful in overcoming the methodological and theoretical limits of widespread psychological models, in favour of a better understanding of human beings as inhabiting a specific world of cultural meanings.
Notes . See Marconi (2001: 34), where the actual acknowledgement of “common sense” is defined “as a bottom problem concerning both philosophy and cognitive sciences”. . The primacy of the phenomenal field is considered by Merleau-Ponty as a basic achievement of Gestalt psychologists in contrast with Piaget’s definition of intelligence. In Piaget’s theory of cognitive development, after the early stage of perceptual and sensorimotor patterns, intelligence reaches its utmost realisation through a process of increasing, equilibrium-driven abstractions, the outcome of which is a set of structures totally independent from concrete situations. In the perspective of Gestalt theory, such indifferent detachment from concrete reality is never reachable. See Merleau-Ponty (1964b: 156). . The target of these criticisim is Bergson: “Bergson’s philosophy could not shake psychologists from their quest for rigorous objectivism”. Merleau-Ponty writes, “as he represented subjective experience as inexpressible” (Merleau-Ponty 1964b: 154). . The Wallon-Merleau-Ponty relationship would deserve a study in itself. Merleau-Ponty and Wallon shared the conviction that no one-to-one correspondence of brain’s areas and mental functions really exists and that damage of certain parts of the brain can result in other parts taking over control. This view was quite popular at the time in France; it went back to the work of Hughlings Jackson and was updated by Goldstein (1934) – another main source for Merleau-Ponty and Wallon’s (1953) work under examination. . See Merleau-Ponty (1964b: 151): “Reflection and history become correlatives, thought is historicity, as far as achievement of one’s own self and its conscius insertion in history.” See also Merleau-Ponty (1945: 213–214): “Thought is no ‘inner’ thing, it does not exist indepen-
Antonella Lucarelli
dently from words. It is deceiving to believe that our thought exists in itself before structured words; by the idea of ready-made thoughts [. . .] we fool ourselves about our inner life”. . See Merleau-Ponty (1945: 226): “a gripping of the throat, some air whistling between tongue and teeth, a certain movement of the body are suddenly overwhelmed by a figurative meaning and signify it in the ouside”. . Cf. Merleau-Ponty (1945: 158): “Underneath concious life [. . .] an ’intentional arch’ projects around us our past and future, our human context, our physical, ideological and moral situation or, to say it better, places us simultaneously in all these relations. This intentional arch unifies senses and intelligence.” . Merleau-Ponty (1964b: 234), Jakobson (1968). . Cf. Merleau-Ponty (2001b: 34–35): “When I speak with the others I don’t speak about my thoughts, but I tell them together with what remains in the middle [. . .] Not from mind to mind, but from a bodily and linguistic being to another bodily and linguistic being, each pulling the other by invisible threads like those moving puppets, making the other think and speak and making him become something that, by himself alone, could never become. So things are said and thought by words and thoughts that we do not owe, but are, even if we do not realize it”. . See Merleau-Ponty (1945: 218): “Conventions are a late form of relationships among human beings; they presuppose earlier means of communication, and language must be put back into this current of intercourse. If we consider only the conceptual and limited meaning of words, it is true that verbal form seems to be arbitrary. But if we took into account the emotional content of each word, its gestural sense [. . .] we would find that words, vowels and phonemes, are, first of all, ways of singing the world, and their original function was to express emotional essence.”
References Boden, M. A. (1989). La simulazione della mente al calcolatore è socialmente dannosa? In R. Viale (Ed.), Mente Umana, Mente Artificiale (pp. 3–17). Milano: Feltrinelli. Bonomi, A. (1967). Esistenza e Struttura. Saggio su Merleau-Ponty. Milano: Il Saggiatore. Braddock, G. (2001). Perception and judgement in Dennett and Merleau-Ponty. http://www.sas.ac.uk/ Philosophy Conferences 2000–2001. Brigati, R. (2001). Le Ragioni e le Cause. Wittgenstein e la Filosofia della Psicoanalisi. Macerata: Quodlibet. Bronfenbrenner, U. (1976). The Ecology of Human Development. Experiments by Nature and Design. Cambridge, MA: Harvard University Press. Bruner, J. (1986). Actual Minds, Possible Worlds. Cambridge, MA: Harvard University Press. Bruner, J. (1990). In Search of Meaning. Cambridge, MA: Harvard University Press. Cole, M. (1996). Culture in Mind. Cambridge, MA: Harvard University Press. Conti, C. (1978). La Causalità nella Psicologia. Milano: Mazzotta. Damish, H. (1964). Avant-propos. Le versant de la parole. Bulletin de Psychologie, 18(3–6), 1–4.
Causes and motivations
De Monticelli, R. & Di Francesco, M. (1989). Lingua degli angeli e lingua dei bruti. Teoria, 9(1), 69–137. De Saussure, F. (1966). Course in General Linguistics. New York: McGraw Hill (French 1st ed., 1913). Dreyfus, H. L. (2001). Merleau-Ponty’s critique of mental representation: The relevance of phenomenology to scientific explanation. http://www.hfac.uh.edu/cogsci (dreyfus html). Dreyfus, H. L. & Dreyfus, S. E. (1990). Making a mind vs modeling the brain: Artificial Intelligence back at a branch point. Daedalus, 117(1), 15–44. Gauchet, M. (1971). Le lieu de la pensée. L’ARC, 46, 19–31. Goldstein, K. (1933). L’analise de l’aphasie et l’essence du language. Journal de psychologie, 1–4, 451–496. Goldstein, K. (1934). Der Aufbau des Organismus. den Haag: Martinus Nijhoff. Harré, R. (1987). The Social Construction of Emotions. Oxford: Blackwell. Harré, R. (1998). The Singular Self. London: Sage. Jakobson, R. (1968). Child Language, Aphasia and Phonological Universals. The Hague: Mouton. Koffka, K. (1935). Principles of Gestalt psychology. New York: Harcourt Brace. Köhler, W. (1925). The Mentality of Apes (2nd ed.). New York: Harcourt Brace. Levin, D. M. (1999). A responsive voice: language without the modern subject. Chiasmi International, 1, 65–102. Lewin, K. (1935). Environmental forces in child behaviour and development. In C. Murchison (Ed.), Handbook of Child Psychology, Vol. 2 (pp. 590–326). Worcester, MA: Clark University Press. Liverta Sempio, O. & Marchetti, A. (1995). Il pensiero dell’altro: La mente, le menti e la dinamica interno-esterno. In O. Liverta Sempio & A. Marchetti (Eds.), Il Pensiero dell’Altro (pp. XIX–XL). Milano: Cortina. Livet, P. (1990). La phénoménologie du <>, face à l’intelligence artificielle. In Sauzeau Boetti, A.-M. (Ed.), 23–33. Lostia, M. (1994). Modelli della Mente, Modelli della Persona. Firenze: Giunti. Marconi, D. (2001). Filosofia e Scienza Cognitiva. Bari: Laterza. Mead, M. (1935). Sex and temperament in three primitive societies. New York: William Morrow & Co. Merleau-Ponty, M. (1945). Phénoménologie de la Perception. Paris: Gallimard. Merleau-Ponty, M. (1964a). L’Oeil et l’Esprit. Paris: Gallimard. Merleau-Ponty, M. (1964b). Maurice Merleau-Ponty a la Sorbonne, 1949. Bulletin de Psychologie, 18–236, 3–6. Merleau-Ponty, M. (1989). Il problema della passività: il sogno, l’inconscio, la memoria. Aut Aut, 232–33, 13–16 (French original, 1954–1955). Merleau-Ponty, M. (1990). La Structure du Comportement (1st ed., 1942). Paris: PUF. Merleau-Ponty, M. (1997). Les relations avec autrui chez l’enphant. In Id. Parcours 1935– 1951, 147–230. Paris: Verdier. Merleau-Ponty, M. (2001a). Signes (1st ed., 1960). Paris: Gallimard. Merleau-Ponty, M. (2001b). Préface (1st ed., 1960). In Id. 2001a, 9–61.
Antonella Lucarelli
Merleau-Ponty, M. (2001c). Le langage indirect et les voix du silence (originally in Les Temps Modernes, 80, 133–144; 81, 70–94). In Id. 2001a, 63–135. Merleau-Ponty, M. (2001d). Sur la phénoménologie du langage. In Id. 2001a, 136–158. Paci, E. (1963). Funzione delle Scienze e Significato dell’Uomo. Milano: Il Saggiatore. Parisi, D. (2001). Nuovi modelli nello studio della mente. In Mente e cervello. Genova: Il melangolo. Politzer, G. (1929). Critique des Fondements de la Psychologie. Paris: PUF. Sartre, J. P. (1961). Merleau-Ponty vivant. Les temps modernes, 304–376. Sauzeau Boetti, A.-M. (Ed.). (1990). La Prosa del Mondo. Omaggio a Merleau-Ponty. Napoli: Quattro Venti. Smith, J. (1995a). Semistructured interviewing and qualitative analysis. In J. Smith, R. Harré, & L. Van Langenhove (Eds.), Rethinking methods in psychology (pp. 9–27). London: Sage. Smith, J. (1995b). Repertory grids: an interactive, case-study perspective. In J. Smith, R. Harré, & L. Van Langenhove (Eds.), Rethinking methods in psychology (pp. 162–178). London: Sage. Stern, D. N. (1987). Il mondo interpersonale del bambino (1st ed., 1985). Torino: Boringhieri. Trevarthen, C. (1991). La genesi della coscienza umana nell’infanzia. In G. Giorello & P. G. Strata (Eds.), L’automa spirituale (pp. 119–136). Bari: Laterza. Trevarthen, C. (1998). Empatia e Biologia (1st ed. 1997). Milano: Cortina. Valsiner, J. (1992). Making of the future: Temporality and the constructive nature of human development. In G. Turkewitz & D. Devenney (Eds.), Time and timing in development (pp. 13–40). Hillsdale, NJ: Lawrence Erlbaum. Van de Veer, R. (1996). Henry Wallon’s theory of early child development: The role of emotions. Developmental Review, 16, 1–27. Wallon, H. (1925). L’Enfant Turbulent. Paris: Alcan. Wallon, H. (1934). Les Origins du Caractère chez l’Enfant (2nd ed., 1949). Paris: PUF. Wallon, H. (1953). L’organique et le social chez l’homme. In Scientia, 17. Reissued in Enfance, 12 (1959), 59–65. Wertsch, J. V. (1991). Voices of the Mind: A Sociocultural Approach to Mediated Action. Cambridge, MA: Harvard University Press. Wittgenstein, L. (1953). Philosophische Untersuchungen. Oxford: Basic Blackwell. Zamboni, C. (1989). Il linguaggio nella riflessione di Merleu-Ponty e i legami con lo strutturalismo. Aut Aut, 17–42.
Chapter 5
Mental causation and intentionality in a mind naturalising theory Sandro Nannini Università degli Studi di Siena
.
Introduction
For many years cognitive sciences and especially cognitive neurosciences have been trying to naturalise the mind, that is, to consider psychological and cultural phenomena as natural phenomena that can be studied by means of the usual procedures and methods of empirical sciences: procedures and methods identical with (or at least having their basis in) physical and biological processes. Therefore naturalising the mind is first of all a work in progress carried out – in very different forms indeed – by many neuroscientists, cognitive psychologists, psycholinguists, cognitive economists, artificial intelligence researchers etc.1 after the ‘revolution’ of behaviourists at the beginning of the xx century and with particular intensity over the last forty years after the birth and growth of cognitive sciences. However, such a scientific research programme has philosophical implications and presuppositions as well. First of all it renews the dispute that blew up between “historicists” (W. Dilthey, W. Windelband, H. Rickert etc.) and positivists at the end of the xix century in Germany and in the xx century was continued under various forms by many continental philosophers (e.g., H. G. Gadamer) and neo-Wittgensteinians (e.g., G. Anscombe and G. H. von Wright) on the one hand and, on the other, logical empiricists (e.g., C. G. Hempel) and other “naturalists”.2 The whole debate was about whether the explanations of natural events have the same logical structure as the understanding of human actions and cultural phenomena.3 Secondly, such an epistemological dispute has obvious ontological presuppositions. For the main reason naturalists have to study psychological and cul-
Sandro Nannini
tural phenomena by the methods of natural sciences is that they think human beings are part of nature. Therefore the epistemological debate between historicists and naturalists is inevitably intertwined with the so called “mind-body problem” and it is not astonishing that the present trend to naturalise the mind is rooted in a view of the human beings as the natural product of biological evolution and in the rejection of any form of Platonic, Christian and Cartesian ‘dualism’ according to which the mind (or the spirit, the soul) and the body are supposed to belong to two distinct and reciprocally autonomous levels of reality, that is, human beings are supposed to be composed by a material body and an immaterial soul. To sum up, naturalising the mind is a trend of scientific research that presupposes a certain philosophical background that could be labelled as ‘cognitive naturalism’, that is, a general philosophical framework common (more or less explicitly) to those philosophers, cognitive scientists, and neuroscientists who see the mind as a part of nature into which, like into any other part of it, one can inquire only by the methods of empirical sciences. Cognitive naturalism has been criticised from many points of view. However, for many years the most important attacks have been coming from the supporters either of ontological pluralism or of epistemological pluralism. The former claim that in the world there are different and irreducible levels of reality. The latter maintain that psychology and social sciences describe and explain their objects by means of concepts that are irreducible to the concepts of natural sciences.4 The earliest origin of ontological pluralism is in Plato’s theory of ideas and Descartes’ dualism (that is, his distinction between res cogitans and res extensa). Dualism has been recently proposed again by Chalmers (1996). However, since the author of the first and still most influential version of ontological pluralism5 in contemporary philosophy is Popper, I shall discuss here his theory. As to the criticisms of cognitive naturalism based on the epistemological pluralism I shall take into account J. McDowell’s recent criticism of ‘bald naturalism’.
. A definition of cognitive naturalism In order to better define cognitive naturalism I shall claim that all of the philosophical naturalistic theories of mind explicitly or tacitly accept or presuppose the following principles whereas all theories that can be reasonably considered non-naturalistic reject at least one of them (Nannini 2000): 1. Minds (as well as consciousness, spirit or subjectivity) are part of nature.
Mental causation and intentionality in a mind naturalising theory
2. There is no knowledge a priori (or obtained by methods that are different from the methods of empirical sciences) of any part of nature. Therefore minds, consciousness and subjectivity, as part of nature, can be known only by means of empirical sciences. 3. The physical world is a ‘closed’ system. Nothing that is non-physical can interfere with the physical world. No cause of a physical event can be a non-physical event. It is obvious that Cartesian dualism rejects all of those three principles: minds and bodies are two different ‘things’ according to Descartes (violation of the first principle). Therefore physics (and more generally every empirical science that studies the outer world, since it can study only the movements of bodies according to a mechanistic view of science) cannot include the comprehension of mental states (violation of the second principle). Moreover, according to Descartes, “the mind can operate independently of the brain; for certainly the brain can be of no use to pure understanding, but only to imagination or sensing”.6 In other words, there are mental acts without brain correlates. However, this claim is incompatible with the “closure” of the physical world if it is combined – as Descartes does – with the possibility of causal interaction between the mind and the body. In order to make clear this point let us assume that a mental act M is a causally necessary condition of the bodily movement P so that if not M then not P: for example, I raised my arm because I wanted to raise it. If I had not wanted to raise it, it would not have risen. In this case, if M has no physical correlate, then no set of physical events is a causally sufficient condition of P, that is, there is an event (P) that comes to be part of the physical world thanks to a non-physical act (M). Therefore the existence of P clearly violates the closure of the physical world prescribed by the third principle of cognitive naturalism. In other words, if the closure of the physical world makes it necessary that only physical events can cause physical events and nevertheless ‘mental causation’ exists, i.e. mental acts can cause physical events, then mental acts are physical events. If they were not physical they could not be the cause of anything in the physical world. The three hypotheses that (a) the physical world is closed, (b) mental acts can cause physical events and (c) mental acts are non-physical, are not compatible. To sum up, the first irreducible opponent of cognitive naturalism seems to be Cartesian dualism and it seems that there is a strong conceptual connection between naturalism and materialism in its broadest sense (including functionalism, provided only physical states can implement mental states).7
Sandro Nannini
However, there are contemporary forms of dualism that do not seem to oppose a naturalistic view of the mind (at least prima facie). One of these, Popper’s pluralism, is not only based on a view of the human beings as part of nature and as the products of a biological evolution but it reverses the conclusion previously gathered from the mind-body interaction: according to Popper the existence of mental causation not only does not exclude mind-body ontological dualism but it is the main proof that materialism is false. Let us see better how this seeming paradox is possible.
. K. Popper’s criticism of materialism In his famous book written with J. C. Eccles and published in 1977, The Self and Its Brain (Popper & Eccles 1985, 2nd ed.), and in other subsequent essays (see especially Popper 1994) Popper proposes a theory of the mind that aspires to be both naturalistic and anti-materialistic. This theory can be so summarised: 1. Life emerged from matter, consciousness from life, and “objective knowledge” (especially scientific theories) from human consciousness (Popper & Eccles 1985: 11). Popper calls the physical and biological world “World 1”, the set of all mental states “World 2”, and the set of all cultural products of human intelligence – from myths to works of art, scientific theories, other abstract objects, and social institutions – “World 3”. Given these definitions he maintains that the second world emerged from the first world and the third world from the second world. In other words, life is the result of physical and chemical processes, consciousness (that Popper seems to identify with ‘mentality’ at large in an Cartesian vein) emerged from life at a certain point of the evolution of species, and only through the intelligent activity of human beings, that is, by repeated mental acts, all products of human culture were brought about. Therefore, there is no life without matter, no minds without brains, and no culture without human minds (Popper & Eccles 1985: 36–39). More in detail, as for the ‘mind-body problem’ Popper, unlike Eccles, does not believe in “the existence of disembodied minds” (206). Therefore it seems reasonable to conclude that Popper, unlike Descartes, does not believe in the possibility of mental states without any brain correlates and maintains that the ‘higher levels’ of reality have been produced by lower levels: life from matter, minds from organisms etc. There is a causal relation from lower levels to higher levels.
Mental causation and intentionality in a mind naturalising theory
2. There is not only a causal relation from lower levels to higher levels but also vice versa a ‘backward’ causal relation from higher levels to lower levels. For example, scientific theories have deeply changed the physical world through their technological applications. Moreover, since science cannot act directly on the physical world such an interaction between the abstract objects of the World 3 and the material objects of the World 1 needs the mediation of the World 2, that is, the mediation of human beings. Therefore there are backward causal relations from the World 3 to the World 2 and from the World 2 to the World 1. For example, through mental processes human beings have conceived scientific theories and devised technological applications that have deeply modified the physical world (Popper & Eccles 1985: 47–48). As for the mind-body problem, this means that mental acts can have effects on the physical world. For example, if you go to a dentist’s because you have a toothache “the caries in your tooth – a material, physico-chemical process – will thus lead to physical effects [the movements of your legs in order to go to the dentist’s etc.]; but it does so by way of your painful sensations, and of your knowledge of existing institutions, such as dentistry” (Popper & Eccles 1985: 36). 3. Only two ‘things’8 that are ontologically distinct can be connected by a relation between cause and effect. Since minds can cause physical events in bodies, and abstract objects of the World 3 can cause mental acts in the minds of human beings, minds and abstract objects are real: “Besides the physical objects and states, I conjecture that there are mental states, and that these states are real since they interact with our bodies” (Popper & Eccles 1985: 36; Popper’s italics). In other words, the existence of mental causation is an empirical proof that the mind is an autonomous emergent level of reality. Materialists, denying the existence of minds as an emergent level of reality distinct from matter, cannot explain how mental acts can be the cause of anything. Only ‘things’ that have an autonomous existence can be the cause of something. Therefore, if materialism were true no backward causation of the mental on the physical would be possible. However, such a mental causation exists, thus materialism is false (Popper & Eccles 1985: 14ff.). 4. A materialist could object that, if backward mental causation really implied a violation of the closure of the physical world, then one should reject mental causation since the closure of the physical world is a well established law of physics. However, Popper replies to such an objection that the closure of the physical world, far from being a well established natural law, “is clearly refuted by the technical, scientific, and artistic achievements
Sandro Nannini
of mankind”, that is, the creativity of mankind proves that the universe is open to novelties (Popper & Eccles 1985: 207). 5. The only dualistic theory of mind that can be compatible with the closure of the physical world is epiphenomenalism: consciousness emerged from a biological basis but cannot react on it. However, epiphenomenalism is an implausible theory since to explain how consciousness emerged from matter in the course of biological evolution is very difficult if one claims that consciousness has no causal efficacy on the physical world and is completely useless for animals in their struggle for life (Popper & Eccles 1985: 74).
. A reply to Popper I think that a materialist should have no objection to the point (1) (except perhaps the use of such expressions like ‘World 1’, ‘World 2’, and ‘World 3’ that are committed by now in an indissoluble way to Popper’s ontological pluralism). Also the point (2), unlike Popper’s opinion, is perfectly compatible with materialism. Why should a materialist challenge that minds exist and have effects on bodies? A materialist will claim, instead, that mental acts must be physical events just because they have physical effects and only what is physical can have effects on the physical world.9 The point (5) is likewise common to Popper and materialists. To tell the truth, one could object that Popper’s argument against epiphenomenalism is not watertight. For consciousness might have no biological function and nevertheless it might have been selected by biological evolution thanks to the fact that it was connected by chance with useful genes in the DNA of some animals (Jackson 1982: 474). However, such a reply makes epiphenomenalism at most possible in principle but not really plausible because it is improbable that a so widespread trait like consciousness has been selected and kept by Mother Nature only as a mere pointless by-product (unless one claims, like Descartes and many other philosophers, that no animals are conscious. However, this seems to be still more implausible when we look at our dog or cat!). Therefore, the points where Popper and materialists really disagree are the points (3) and (4). As to the point (3), Popper’s premise that a mental cause must be distinct from its physical effects belongs to a Humean view of causality that all empiricists and also a lot of materialists accept.10 However, it does not follow from such a premise that a mental cause must necessarily be something of non-physical. On the contrary – as I have already remarked – a mate-
Mental causation and intentionality in a mind naturalising theory
rialist could claim that a mental cause is certainly physical just because it has physical effects. One could object that such a reply misses the focal point of Popper’s argument. Popper does not want to deny that materialists claim that mental acts, being identical with neural events, can cause bodily movements. He wants to affirm that the causal role ascribed in such a way by materialists to mental acts is only seeming and that there is no deep difference between materialism and epiphenomenalism. Let us see why it is so according to Popper. Let us assume that I am raising my arm because I want to do so. According to a causal theory of action,11 my intending to raise my arm (a mental act) is the cause of its rising (a physical event). Is there a set of neural (or anyway physical) events in my brain (and the rest of my body) sufficient to cause the rising of my arm? If the answer is affirmative then the principle of the closure of the physical world is certainly respected but (at least prima facie) my intention has no function whatsoever: my arm rises because my brain gives it the impulse to rise and it would rise even if I did not intend to raise it. My intention plays no causal role (or it is only a useless over-determining cause). In other words, if the physical world is closed then either my intending to raise my arm is epiphenomenal or the movement of my arm is over-determined both by my intention and by the firing of my neurones. However, epiphenomenalism must be rejected for the reasons mentioned above whereas over-determination needs a certain ‘parallelism’ between the mind and the body that is difficult to explain. Why do right neurones fire exactly when I intend to raise my arm? If the firing of my neurones and my mental acts are two different ‘things’ and there is no causal interaction between them, then only heavy metaphysical hypotheses – like e.g. Leibniz’s ‘pre-established harmony’ or ‘neutral monism’ (with its panpsychistic implications) – can explain such a coincidence. Therefore the only viable hypothesis compatible with the closure of the physical world is materialism (in the broad sense defined above): either my intention and the firing of my neurones are just the same event in this case and in all similar cases (‘type-type identity theory’) or my intention is a ‘functional state’ implemented by the firing of certain neurones in this case and possibly by the firing of other neurones in other similar cases (‘materialistic functionalism’, that is, functionalism combined with ‘token-token identity theory’).12 It seems that my intending to raise my arm is the cause of the movement of my arm because it is identical with the neural events that cause that movement. If it is so, is there a deep difference between materialism and epiphenomenalism? According to Popper, there is no difference! For, although materialism is a monistic solution of the mind-body problem whereas epiphenomenalism
Sandro Nannini
is a dualistic theory (Popper & Eccles 1985: 74), both theories seem to grant the mind no true causal efficacy on the control of bodily movements. Epiphenomenalism asserts that mental states are causally ineffective by-products of physiological processes. Materialism seems to recognise that mental acts are causes of bodily movements but in fact it reduces them to be neural events under another description. Therefore mental acts are causes of physical events only ‘per accidens’ (according to the old Aristotelian terminology) in the framework of materialism. Popper concedes to materialists that they admit a certain form of mental causation because, if my intending to raise my arm is identical with certain neural events that cause the rising of my arm, then my very intention causes my bodily movement. However, the mental causation admitted by materialists is only seeming, according to Popper, because my mental acts cause my movements qua neural events and not qua mental acts. For example, if a white billiard ball knocks another ball and moves it then the white ball causes that movement qua knocking object and not qua white object. Similarly, if my mental acts are brain processes then it is true that they can cause the movement of my body but they cause it qua neural events and not qua mental acts. Therefore, according to Popper, materialism, despite all appearances, concedes to the mind no real causal role. According to Popper the only solution of the mind-body problem that is compatible with the undeniable existence of mental causality is the good oldfashioned Cartesian dualism, that is, the ‘interactionistic’ dualism: some physical events cause some mental acts and some mental acts cause some physical events (the roaring of a lion brings about my fear, my fear brings about the escape movement of my legs). Maintaining such a kind of interactionistic dualism, that gives human beings free will in the traditional Christian sense, is the rationale of Popper’s argument. The point (4) is the logical consequence of this choice in favour of interactionistic dualism: according to Popper, if one wants to admit the existence of mental causation one must pay a price and renounce the closure of the physical world. If I seriously assert that my arm rose because I(not my brain) wanted to raise it and I maintain that only this fact grants that my voluntary acts are free and I am morally responsible for their consequences then I must admit that no set of neural events in my brain could be causally sufficient to cause the rising of my arm without the free intervention of my mind. Interactionistic dualism and closure of the physical world are not compatible. A similar incompatibility cannot be extended anyway to the constant correlation between mental and neural events asserted by Popper at the point (1). According to Popper’s pluralistic perspective, no set of neural events of my
Mental causation and intentionality in a mind naturalising theory
brain can be a causally sufficient condition of the rising of my arm before (and independently of) my decision to raise it if my act is really voluntary and free. However, it is perfectly possible, also for a dualist, that my free acts of will have always neural correlates provided such neural correlates are not their causes but their effects: certain neurones fire because I want to raise my arm and not vice versa. Popper’s pluralism, unlike Descartes’ dualism, is compatible with a perfect correlation between mental and neural events. To sum up, Popper’s ontological pluralism seems to grant human beings the capability of freely and voluntarily acting, in accordance with the traditional Christian image of the man, without renouncing the theoretic framework inspired by modern science except the violation of the law of the closure of the physical world (that Popper does not anyway consider a dogma – as we shall see later). Marvellous reconciliation of Christian tradition and modern science! However, Popper’s argument against materialism is not so strong as it seems to be. Firstly, it presupposes a conception of mental causation (point 3) that can be challenged. Secondly, the price paid to renounce the closure of the physical world (point 4) is much higher than Popper thinks it is. As to the existence of mental causation, it is not true that mental causation is not compatible at all with materialism or it is compatible only in a seeming form. Let us see why it is so. A Humean concept of causality like that accepted by Popper needs the logical independence of a cause and its effect but such a logical independence does not necessarily imply an ontological independence as well. The two forms of independence coincide only if one presupposes that causal explanations are necessarily ‘transparent’, that is, only if one thinks that, if A is the cause of B under a certain description of A and B, then it remains such a cause under all possible descriptions.13 For example, let us assume that M is my intending to raise my arm, P is the rising of my arm, N is the neural correlate of M, and → means the cause-effect relation. If the sentences M = N (materialism) and M→P (mental causation) are true and the latter sentence is transparent (i.e., it admits the substitution of co-referential terms salva veritate) then one can deduce N→P from the previous sentences, that is, that the rising of my arm is caused by the firing of certain neurones. However, N→P seems either to take away any causal efficacy from M or to put a difficult question about the possibility condition of an overdetermination case where P is caused both by M and N. According to Popper the only way to avoid both problems seems to be a combination of mental causation with interactionistic dualism: M→N→P (my intending to raise my arm
Sandro Nannini
causes a certain neural event that, in its turn, causes the rising of my arm). This solution, claiming that M causes N (M→N), presupposes that M and N are two different ‘things’: mental causation presupposes dualism. However, there is no decisive reason that urges us to see M→P as transparent, that is, to see relations of mental causality as extensional relations that belong to the world independently of the description of their terms. On the contrary, most sentences that contain psychological terms are opaque.14 Why should psychological causal explanations be an exception? For example, one can have good grounds to say ‘She opened the window because she was hot’ although one has not even the faintest idea of the neural events (and other biological processes) that caused the movements of her body necessary to open the window. Moreover her intending to open the window explains those bodily movements (let us name them ‘P’) only if they are described as the action of opening the window (let us name this action ‘A’). Therefore, M→A and N→P are two distinct assertions from a methodological and epistemological point of view, even if M is identical with N from an ontological point of view: ‘she opened the window because she was hot’ sounds otherwise as (and is logically15 independent of) ‘The firing of neurones in the cortex area x caused the contraction of the muscles y and z . Ontological reductionism does not imply methodological reductionism. Materialism does not make psychology useless. Even if mental acts are neural events it might be that psychology is irreducible to neurosciences. This does not certainly mean that, within a materialistic ontological framework, psychology and neurosciences are completely independent: neuroscientific theories set limits to psychological explanations and these limits will become the narrower the more neurosciences progress. It is plausible to assume that if M→A and A = P then there is a (usually unknown) neural event N that it is the neural correlate of M and the cause of P (N→P). For example, if she opened the window because she was hot it is practically sure that the movements of her body, when she opened the window, were led by brain events among which there was also a set of neural processes correlated to her subjective experience of heat. The existence of a neural correlate for each mental act is conceded by Popper as well. However, does M keep its causal power even if M has no ontological autonomy from N? Popper’s answer is negative but the possibility that M→A is opaque invites us to consider again the whole matter. There are three forms of materialism with regard to the relationship between a mental act M and its neural correlate N: (a) identity theory (M = N); (b) materialistic functionalism (M is implemented by N); (c) eliminativism (although no neural event N can be exactly alike a mental act M this is exclusively
Mental causation and intentionality in a mind naturalising theory
due to the fact that ‘M’ is a pre-scientific concept that will be replaced for scientific purposes by ‘N’ sooner or later thanks to the progress of neurosciences). Do these three hypotheses make the existence of mental causation impossible? The answer depends on what form of mental causation one is taking into account. For there are three major forms of mental causation: (a) a mental act causes an action or another mental act (M→A or M→M’); (b) a mental act causes an observable bodily movement (M→P); (c) a mental act causes its neural correlate (M→N). Now, materialism (in all forms of its own) is perfectly compatible with the survival of M→A in everyday life, in ethical or legal reasoning, and may be in scientific research at least as a heuristic instrument. For example, to think that the progress of neurosciences will lead us to say in the future ‘The firing of her neurones in the cortex area x caused the contraction of her muscles y and z instead of saying ‘she opened the window because she was hot’ is absurd! Moreover, the second way of speaking is irreplaceable in the context of moral and legal reasoning because moral and legal norms are formulated in terms of everyday language. Therefore, to speak of voluntary and free actions or of moral and legal responsibility for practical purposes in the context of ethics and jurisprudence would still be perfectly reasonable in a world where materialism became the received view provided most people accepted Hume’s Law (i.e., is-sentences do not imply ought-sentences) and saw ethics and jurisprudence independent of any metaphysical presupposition.16 However, such a logical independence between folk psychology and neurosciences has limits for a materialist. For mental acts and actions are, according to materialists, only re-descriptions (in a form that is practically useful but scientifically unreliable) of neural events and bodily movements. Mental acts are virtual entities that exist so far they are somehow implemented by real physiological processes. Consequently, also the causal relation from a mental act to an action (M→A) exists, in the universe of discourse of folk psychology, as a virtual reality that must be somehow implemented by some causal relation between physiological processes (N→P) although its opacity preserves the former relation from being reducible to the latter relation from an epistemological and methodological point of view. Therefore, such a non-reducibility of virtual causal relations finds a limit in the possibility of their physiological implementation: psychology cannot hypothesise virtual entities or processes or events whose implementations would clash with established laws of natural sciences. However, within such limits, to speak of mental causation from mental acts to actions (M→A) is perfectly compatible with any form of materialism.
Sandro Nannini
Other observations must be made about M→P. For, whereas to say that I am raising my arm because I want to raise it is perfectly correct, to say that my arm is rising because I want to raise it is rather odd. Whereas a lot of evidence gathered from folk psychology (empirical generalisations etc.) can support the statement that certain desires, beliefs, motives, intentions, passions etc. caused a certain action (that is, a certain bodily movement described as intentional), no evidence at all supports the hypothesis that such mental acts (or states etc.) caused a certain bodily movement described as the terminal state of a physiological process. Similar objections can be raised also against the symmetrical hypothesis that a certain neural event causes a certain action (N→A). To sum up, mental acts (or states etc.) cannot cause bodily movements but actions; vice versa neural events cannot cause actions but bodily movements and physiological processes. Therefore, the second form of mental causation (M→P) is not compatible with materialism unless one reads M→P, as it were, de re as following: ‘A certain neural event that I know only under the psychological description M is as a matter of fact (but not qua M) the cause of P’. Instead, not even such a way out exists to make materialism compatible with the third form of mental causation: M→N is an absurd sentence for materialists because M and N are according to them two distinct descriptions of the same event and therefore cannot be the terms of a cause-effect relation since they do not have the required logical independence. Saying that a mental act is the cause of its neural implementation is for a materialist so absurd as saying that Cicero was M. Tullius’ father! To sum up, is the compatibility (within the limits previously mentioned) between materialism and M→A sufficient to claim that materialism is compatible with the existence of mental causation? Or, since to renounce the possibility of other forms (or more extended forms) of mental causation is the price that one has to pay if one wants to be a materialist, is this price so high that it practically implies the renunciation of the existence of mental causality? Materialists must certainly pay a price. For, although mental acts are not deprived of any practical function in a materialistic perspective, nevertheless it is true that a materialist privileges the description given by physics (and other natural sciences) as the ‘right’ description of the world and degrades descriptions that include psychological terms to be only descriptions of virtual realities and sometimes also of illusions: very useful and may be for ever irreplaceable illusions. Nevertheless illusions! For example, I sometimes think (even better, I feel) that I freely raised my arm. However, is this thought (or feeling) reliable if I know that my intending to raise my arm was ontologically identical with
Mental causation and intentionality in a mind naturalising theory
some neural events that were caused in their turn by other previous neural events out of my conscious control? If the only available alternative to materialism (in a broad sense) is interactionistic dualism, is the price paid by materialists higher than the price paid by dualists? Do dualists pay no price? Sure, they pay a huge price! They renounce the closure of the physical world (Popper’s point (4)), that is, they renounce the first law of thermodynamic: the law of the conservation of energy. If a ball is shocked by another ball the shocking ball gives the shocked ball a certain amount of kinetic energy. More generally, if a body acquires a certain amount of energy (under any form whatsoever) another body (or set of bodies), in the universe, loses the same amount of energy transmitted to the former through a causal relation.17 Therefore, since also N, the firing of certain neurones, implies the acquisition of a certain amount of energy, such an amount must be given to such neurones by other parts of the physical world (including other neurones of the same brain) in which changes happen that are the cause of N. If N were not caused by physical events but by the non-physical act M then the energy acquired by the neurones whose firing is equal to N could not come from a physical event: the physical world would be open, that is, the law of the conservation of energy would be violated because the amount of energy acquired by the brain during N would not be compensated by the loss of an equal amount of energy in another part of the physical world (unless a part of the physical world loses the right amount of energy at the right time even if no cause of N obtains in it. However, this would be a sheer miracle!). Popper discussed such an objection to dualism (or pluralism) and gave it a reply articulated in three points (shared by Eccles as well). First of all, dualism can be rescued by quantum mechanics: the fact that the movements of particles are not completely determined by other previous physical events does not violate the first law of thermodynamics. Therefore the mind could determine, in the brain, by means of a procedure of selection what the laws of physics leave undetermined (Popper & Eccles 1985: 540–541). Second, Popper remarked that the law of the conservation of energy might be valid only statistically (p. 541). Third, even if brain processes violated such a law the deviations from it would be so small that “we should not be too worried about a prima facie violation of this law: somehow we may be able to smooth it all out” (p. 542). Well, the first reply is very weak: no one has the faintest idea of how the mind could determine what quantum mechanics leaves undetermined18 and, anyway, it is not clear what kind of room the indeterminacy of quantum mechanics could leave for the mind and its free will since such an indeterminacy disappears at the macroscopic level of the laws of chemistry that are the most
Sandro Nannini
relevant ones for brain processes. The second reply is an appeal to a miracle of the kind mentioned above. The third reply is . . ., well, let us say, at least, that it is not sufficiently justified! The weakness of Popper’s replies is a cue of the huge difficulty that dualists find to avoid the objection that their theory violates the closure of the physical world. They might reply again that also materialists have a big problem: to convince people that their feeling that they have free will and are free agents is an illusion. Well, this is the situation: either one renounces free will from a metaphysical point of view (although one can continue to live as if free will existed from a practical point of view, including one’s reasoning about moral and legal matters) or renounces the validity of a well established law of physics. Such a choice is really difficult and painful. However, the two alternatives are not equally difficult to accept and the scales are tipped in favour of materialism because, whereas materialists can rescue free will for practical purposes through an ‘as-if-hypothesis’, dualists have no way to avoid the violation of a fundamental law of physics, the best achievement of human knowledge.
. Two kinds of epistemological pluralism: H. Putnam and J. McDowell Popper founded his criticism of materialism and his defence of ontological pluralism on the existence of mental causation. Instead, other philosophers who have defended various forms of epistemological pluralism have attacked materialism and naturalism for the opposite reason: according to them the relation between mind and world is not reducible to a causal relation. Two major kinds of epistemological pluralism have been maintained by such philosophers. Let us take as representatives of these kinds of epistemological pluralism Hilary Putnam and John McDowell. Putnam has founded his criticism of naturalism on a form of anti-realism inspired by N. Goodman: every kind of description and explanation of phenomena has its own inner object and there is no reason to believe that the respective objects of distinct kinds of description and explanation belong to a unique world. Descriptions and explanations that include normative and semantic concepts such as ‘rationality’, ‘truth’ or ‘reference’ cannot belong to that image of the world that is called ‘nature’. Distinct and irreducible kinds of description and explanation construct, as it were, distinct images of the world. To wonder whether these different images are compatible and form the image of a unique real world would mean to ask for the solution of a meaningless metaphysical question. It is a metaphysical prejudice to think that one may speak
Mental causation and intentionality in a mind naturalising theory
of an absolute reality that has its own properties independently of any kind of description of it. The mind-body problem becomes a pseudo-problem in such a theoretic framework because mind and body are considered as belonging to two logically independent ‘versions’ of the world that can be both right although they are reciprocally incompatible (Putnam 1981, 1983). However, such a way to avoid the mind-body problem obliges to renounce the uniqueness of the real world and leaves unexplained and unexplainable many obvious cases of correlation between mental acts and physical events: if the ‘world’ of mental phenomena is logically and causally independent of the ‘world’ of physical phenomena why, in normal conditions, does my arm rise whenever I want to raise it? This objection cannot be raised against some other versions of epistemological pluralism such as McDowell’s theory of the mind-world relationship called by him “a naturalism of second nature” (McDowell 1996: 110). McDowell, like Putnam, distinguishes two irreducibly different ways to describe and explain the reality: on the one hand, nature – so far it is the object of natural sciences – is the “realm of law” (that is, the sphere of phenomena covered by natural laws). On the other hand, if for example I am saying that a certain action was right or rational – and more generally I am referring to values, norms, justifications, evaluations, freedom, rationality or other concepts typical of ethics and the study of man or to concepts such as truth or reference typical of semantics – then what I am speaking of belongs to the “space of reasons” (McDowell 1996: 73). Let us assume that I see a cat. Why do I see just a cat and not, for example, a dog? An answer that sounds perfectly sensible is: I see a cat because there is just a cat in front of me. The presence of a certain object in the world is the cause of my having a perception of it in my mind, that is, O→M(O*) (if the cat is named ‘O’ and my perception of it ‘M(O*)’19 ). There is a causal relation from the world to my mind by means of which one can explain why I have a certain perception. Moreover, if, according to a materialistic hypothesis, M(O*) is identical with its neural correlate (let us call it ‘N’) then the causal relation expressed by the sentence O→M(O*) is likely covered by some natural laws and is part of the ‘realm of law’, that is, it is implemented by the physical causal relation expressed by O→N: the presence of a cat in front of me (plus the reflection of light on its body etc.) causes the firing of certain neurones in my retina and this causes a certain set of neural events called ‘N’ in its turn. However, does the fact that M(O*) is caused by the presence of O grant that M(O*), representing O as an O*, represents it as it really is? In other words,
Sandro Nannini
does the truth of ‘O→M(O*)’ grant that O = O* and that thus the semantic relation expressed by the sentence ‘M(O*) 8 O’20 is satisfied? Let us assume that O is a cat but I think it is a squirrel because of insufficient illumination. In that case my perception-of-a-squirrel is caused by the presence of a cat. Thus, the fact that a certain perception is caused by the presence of a certain object does not grant that the perception is true. The causal relation expressed by the sentence ‘O→M(O*)’ is different from the semantic relation expressed by the sentence ‘M(O*) 8 O’. The first sentence can be true and the second sentence not satisfied or vice versa. In the previous example, ‘cat→M(squirrel)’ is true since I see a squirrel because of the presence of a cat whereas ‘M(squirrel) 8 cat’ is not satisfied because my perception-of-a-squirrel actually ‘points’ to a cat. A perception caused by a cat can be the false perception-of-a-squirrel. Therefore, inquiring into the truth of a perception (or also a belief etc., more generally let us say a mental representation) needs arguments and procedures that do not belong to the ‘realm of law’ but to the ‘space of reasons’: justifying something by means of good reasons is conceptually different from explaining something by means of its causes. One can explain the taking place of M(O*) by the causal relation O→M(O*) but one needs reasons of another kind to justify the satisfaction of the semantic relation M(O*) 8 O. For example, I see O as a cat: if O is mewing then this is a good reason to think that it is really a cat and that my perception is true. To sum up, McDowell thinks that no justification by means of reasons that establishes the satisfaction of a semantic relation can be reduced to a causal explanation. Justifications and causal explanations belong to two completely different approaches to reality: the ‘space of reasons’ and ‘the realm of law’. Since the ‘realm of law’ practically coincides with modern science this means that there is a ‘logical space’ concerning the study of man that is in principle out of the scope of natural sciences. McDowell claims the validity of an epistemological pluralism that is not very far from Putnam’s ideas and that is founded, unlike Popper’s ontological pluralism, on the fact that there are some relations from the mind to the world – such as semantic relations expressed by sentences of the kind of ‘M(O*) 8 O’ – that are not causal and are logically different from the causal relations that go from the world to the mind: for example, the causal relation from an object to a perception (O→M(O*)). However, McDowell, unlike Putnam, thinks that the ‘realm of law’ and the ‘space of reasons’ must complement each other in an unitary image of the world. How is this possible? Three solutions have been given to this problem and according to McDowell the three of them are wrong.
Mental causation and intentionality in a mind naturalising theory
The first solution (proposed, for example, by Garrett Evans) is based on the idea that empirical evidence, on the one hand, belongs to the ‘realm of law’ as the raw material that it is passively received through senses and from which we gather our beliefs about the world but, on the other hand, it belongs, at the same time, to the ‘space of reasons’ as the touchstone to evaluate the truth or falsity of our very beliefs. If I see a cat from close up in normal conditions (my sight is good, the illumination is sufficient etc.), my perception of the cat is both the cause of my belief that there is a cat and a good reason to think that my belief is true (McDowell 1996: 50ff.). However, according to McDowell this solution of the problem is erroneous because – as already W. Sellars has shown – it is based on the “Myth of the Given” (McDowell 1996: xiv). A perception has the power to justify a belief only if it is considered in the background of other beliefs and not qua raw material independent of any image of the world whatsoever. Therefore, no evidence is a reason justifying the truth of a belief only thanks to the fact that it is its cause. The second solution (proposed for example by D. Davidson) goes to the opposite direction: empirical evidence belongs to ‘the realm of law’ and can play no role in the ‘space of reasons’, that is, in the justification of beliefs. A belief can be justified only by other beliefs. One’s belief is true only if it is coherent with the rest of one’s other beliefs. Human beings are free to differently interpret, in the light of the whole system of their beliefs, the empirical evidence that they receive from their senses. However, in this way, the systems of beliefs are not any more grounded on empirical evidence and this is very implausible according to McDowell (1996: xvi ff.). The third solution is called by McDowell “bald naturalism” and it is the worst solution according to him (McDowell 1996: 72ff.). ‘Bald naturalism’ completely renounces the ‘space of reasons’ (including one’s freedom to freely interpret empirical evidence in the light of one’s own beliefs) and reduces it to the ‘realm of law’. In other words, the rational justification of the truth of the semantic relation M(O*) 8 O is completely reduced to the causal explanation of the taking place of M(O*) through the causal relation O→M(O*). The fact that a perception of mine is caused by the presence of a cat in front of me is identical, in normal conditions, with a reason justifying the truth of my perception.21 Such a theory is completely wrong according to McDowell because it reduces the semantic relation between a mental representation and its reference to a causal relation between an external object (or event etc.) and its effects in a mind: it identifies the satisfaction of ‘M(O*) 8 O’ to the truth of ‘O→M(O*)’. ‘Bald naturalism’ eliminates the ‘space of reasons’ in favour of the ‘realm of
Sandro Nannini
law’, it is blind to the obvious difference of these two ‘logical spaces’ and denies the existence of human freedom only to be faithful to a groundless scientism. McDowell privileges a fourth solution: the ‘naturalism of the second nature’. For, according to him, the major error of ‘bald naturalism’ is not to search for a link between the ‘space of reasons’ and the ‘realm of law’ but to reduce the former to the latter. Such a reduction can be avoided according to McDowell if the common ground to which both logical spaces belong is ‘nature’ in a larger and more comprehensive sense than the sense privileged by natural sciences. They were born in the xvii century thanks to a theoretic revolution that, rejecting Aristotle’s physics, removed final causes and any kind of anthropomorphism from the concept of nature: nature was completely ‘disenchanted’ and reduced to the ‘realm of law’ where there is no room for human freedom any more. However, if one comes back to Aristotle and generalise his view of human nature (according to which human beings are rational animals) from ethics to epistemology one can plausibly claim that human beings interact with a ‘nature’ that is deeply impregnated and modified by human culture. The ‘nature’ with which human beings interact is their ‘second nature’, that is, it is the ‘realm of law’ plus costumes and habits of thought, different from innate instincts, that human beings acquire through education and, more generally, through their life in a society. Consequently, it is true that human beings are animals and their actions are led by their ‘nature’. However, such a nature, although it is compatible with the realm of law, is not reducible to it because human beings, even if they are animals, are rational and educated animals. Their partially acquired nature is a second nature that adds to the ‘realm of law’, where everything is determined by scientific laws, a cultural dimension open to the normative and semantic properties of the space of reasons where there is room for freedom (McDowell 1996: 78–86). To sum up, McDowell thinks that he has managed, thanks to the concept of ‘second nature’, to propose a form of naturalism that avoids both the scientism of ‘bald naturalism’ and the risk to fall into the ‘supernaturalism’ of ‘rampant’ Platonism: “Second nature could not float free of potentialities that belong to the normal human organism. This gives human reason enough of a foothold in the realm of law to satisfy any proper respect for modern natural science” (McDowell 1996: 84).
Mental causation and intentionality in a mind naturalising theory
. A reply to McDowell: Intentionality naturalised Is the naturalism of second nature a kind of cognitive naturalism? The answer seems to be affirmative prima facie. McDowell has “respect for modern natural science” and the objections previously addressed to Popper do not seem to touch him. For he does not asserts the existence of any backward mental causation that interferes with the closure of the physical world. On the contrary, his epistemological pluralism is founded on the fact that semantic relations are not reducible to causal relations and he seems to claim for the ‘space of reasons’ an autonomy with regard to the ‘realm of law’ similar to the autonomy that also a materialist can concede to the descriptions in terms of mental acts and actions (such as M→A) with regard to the descriptions in terms of physiological processes (such as N→P). However, this similarity is only an appearance. For a materialist thinks that M→A is implemented by N→P and its autonomy is limited by the constraints imposed by such an implementation. On the contrary, McDowell thinks that some properties of the ‘space of reasons’ such as the existence of human freedom make the ‘second nature’ different from the ‘realm of law’. Therefore, McDowell cannot avoid the following dilemma: either the difference between the ‘second nature’ and the ‘realm of law’ is similar to the difference that also a materialist can concede for practical purposes between mentalistic and physiological descriptions of the same physical reality, or human beings, thanks to their ‘second nature’, can modify the course of physical events through their free and voluntary actions. In the first case McDowell would not concede to human beings more freedom than materialists and his ‘naturalism of the second nature’ would collapse into ‘bald naturalism’. In the second case – the option actually chosen by McDowell – his theory violates the closure of the physical world, like Popper’s ontological pluralism, and it collapses into ‘rampant Platonism’. One could object that such a criticism does not touch the main point of McDowell’s argument: ‘bald naturalism’ confuses semantic relations from the mind to the world with causal relations from the world to the mind, that is, bald naturalists can explain only O→M(O*) but they think this is sufficient to justify M(O*) 8 O as well. However, such an objection to ‘bald naturalism’ is not valid because ‘bald naturalism’ is not so bald as McDowell thinks it is! McDowell would be right if no kind of naturalism that identifies nature to the ‘realm of law’ could offer a naturalistic explanation of semantic relations. However, such an explanation exists and it is known under the name of ‘consumer’s semantics’.22 Its essential core – discarding all details and differences among its several formulations –
Sandro Nannini
presupposes the view that human intelligence is a product of biological evolution and finds its roots in the sensorimotor coordination. According to such a view a mental representation is an intermediate step in a causal chain of neurological processes that go from a stimulus to its motor response. It does not matter if the mental representation is a mere perception or an act of intellectual comprehension: if I see a tile falling on my head I shall move out of the way; if I am asked for the name of Plato’s teacher I shall move my mouth (and tongue etc.) in such a way that I shall utter the sound ‘Socrates’. In both cases there is a distal stimulus from the physical world (the fall of a tile, the utterance of certain sounds) that causes, through a causal chain of physical events, certain brain processes that implement a certain mental act. These brain processes (plus the general state of my brain in that moment, including the dispositions I have acquired through previous experience and learning) cause a certain behavioural output. That is, according to another way of speaking, a certain mental representation of mine (plus my whole personality, cognitive capabilities, customs, habits etc.) causes a certain action of mine. The whole process (described in mentalistic terms) is as following: O→M(O*) [+D(O*/O’*)+M’]→A(O*/O’*)→O’
In other words, a certain real object (or event, process, situation etc.) O causes a certain mental representation M(O*) which – combined with the desire D(O*/O’*) that the object O (under the description O*) is changed into an object O’ (describable as an O’*) and with other mental states M’ (including memories, emotions, cognitive capabilities etc.) – causes, in its turn, the action A(O*/O’*). This action is a certain behaviour – described ‘under the intention’23 to change a certain object described as an O* into an object described as an O’* – which causes O’ in the real world. M(O*) is an intentional state (in Brentano’s sense), that is, it is a mental representation that describes O as an O* and therefore is ‘directed’ to O through its content O* (M(O*) 8 O). Such a description throws light on a semantic relation between the mental act M(O*) and the physical object O and it is true if and only if O* = O. Similarly, the action A(O*/O*’) is a bodily movement carried out with the intention to change a certain real object described as an O* into a real object identical with the purpose of the desire D(O*/O’*) to change the object described as an O* into an object describable as an O’*. Therefore the action A(O*/O’*) is successful only if “O’* = O” is true. For example, let us assume that O is a cat, I see it as a squirrel (O*) and I want to capture a squirrel (that is, I want the object that I see as a free squirrel (O*) to become an object that I represent as a captured squirrel (O’*)). I shall capture O intending to capture a squirrel (that is,
Mental causation and intentionality in a mind naturalising theory
I shall carry out the action A(O*/O’*)) but I shall not be successful because I will not capture a squirrel but a cat (that is, I shall change O (a free cat) into O’ (a captured cat) but O’ will not be identical with O’* (a captured squirrel). In this naturalistic perspective, the mental act M(O*) is functionally defined as that effect of O which is also, if it is combined with the desire D(O*/O’*) and other mental states M’, the cause of an action that brings about O’ in the real world. The success of an action depends on the ‘right’ description of the object on which the action operates. Even better the description is ‘right’, that is, it is true only if it contributes (at least usually and in normal conditions) to the success of the action. The success of the action A(O*/O’*) is a truth condition of the mental representation (MO*). The mental representation of O as an O* is true only in the light of the possibility to realise the purpose O*/O’* to change O into O’. Biological evolution selected only those cognitive powers that are useful for survival. Therefore also we, human beings, can nowadays discriminate by our perceptions only those forms, sizes and colours of material objects that were useful for our ancestors in order not to confuse food and poison, preys and plunderers, etc. To ask whether a mental representation is absolutely true (i.e. whether it absolutely ‘corresponds’ to its object) without specifying the purposes with which it is usually connected is absurd. Therefore, the truth condition of the mental representation M(O*), that is, ‘O* = O’, is satisfied only if the success condition of the action A(O*/O*’), that is, O’* = O’, is (at least usually) satisfied as well. It is easy to see now that McDowell is perfectly right when he says that the causal relation O→M(O*) is insufficient to naturalise the semantic relation M(O*) 8 O, but it is wrong when he does not see that a ‘bald naturalist’ has the possibility to naturalise semantic relations thanks to a broader view of mental representations, that is, seeing them both as the effects of a perception and as the cause of an action. In other words, a naturalist can functionally define mental representations as a necessary ring in a causal chain of the following kind: (s) O→M(O*) [+D(O*/O’*)+M’]→A(O*/O’*→O’
To sum up, a naturalist can plausibly claim that the sentence ‘M(O*) 8 O’, expressing a semantic relation, is satisfied (that is, the mental representation M(O*) correctly represents the object O) if and only if (at least usually) the sentence (s), describing a causal chain of natural events, is true. Moreover, (s) describes a virtual process that can exist only if it is implemented by a causal chain of physical events of the following kind:
Sandro Nannini
(s’) O→N+N’+B→P→O’
(N is a neural event that implements (MO*), N’ is a neural event that implements D(O*/O*’), B is the whole state of the brain that implements M’, and P is the bodily movement that implements the action A(O*/O’*). In other words, a naturalist can recognise that (s) is not translatable without a rest into (s’) and that not only (s) but also sentences expressing semantic relations (and other concepts belonging to the space of reasons) are irreplaceable for practical purposes in everyday life and sometimes in the context of scientific psychology as well. However, the possibility of finding out causal chains of natural events that have the same truth conditions as sentences expressing semantic relations gives a cue of how the whole space of reasons, after having been ‘purified’ by (may be practically useful) illusions such as the existence of human free will, might be considered as a virtual reality implemented by an interaction between those animals that are called human beings and their natural environment. McDowell’s error is to have a too ‘intellectual’ view of the human mind. He thinks that the main function of the mind is to represent the world. He does not see that the capability of representing the world has been selected by biological evolution only because of the function that representations have to lead actions. McDowell thinks that semantic relations from the mind to the world can be naturalised only if they are reduced to causal relations from the world to the mind (that is, perceptions): since such a reduction is not possible then semantic relations cannot be naturalised. However, although the premise of this argument is true the conclusion is false. Semantic relations can be naturalised if one does not consider only the causal relations from the world to the mind (perceptions) but the whole cycle of causal relations from the world to the mind and from the mind to the world (actions), that is, if one does not see human beings as creatures whose ‘essence’ is an Aristotelian Bios theoretikòs (life devoted to the disinterested knowledge of the Truth) but as animals (very developed animals indeed, nevertheless animals) that interact with their environment in a repeated causal cycle of this kind: ‘world→perception→action→ (modified) world→ etc.’. If it is objected that neither neurosciences nor scientific psychology have offered so far the faintest idea of how such an implementation of psychological processes of the kind described by (s) could be realised by means of physiological processes of the kind described by (s’) a naturalist can safely respond that this is not true. Many new trends of scientific research show that to consider mental phenomena as higher-order (functional) descriptions of intermediate
Mental causation and intentionality in a mind naturalising theory
brain processes, in the information processing from sensory input to motor output, is very plausible. First of all, artificial neural networks (and the whole trend of ‘connectionism’ inaugurated by Rumelhart and McClelland (1986) in AI and now developed also in ‘artificial life’24 ) seem to be particularly suited for the simulation of brain processes and for giving a cue of how mental representations might be an intermediate stage in processing information from the sensory input to the motor output. For example, a three layer ‘forward’ network, after its ‘training’, codes the input given to the first layer of ‘neurones’ by means of a certain pattern of activation of the ‘hidden neurones’ of the second layer in such a way that the desired output is produced at the third layer.25 The pattern of activation of hidden neurones, at the end of the training, is the result both of the inputs chosen for the training and of the corrections carried out on the network by the external ‘supervisor’ in order to minimise the gap between desired and obtained outputs at each step of the training. If one compares such an intermediate state of hidden neurones in an artificial network with the activity of cortex areas placed between sensory cortex and motor cortex then to see the pattern of activation of hidden neurons as a sort of a ‘proto-mental’ state where the perception of the world (the coding of the input) is merged with the intention to act in a certain way (production of the output) is very plausible. Moreover, the role of the external supervisor might be played, in the interaction between animals and environment, by the latter: if the result of an action is not right the cause of this error (to be found, first of all, in an erroneous coding of the input, that is, in a false representation of reality) will be eliminated either through the death of the animal who carried out the wrong action (genetic selection) or, in more sophisticated animals, through pain (an ‘alarm system’ that prevents the repetition of false behaviour). For that matter, the efficacy of ‘genetic algorithms’ has been proved also by simulations of ‘artificial life’ (Parisi 1999). Moreover, Edelman (1987, 1989, 1992) has sufficiently proved that the brain develops through a mechanism of inner ‘somatic selection’ between neuronal groups that compete to control behaviour. Finally, mathematical and physical theory on complex systems (describable only by means of non-linear equations) have been applied both by psychologists who propose a ‘dynamical approach’ to the study of the mind26 and brain scientists who search for a dynamics of the human brain capable of explaining how human beings can be conscious (Hardcastle 1995; Jibu & Yasue 1995).
Sandro Nannini
Notes . A brief survey can be found in Nannini (2002: 156–203). . Let us call ‘naturalists’ here those who see no deep theoretic difference between ‘explaining’ and ‘understanding’ and claim the so called ‘methodological unity of science’. . Cf. Apel (1979, 1982). Cf. also Nannini (1992). . The distinction between ontological and epistemological pluralism is similar to the distinction, among the opponents of positivists at the end of the xix century in Germany, between W. Dilthey and W. Windelband. The former distinguished Natur- und Geisteswissenschaften (natural sciences and humanities) through their respective objects: Natur und Geist (nature and spirit (mind)). The latter distinguished ‘nomothetic’ and ‘idiographic’ sciences (sciences that find out laws and sciences that study single cases) through their respective methods. See the classical study of Pietro Rossi about German ‘historicism’ (Rossi 1956). . Pluralism includes dualism. For dualists maintain that minds and bodies belong to two different levels of reality whereas pluralists add that cultural abstract objects are a third independent level of reality. . Descartes (1969, i: 190). . Cf. Nannini (2002: 204–207) for a definition of materialism in such a broad sense. . To decide what kinds of ‘things’ belong to the world is the main problem of philosophical ontology: are there, in the world, only individual substances or also universal properties, states of affairs, events, processes etc.? Also dualism can be seen as substance dualism or property dualism. (For a discussion of philosophical ontology with regard to the philosophy of mind see Steward 1997.) It is not clear what kind of dualism (and pluralism) Popper’s theory is. However, I think that the troubles of dualism I am going to discuss are common to all of its forms. . One could object that the most influential form of materialism is ‘eliminativism’ nowadays. However, to think that eliminativists don’t believe in the existence of mental states and especially of consciousness is a big misunderstanding even if such an interpretation of their theory is very common. They do not want to eliminate mental states but the concepts of folk psychology by which such states are falsely described and explained. Cf. P. M. and P. S. Churchland in McCauley (1996) and Nannini (2002: 186–187). . For mental causation see Kim (1998) and its review by Vaas (2002). . The non-causal action theory proposed by those neo-Wittgensteineans (e.g. G. H. von Wright 1971) who see a conceptual link between intention and action is still less in favour of Popper. . Cf. Bechtel (1988), Macdonald (1989), and Nannini (2002: 105–113, 128) about the distinction between type-type and token-token identity theories. . By the distinction ‘transparent/opaque’ I refer, of course, to Quine (1960). . See for example Quine’s famous analysis of propositional attitudes (Quine 1960). . This does not mean – as we shall see – that it is really independent.
Mental causation and intentionality in a mind naturalising theory . Cf. Nannini (1998) for a defence of Hume’s law. For a general survey see Celano (1994). . If it is objected that the concept of cause plays no important role any more in modern physics, the causal relation in question can be replaced by an empirical regularity covered by a natural law (that is expressed in mathematical terms). . Eccles (1989) has suggested a hypothesis based on the interaction between certain ‘open modules’ of the brain and some alleged immaterial entities (“psychones”). However, such a hypothesis is ad hoc and implausible. . O is the real external object whereas O* is the content of my perception. I have added these and just a few other symbols to easier expose McDowell’s theory. . ‘M(O*) 8 O’ is to be read: ‘the mental representation M(O*) refers to O, that is, O is its reference in the real world and it aims to correctly describe O as an O*’. M(O*) is an ‘intentional state’ in the sense of Franz Brentano’s intentionality’. Therefore, ‘M(O*) 8 O’ is satisfied if and only if ‘O = O*’ is true. . The possibility of an error is explained by ‘wayward causal chains’. Even if it is usually the case that ‘O→M(O*)’ it can sometimes happen that it is the case that ‘O→M(O’*). Although the presence of a cat in front of me usually brings about that I see a cat it can happen (for example if the illumination is insufficient) that I mistake the cat for a squirrel. . Cf. especially Millikan (1984), Dennett (1995, 1996), and Carruthers (2000). . This way of considering an action as a certain bodily movement described ‘under the intention’ to realise a certain goal was introduced by some neo-wittgensteineans like G. Anscombe (1957) and G. H. von Wright (1971). . See in Italy especially Parisi (1999). . See for example Churchland (1995). . For the dynamics of complex systems see Kaplan and Glass (1995). For the application of this approach to neurosciences and psychology see Hopkins, Beek, and Kalverboer (1993), Freeman (1999), and Luccio, Ieri and Salvadori (2002).
References Anscombe, G. E. (1957). Intention. Oxford: Blackwell. Apel, K. O. (1979). Die Erklären/Verstehen-Kontroverse in transzendental-pragmatischer Sicht. Frankfurt a. M.: Suhrkamp. Apel, K. O. (1982). The Erklären-Verstehen Controversy in the Philosophy of Natural and Human Sciences. In G. Floistad (Ed.), Contemporary Philosophy. A New Survey Vol. II (pp. 19–49). The Hague: M. Nijhoff. Bechtel, W. (1988). Philosophy of Mind. An Overview for Cognitive Science. Hillsdale, N.J.: Lawrence Erlbaum Associates Inc. Carruthers, P. (2000). Phenomenal Consciousness. A Naturalistic Theory. Cambridge: C.U.P. Celano, B. (1994). Dialettica della Giustificazione Pratica. Saggio sulla Legge di Hume. Torino: Giappichelli.
Sandro Nannini
Chalmers, D. (1996). The Conscious Mind. In Search of a Fundamental Theory. Oxford: O.U.P. Churchland, P. M. (1995). The Engine of Reason, the Seat of the Soul. A philosophical Journey into the Brain. Cambridge, MA: The MIT Press. Dennett, D. C. (1995). Darwin’s Dangerous Idea: Evolution and the Meanings of Life. New York: Simon & Schuster. Dennett, D. C. (1996). Kinds of Minds: Towards an Understanding of Consciousness. New York: Basic Books. Descartes, R. (1969). The Philosophical Works of Descartes. Cambridge: C.U.P. Eccles, J. C. (1989). Evolution of the Brain: Creation of the Self. London: Routledge. Edelman, G. M. (1987). Neural Darwinism. The Theory of Neuronal Group Selection. New York: Basic Books. Edelman, G. M. (1989). The Remembered Present: A Biological Theory of Consciousness. New York: Basic Books. Edelman, G. M. (1992). Bright Air, Brilliant Fire: On the Matter of the Mind. New York: BasicBooks. Freeman, W. J. (1999). How Brains Make Up their Minds. London: Weidenfeld & Nicolson. Hardcastle, V. G. (1995). Locating Consciousness. Amsterdam-Philadelphia: John Benjamins. Hopkins, B., Beek, P. J., & Kalverboer, A. F. (1993). Theoretical issues in the longitudinal study of motor development. In A. F. Kalverboer, B. Hopkins, & B. R. Geuze (Eds.), Motor Development in Early and Later Childhood: Longitudinal Approaches (pp. 343– 371). Cambridge: C.U.P. Jackson, F. (1982). Epiphenomenal Qualia. Philosophical Quarterly, 32, 127–136. Jibu, M. & Yasue, K. (1995). Quantum Brain Dynamics and Consciousness. An Introduction. Amsterdam-Philadelphia: John Benjamins. Kaplan, D. & Glass, L. (1995). Understanding Nonlinear Dynamics. New York: Springer. Kim, J. (1998). Mind in a Physical World: An Essay on the Mind-Body Problem and Mental Causation. Cambridge, MA: MIT Press. Luccio R., Ieri, C., & Salvadori, E. (2002). Contro l’integrazione. Giornale italiano di psicologia, 39(2), 281–287. Macdonald, C. (1989). Mind-Body Identity Theories. London–New York: Routledge. McCauley, R. N. (Ed.). (1996). The Churchlands and their Critics. Oxford: Blackwell. McDowell, J. (1996). Mind and World. With a New Introduction. Cambridge, MA: Harvard University Press. Millikan, R. (1984). Language, Thought and Other Biological Categories. Cambridge, MA: The MIT Press. Nannini, S. (1992). Cause e Ragioni. Modelli di Spiegazione delle Azioni umane nella Filosofia Analitica. Roma: Editori Riuniti. Nannini, S. (1998). Il Fanatico e l’Arcangelo. Saggi di Filosofia Analitica Pratica. Siena: Protagon. Nannini, S. (2000). Cognitive naturalism in the philosophy of mind. In S. Nannini & H. J. Sandkühler (Eds.), Naturalism in Cognitive Sciences and the Philosophy of Mind (pp. 41– 62). Frankfurt am Mein: Peter Lang. Nannini, S. (2002). L’anima e il Corpo. Un’Introduzione Storica alla Filosofia della Mente. Roma-Bari: Laterza.
Mental causation and intentionality in a mind naturalising theory
Parisi, D. (1999). Mente. I Nuovi Modelli della Vita Artificiale. Bologna: Il Mulino. Popper, K. R. (1994). Knowledge and the Body-Mind Problem. In Defense of Interaction. London–New York: Routledge. Popper, K. R. & Eccles, J. C. (1985). The Self and Its Brain. New York: Springer. Putnam, H. (1981). Reason, Truth and History. Cambridge: C.U.P. Putnam, H. (1983). Why reason can’t be naturalized. In H. Putnam (Ed.), Realism and Reason. Philosophical Papers, Vol. III (pp. 229–247). Cambridge: C.U.P. Quine, W. V. O. (1960). World and Object. Cambridge, MA: The MIT Press. Rumelhart, D. E. & McClelland, J. L. (Eds.). (1986). Parallel Distributed Processing: Exploration in the Microstructure of Cognition. Cambridge, MA: The MIT Press. Rossi, P. (1956). Lo Storicismo Tedesco Contemporaneo. Torino: Einaudi. Steward, H. (1997). The Ontology of Mind. Events, Processes and States. Oxford: Clarendon Press. Vaas, R. (2002). Problems of mental causation – whether and how it can exist, on line . Von Wright, G. H. (1971). Explanation and Understanding. Ithaca, N. Y.: Cornell University Press.
Chapter 6
The envious frog Marco Salucci Florence, Italy
Rana rupta et bos In prato quondam rana conspexit bovem, et tacta invidia tantae magnitudinis rugosam inflavit pellem: tum natos suos interrogavit an bove esse latior. Illi negarunt. Rursus intendit cutem maiore nisu, et simili quaesivit modo, quis major esset. Illi dixerunt “bovem”, novissime indignata, dum vult validius inflare sese, rupto iacuit corpore. Phaedrus
.
Identity theory and mental causation
The after-behaviourist era in philosophy of mind was opened by Feigl’s, Place’s and Smart’s identity theory in late fifties.1 Ever since, several objections to such a theory and many other theories of mind have appeared on the scene. Objections such as these were founded on Leibniz’s law, on the multiple realization thesis, on modal arguments, and on the knowledge argument; and as for the theories one can mention anomalous monism, functionalism, neo-dualism, property dualism, epiphenomenalism, supervenience and non-reductive physicalism. However, because of both the effectiveness of identity theorists’ replies and the suitability of the theory itself to facing the problems, identity theory is not at all out of date. On the contrary, in view of the current debate on mental causation, identity theory provides the best account; maybe the unique satisfactory account of mental causation.
Marco Salucci
Since early identity theorists’ essays, the mind-body relation was excluded to be a mere correlation. Mere correlation should be, in fact, consistent with theories in competition with the identity one, mainly dualism: the major opponent to the identity theory. Moreover correlations are consistent with causal relations too. Place wrote: What is, therefore, that leads us to say that two sets of observations are observations of the same event? It cannot be merely the fact that the two sets of observations are systematically correlated such that whenever there is lightning there is always a motion of electric charges. There are innumerable cases of such correlations where we have no temptations to say that the two sets of observations are observations of the same event. There is a systematic correlation, for example, between the movements of the tides and the stages of the moon, but this does not lead us to say that records of tidal levels are records of the moon’s stages and vice versa. We speak rather of a causal connection between two independent events or processes. (Place 1956: 33)
And Smart added that these [sensations] should be correlated with brain processes does not help, for to say that they are correlated is to say that they are something “over and above”. You cannot correlate something with itself. (Smart 1959: 53)
Identity and correlation were thought, therefore and thereafter, mutually exclusive. Moreover correlation was viewed as a direct route to dualism. But in 1966 Lewis introduced in philosophy of mind the principle of causal closure of the physical world: any cause of physical effects is a physical thing too.2 Since mental states have physical effects – i.e. bodily effects – they are physical events. With this principle in the identity theory’s paraphernalia, any possible sliding from correlation to dualism is blocked. But, to be sure, identity is not correlation, and the identity theory is not a mere correlation theory. Correlation is weaker than identity, and identity is the strongest form of correlation. Place claims that correlation does not suffice for identity, Smart stresses that correlation is not compatible with identity since you cannot correlate something with itself. Since Place’s, Smart’s and Lewis’ early formulations of the identity theory the debate in philosophy of mind has continually enriched itself. Here I shall deal with two subjects, mental causation and the argument of knowledge, to revisit the place correlation occupies – or better, should occupy – in formulating the identity theory. Nowadays mental causation receives most of the attention in philosophy of mind.3 Given the causal closure principle, the way a theory accounts of mental
The envious frog
causation is tantamount to testing if it is a physicalist theory or not. Indeed the causal closure principle is, at present, the main available resource for physicalism. Non-reductive physicalism, or supervenience physicalism, as well as functionalism, fail to match the causal closure of the physical world and hence they are wrong. There is, in fact, a great amount of literature about the questions concerning causal overdetermination and causal exclusion in mental causation; such questions have, however, their source uniquely in non-reductive physicalism. No satisfactory solution is available yet. With his usual skill, Kim outlines the state of the art:4 [. . .] the causal powers of M are wholly derived from the causal powers of its realizer P: this instance of M causes whatever its physical realizer P causes. Since whatever causes P to be instantiated also causes M to be instantiated thereby, it follows that the given instance of M enters into exactly the same causal relation that the corresponding instance of P enters into: something is a cause or effect of the M-instance if and only if it is a cause or effect of the P-instance. There are no new causal powers that magically accrue to M over and beyond the causal powers of P. The approach to mental causation last pictured, therefore, is essentially reductionist: no new causal powers emerge at higher levels, and this goes against the claim of the emergentist and the not-reductive physicalist that higher-level properties are novel causal powers irreducible to lower-level properties. (Kim 1996: 232) If we reject reductionism, we are not able to see how mental causation should be possible. (Kim 1996: 237)
Therefore a direct and simple strategy is to bypass the questions and adopt an utterly reductive physicalism, as Place’s and Smart’s one was. No doubt the causal closure principle directly gets reductive physicalism. Mental states can have physical effects since they are physical states.
. The knowledge argument teaching So far so good (maybe). Reductive physicalism has to face the problem of qualia (but also any other theory of the mind-body relationships has to face it), particularly the knowledge argument, which stresses the subjective feature of mental states and the private access to them.5 The replies to the knowledge argument are oriented to utter rejection,6 and supporters and opponents of such an argument are sharply divided. My view is that the knowledge argument is not conclusive; nevertheless, I think it is useful to state a striking feature of the
Marco Salucci
mind-body relationships – a feature physicalism must take as seriously as to reconsider the status of “correlation” within the identity theory. The privateness of mental states is by itself no trouble for reductionism,7 since that feature depends on the particular physical realization of mental states: my own mental states are physically realized in my own brain states, and so it is as regards yours.8 But privateness constrains the way mental states can be known: as there is no way to know mental states objectively, there is no way to obtain a public knowledge of them. In order to appreciate the consequences of the private access to mental states for physicalism, it is useful to consider two ways the knowledge argument may be refuted. First, it would be falsified if the physical or physiological description of a certain sensation (e.g. the sensation of red, as in the Mary Gedankenexperiment by Jackson 1982) resulted in having the sensation, too. Second, the knowledge argument would be falsified if having a sensation resulted in getting a physical or physiological description of the sensation, too. Clearly, both conditions are impossible to fulfil. Fulfilling the former would be like tasting some food simply by reading the recipe; fulfilling the latter would mean knowing which neurons are firing upon seeing red. Therefore the knowledge argument is not conclusive, since it relies on two different ways of knowledge: “knowledge by acquaintance” and “knowledge by description”, to use Russell’s terms, or knowing by experiencing and knowing by theorizing. According to a similar distinction between different meanings of “knowledge”, physicalists have faced the challenge shouted by the knowledge argument. But, notwithstanding they answered the attacks in several and articulated ways, Chalmers (1996) claims that the identity theory is wrong presenting Nagel’s and Jackson’s styled arguments (among others). Clearly, the reductionists’ replies were not convincing, because the knowledge argument is grounded on the fact we really have different access to mental and brain states. Below I provide seven points characterizing the identity theory to show why Nagel’s and Jackson’s styled arguments are not conclusive but also to show why our intuition in favour of their effectiveness is so strong.9 1. Some brain states are mental states (obviously, there are brain states which are only brain states such as those controlling physiological functions). 2. All mental states are brain states. 3. Subjective experiences (“what-it-is-like-to-be” kind experiences) are mental states. 4. Subjective experiences are brain states, from 2 and 3.
The envious frog
5. We have experience of our mental states, but we have not experience of our brain states. 6. From 5, clearly, does not follow that mental states cannot be brain states. 7. Nagel’s and Jackson’s styled arguments are based on 5, then they are invalid because of 6. Point 7 needs some remarks. To my view, the knowledge argument requires us to experience how our brain states are identical to (or become) mental states. That is because of the fact that the argument is falsifiable only if our perceiving mental states were also experiencing our brain states. But we cannot have such an experience because the brain has not a special function or organ detecting its own states, qua brain states. Therefore, the knowledge argument simply points out this natural fact: the lacking of such a special organ or function. Nagel’s and Jackson’s antireductive argument is similar to attacking the tenability of optical theory because of we do not have eyes that see infrared radiations. Lacking such an ability, we can however elaborate a physical theory of electromagnetic radiation. Maybe something similar could happen in the mind-body case. Nevertheless, point 5 explains why the knowledge argument is as attractive as it really is, it is founded on the special access to mental states we have: the private-subjective-privileged one. In 1714 Leibniz presented (in the Monadology, sec. 17) the prototype of each future form of the knowledge argument. One is obliged to admit that perception and what depends upon it is inexplicable on mechanical principles, that is, by figures and motions. In imagining that there is a machine whose construction would enable it to think, to sense, and to have perception, one could conceive it enlarged while retaining the same proportions, so that one could enter it, just like a windmill. On this supposition, one should, when visiting it, find only parts pushing one another, and never anything by which to explain a perception. Thus it is in the simple substance, not in the composite or in the machine, that one must look for perception.
The visitor in the windmill observes nothing but the physical properties of the parts (forms, sizes, motions and so on) and their relations. But no explanation of mental states can possibly be deduced from these physical properties. No matter how complex the machine (windmill or brain) is, nothing about it being observed reveals if it has mental states. Hence materialism is false for there is no possible description of merely physical properties that can account for mental phenomena. Churchland (1995) manages such an argument by stressing that the visitor finds nothing concerning “mentality” since he does not know what
Marco Salucci
he is looking for. But, in my view, the point of the argument (apart from its ontological flavour) is that “objective” observation is a inadequate for mental properties. Mental states are only subjectively observable, whereas the visitor in the windmill, as a biologist studying bats’ sonar system and Mary’s knowing the physiology of colour perception, is performing an objective observation. So the observation of physical properties is inadequate to reveal mental properties not just because the latter cannot be properties of a physical system, but because they are not objectively observable.10 This outcome is not against physicalism. From the knowledge argument the conclusion to be drawn is tantamount to stating the private access to mental states; no result from the argument regards the ontological features of mental states. However, the argument poses the question as to how it is possible to obtain empirical evidence for physicalism. Also McGinn (1989) presented a Leibniz’ styled argument. Let us assume there is a brain property in virtue of which mental states spring from the brain: [. . .] does the introspective faculty reveal [such a] property [. . .]? Can we tell just by introspecting what the solution to the mind-body problem is? Clearly not. We have direct cognitive access to one term of the mind-brain relation, but we do not have such access to the nature of the link. Introspection does not present conscious states as depending upon the brain in some intelligible way. We cannot therefore introspect P. [. . .] the property of consciousness itself (or specific conscious states) is not an observable or perceptible property of the brain. You can stare into a living conscious brain, your own or someone else’s, and see there a wide variety of instantiated properties – its shape, colour, texture, and son on – but you will not thereby see what the subject is experiencing, the conscious state itself. Conscious states are simply not, qua conscious states, potential objects of perception: they depend upon the brain but they cannot be observed by directing the senses onto the brain. You cannot see a brain state as a conscious state. (McGinn 1989: 533)
As my diagnosis of the knowledge argument suggests, I agree with McGinn. However, since McGinn derives from his remarks a “mysterian” point of view about the mind-body problem – with which, on the other hand, I disagree – some clarifying remarks are needed. My first remark concerns the notion of “relation”, involved in the property by which, according to McGinn, mental states should spring from the brain. I wonder whether relations are observable: can we observe the bookstands-on-a-desk relation? But the point is that no relation of identity is observable. According to identity theorists, mental and brain states are just the
The envious frog
same, thus no relation over and beyond identity is observable.11 That is because no relation over and beyond identity does exist. Some Wittgensteinian-minded reader might note that the observation of my own brain states would not reveal whether such states are states of mine. He might add my own observed brain states do not reveal they are in relation with mental states. That is all right, given that mental states are just brain states. No relation (between mental and brain states or between mental states and me) can be observed, and the reason is simple: no such relation exist. Nonetheless, there is a sense of “observing a correlation” relevant for identity theory too, since it points to our tentative and preliminary way to approach the identity thesis – before stating it. For instance, we can say “the properties of water are in relation with those of a given compound of H2 O” as a tentative summary of some experiences aiming at the final justification of the sentence “water is H2 O”. Second, McGinn emphasizes that the observation of the mind is not, by itself, the observation of the brain and vice versa. For introspection – which is the only way to know mental states – is not observation,12 and observation – which is the usual way we know physical objects – is not introspection. Thus, as regards the terms to be “related”, i.e., mind and brain, introspecting the former is not observing the latter and observing the latter is not introspecting the former. This is no obstacle to support identity theory, since we are familiar with similar situations: it is a matter of fact that Don Diego is nowhere to be found each time we see Zorro, Don Diego being identical with Zorro. The issue concerns rather which criterion we should adopt to accomplish identifications. It is not always true that, in order to accomplish an identification, there must be a simultaneous observation of the items to be identified,13 a case in point being that of water and H2 O. The way to identify mental and brain states is not of the same kind we follow when we observe two instances of the same object. Chemists identified H2 O with water by observing the properties of two samples of the same matter. Briefly, they identified two components of water with different properties (weight, density and so on) and named them “H” and “O”; it was then checked that H has properties, say, x, y, z and O has x’, y’, z’. Through suitable reactions, H and O form a special kind of matter. Be “W” the name of the matter so obtained and x”, y”, z” the properties of W. It turned out that two parts of H react with one part of O to form W. But x”, y”, z” are the same properties of the water. Therefore W is water. Since W is compounded by two parts of H and one part of O, water is too. It follows that water is H2 O. In the case of the mind-body problem we cannot observe two samples of the same material. To be sure, that might depend on facing two different ma-
Marco Salucci
terials. But, given the very special status of private access to mind, the impossibility to observe two samples of the same material composing mental and brain states is due to observation being available for brain states only. This notwithstanding, the identity claim may be justified by a theory explaining it. Place first advanced such a possibility, as in his text quoted above. As for which way to discern between causality and identity – both of them consistent with correlation – we could resort to the time gap between mental and brain states. According to McGinn, we are cognitively closed entities. But that does not follow from his remarks quoted. It only follows that we are perceptively closed: we cannot observe mental states and introspect brain states. Observing and introspecting are ways of perceiving. Thus, even assuming the perceptive closure as ascertained – as I do –, cognitive closure would not follow. Here, by “cognitive” I mean “theoretical”, i.e., I refer to the human ability to build and linguistically express theories. Our theories exceed our perceptive skills: quantum mechanics and relativity theory are clear examples. Therefore, even though the mind-brain identity were unachievable by observation, it would be theoretically justifiable. Neither observation nor introspection alone does all the work; what we must demand is that we can correlate the results of observing and introspecting, in the way we shall see below. If identity theory has to be justified itself in compliance with the standards posed by McGinn’s and other Leibniz’ styled arguments, then any other reduction statement has to comply with the same standards. Therefore a sentence as “light is electromagnetic radiation” ought to be justified by perceiving light as electromagnetic radiation. That is impossible: the identification is obtained in a theoretical way. What is paradoxical about the knowledge argument is that it should be proposed again even if mental states were really reducible to brain states. Even in the case future neuroscience will explain why and how mental states spring from brain states, a neuroscientist like Mary might still tell she did not know colours before her releasing and so. . . Thus, is the whole argument sound? Or does the lack of potential falsifiers imply there is something wrong with it?
. Correlation and identity I have already suggested that the real core of the knowledge argument is in the subjective access to mental states. I added that there is no problem for physicalism since a physicalistic explanation is available for it: the lack of a special
The envious frog
faculty to detect brain states. It is time to elucidate the constraints this view implies for the mind-body identity. Feigl noted the “auto-cerebroscope” would be the most direct confirmation of the identity theory. (Note the difference between Feigl’s stance on correlation and Place’s and Smart’s one.) The most direct confirmation conceivable would have to be executed with the help of an autocerebroscope. We may fancy a “compleat autocerebroscopist” who while introspectively attending to, e.g., his increasing feelings of anger [. . .] would simultaneously be observing a vastly magnified visual “picture” of his own cerebral nerve currents on a projection screen [. . .] According the identity thesis the directly experienced qualia and configurations are the realities-in-themselves that are denoted by the neurophysiological descriptions. This identification of the denotata is therefore empirical, and the most direct evidence conceivably attainable would be that of the autocerebroscopically observable regularities. (Feigl 1958: 456–457)
In spite of his reductionistic scepticism, Brandt agreed with Place and Smart about the correlation thesis. The identity theory is a stronger theory than the correspondence hypothesis. There is no empirical evidence for it beyond the evidence for the correspondence hypothesis. Unless there are other advantages in the identity theory, we are in a better-entrenched position to support the correspondence theory and leave open questions of identity, refraining from commitment. (Brandt 1960: 63)
The importance of this remark parallels that of Feigl’s considerations. I wish to argue, following Feigl, that a correlation/correspondence hypothesis is the only viable way to obtain the identity. For the knowledge argument requires the identity theory to be grounded on empirical evidence beyond mind-body correlations. Since supporters of the knowledge argument believe there is no such possible evidence – otherwise they would not support it –, they impose an oppressive burden on physicalism. Identity theorists, on the other hand, must face the fact that empirical evidence beyond correlations is not available. If private access is an essential feature of knowing-mental-states, then evidence beyond correlation will be unavailable forever: this is what already emerges from the discussion about the auto-cerebroscope imagined by Feigl, and it is confirmed by the present debate on the knowledge argument. As noted above, though introspection cannot detect brain states and observation cannot detect mental states, we can use both to assess a correlation. If we introspect a given mental state while observing a given brain state, we are
Marco Salucci
justified in claiming their identification. Sure, this is a common sense solution but it is the only one available (in fact, one more satisfactory than most “philosophical” solutions). Obviously, I am not claiming that identity is just correlation. Nor am I reducing identity theory to a mere correlation theory. What I mean is that the correlations can justify the identity statements. Moreover, evidence about correlation is the only empirical evidence in favour of identity theory. To demand justification for identity over and beyond the correlations might mean to look for a fancy glue cementing mental and brain states. No such glue can be found if mental and brain states are identical. Concerning the causal correlations, Schlick wrote that no “metaphysical glue” is available to correlate causes and effects over and beyond the relations between phenomena established by physics. The fact the correlations are consistent with theories other than identity theory is not a mark of untenability of identity. Simply, it is a further instance of underdetermination of theories with respect to empirical evidence. Given two theories compatible with the same empirical evidence, both may be considered to be true. (As a consequence, the objection that correlations provide too weak a support to identity, as they are consistent with dualism, does not prove that identity theory is wrong.) As it is usual in the scientific enterprise, we can choose one among competing theories by appealing to different criteria from the success in explicating empirical data. As regards identity and dualism – both assuming the correlations between mind and body – here are some criteria: – – –
– –
identity theory better merges with the entire corpus of science than dualism does; identity theory explains mental causation, whereas dualism (and nonreductive physicalism) cannot; identity theory explains the existence of correlations (as a trivial consequence of the identity between mental and brain states); on the contrary, dualism (and not-reductive physicalism) assumes the correlation as a brute fact; identity theory explains correlations without requiring any explanation in turn;14 identity theory is more intelligible than dualism; notwithstanding that the notion of identity is far from being clear, it is less problematic than the notion of interaction between physical and not-physical entities; the history of the mind-body problem after Descartes is the history of the failure of interactionistic dualism; moreover, splitted solutions were found, through
The envious frog
– – –
– –
either non-dualistic interactionism (monism in all its different forms) or non-interactionistic dualism (parallelism in all its different forms);15 identity theory requires a simpler ontology than the dualism; identity theory explains the empirical results and the effectiveness of neuroscientific research and clinical practice; dualism implies an expensive revision of the overall scientific picture: the causal closure of the physical world, together with mass-energy invariance principles ought to be abandoned. This would risk getting rid of natural science; however tentative, some reductive strategies are already available, as those advocated by the “neural networks” approach;16 identity theory can appeal to the “simplicity of explication” principle.17
. Getting rid of multiple realizability Correlations provide justification for identity by showing biconditional correlations between mental and brain states. These correlations are constant and systematic, as seventeenth century’s interactionists and parallelists knew. Given the causal closure principle, if we really think correlations are not enough for identity we all ought to be parallelist, not interactionist. But, chiefly, the correlations are empirically detectable, as witnessed by neurosciences and related clinical practice. If each time I am in a given mental state I am also in a given brain state and vice versa, then a constant and systematic correlation of biconditional form occurs. My being in a given mental state is detectable only by my own (verbal) reports – because of the private access. But my being in a given brain state is detectable by some physical device such as, e.g., EEG, TAC, PET and whatever better device will be found in the future. Nonetheless, biconditional correlations of P ↔ M form are not for free: the price to pay is getting rid of the multiple realization thesis. That means getting rid of the supervenience notion too. Actually, multiple realization and supervenience assume a simple correlation as P → M for, being M multiple realizable in , P is a sufficient, but not necessary, condition for M. On the contrary, if P ↔ M, then P is a sufficient and necessary condition for M. Although the multiple realization thesis has received wide consensus, there is still some scepticism.18 There are several reasons to cast doubt on the multiple realization thesis. Below I shall make some of these explicit. My scepticism about multiple realizability does not exclude that organisms differing from human beings can really have mental states. It excludes that a human being’s men-
Marco Salucci
tal state can be realized in a physical structure fundamentally different from that of human beings. I sympathize with a form of species-specific identity or local reductionism. Identity theory has been challenged in several ways by advocates of multiple realization. Here are the prominent ones. Non-human beings have (as animals) or may have (as artefacts or aliens) mental states. Since these beings do not have a human brain, then mental states cannot be identical with brain states, and this applies to human beings too.19 If this argument is empirically viewed then its value is very doubtful. For we do not know any beings having mental states which do not have also a nervous system. In fact, each being to which we attribute mental states does have a nervous system. On the other hand, if the argument is conceptually viewed, as a logical possibility, my reply is that it derives from our ability to imagine perceiving or thinking beings devoid of a nervous system. But what we can imagine essentially depends on what we know and, thereby, on what we believe to be possible. As we have lost, in a recent past, the possibility of imagining metals devoid of electrical conductivity, so we could lose, because of the future discoveries of neurosciences, the possibility of imagining mental states not realized in nervous systems.20 Mental states do not fluctuate in a vacuum: they undergo several kinds of constraints. Biological, bodily, perceptive, motor, ecological and other kinds of constraints affect, and in an essential way, the mental states of organisms.21 The theories of mind, which agree on multiple realizability (and supervenience), do not provide a specific psychology of human beings. They implicitly claim to be able to provide a universal psychology, a psychology suitable to every actual or possible being. Sure, multiple realizability’s advocates have charged the identity theory as being chauvinistic; but, were it possible to achieve any universal psychology, we ought to have some standards to detect mental states. These standards cannot be grounded other than in human beings’ psychology. Thus, some degree of anthropocentrism is inevitable; once anthropomorphism is (duly) avoided, traces of anthropocentrism necessarily remain. The conviction that we can construct an indeed universal psychology falls into what I name Phaedrus’s fallacy, i.e., the fallacy to attribute human mental states to non human beings, as in Aesop’s and Phaedrus’s tales or in Disney’s cartoons. Local reductionism avoids this fallacy by attributing a specific psychology to each specific kind of organisms. Neurobiological research has found empirical evidence showing the role of material structures (biochemical, biological, neural) constraining the ways animals can perceive and react in their own environment. Animals’ ways of
The envious frog
perceiving are often very different from ours. Bats’ sonar system; dogs’ sharp olfactory skill; snakes’ sensitivity to infrared radiation; bees’ skill to return to hive and detecting wave lengths in regions of the electromagnetic spectrum we cannot access; octopuses’ and cuttlefishes’ reactions to particular geometrical features of objects: all these are only samples of the “biodiversity” in perceiving.22 This evidence also shows that, if animals have mental states, these are very different from ours; for sure, they have qualia different from ours and, possibly, they have “propositional attitudes” very different too. Given that frogs perceive flies only if flies are moving and not by their shape (according to frogs, a black dot moving is a fly and a stationary fly is not a fly), it follows that, if frogs had “propositional attitudes” or beliefs about flies, frogs would not take recognize flies as in-sects. We can also build some fictitious cases in order to show how a universal psychology is hard to be reached. E.g., the two-dimensional characters in Abbot’s novel Flatland cannot get any notion of rotation in 3space. They cannot understand the notion of “tie one’s tie”. Knots (as usually intended) require 3-space; understanding what “knot” means requires it too.23 The moral of the story is opposite to the functionalist’s one. We should try to understand how mental states essentially depend on physical structures, rather than neglect the physical structures as the multiple realization thesis implies. Another charge against the identity theory concerns the description of mental states in physical terms. Supporters of the multiple realization subscribe the full autonomy of psychology from physical sciences. It would be absurd to use physical or physiological descriptions to grasp psychological states, just as it would be absurd to refer to physical properties of coins in order to describe economic phenomena.24 Unless this objection is viewed as grounded on some actual ontological difference between physical and mental states – in such a case, however, the above considerations about the actual multiple realizability of human mental states apply –, it can only justify a sort of methodological antireductionism, not an ontological one. But methodological antireductionism is consistent with ontological reductionism. Maybe, methodological antireductionism must be assumed, since it is impossible, as a matter of fact, to describe, say, the battle of Waterloo by describing the brain states of the soldiers taking part in the event. Finally, according to many scholars there is the difficulty of proving that the same mental state corresponds to the same brain state, even during a single organism’s life. Sure, given the very large number of possible connections among neurons (about 1013 ), the human brain has sufficient resources to be in unrepeatable states. That means, however, that mental states are unrepeatable too: no two occurrences of, e.g., thinking of Vienna, may be exactly identical.
Marco Salucci
So, if we are entitled to refer to mental states by a finite lexicon, we are also entitled to claim the identity between mental and brain states. Mental states are unrepeatable because brain states are unrepeatable, too. Supervenience of duplicates of P with the same M-supervenients is an extreme case (depending on atom-by-atom duplication of such a complex organ as the brain). If the advocates of supervenience and multiple realization thesis present an extreme case as an ordinary case, a fortiori identity theorists may use it as an extreme case.
. Between universal and individual The multiple realization thesis is opposite to the idea – attributed to the identity theory – that the property characterizing mental and brain states must be unique for each occurrence of those states. But the identity theory does not need such an idea. Identity theory can co-exist with the one-many correspondence relation claiming that what characterizes a type of state is a family of similar, rather than identical, properties. “Pain” does not characterize the very same property shared by each pain occurrence, but a set of similar properties. A mental state can vary just as brain states can vary. In order that mental states to be occurrences of the same mental state, it is necessary, however, that the range of variation be limited. The range is set by a prototypical property the single occurrences resemble (at different degrees). We are familiar with such a procedure in categorizing objects: exactly identical trees do not exist, notwithstanding we can identify trees according to their likeness. Sentences referring to a certain mental state do not refer to a unique property shared by each occurrence of the mental states, but to a sort of likeness constrained within a definite range. As noted above, the occurrences of a mental state (and, according to the identity hypothesis, of a brain state too) are similar with regard to some relevant features, but not exactly identical. If it is implausible to think that various occurrences of one and the same mental state have the same brain correlates, this is because, as mental states, they have various features all around the core of the characterizing ones. But the prototypical features of a given mental state are identical to the prototypical features of the corresponding brain state. That means there are, at least, typical, or better, prototypical features of brain and mental states ensuring their systematic correlation. Prototypical features ensure we can recognize the occurrences of a given mental state as occurrences of the same kind of mental state. The token-identity theories (anomalous monism and identity of occurrences) deny not only type-identity, but also exclude pro-
The envious frog
totypical features. So they prevent themselves from identifying any mental state beyond its different occurrences. Connectionism favours the above approach. An identical neural network is not in the same state each time it is performing the same task. The same neural network reacts in a very similar but not identical way to the same stimulus (assume the optimal response to be 100, actual responses can be, e.g., 98.3, 98.7, 99.1 and so on). Properties of neural networks like that just mentioned are empirical evidence supporting both my response against the one-many correspondence objection and the prototypical features of mind-brain states hypothesis.
. Concluding remarks: Featuring mental states Finally, I wish to suggest that from the fact that we have experience of our mental states but we do not have experience of our brain states (as stated in section 2), it is possible to trace a research line concerning the definition of mental states, as a step to contribute to the solution of the mind-body problem.25 Def. Mental states are brain states which are owned by individuals and appear in some specific form (e.g., qualia) only to their owners.
Qua appearances, mental states differ from brain states of which they are appearances, so I think a way is at hand to save mental realism. Reductionism has been accused of collapsing into eliminativism26 since if mental states were brain states they could not have an autonomous existence as mental states. But, because of the fact only some brain states are also mental states (as also noted in par. 2, that is the reason the mind-body identity is not reversible), the set of the mental states is a subset of the brain states’ one. Therefore, some feature of mental states is needed in order to distinguish the members belonging only to the subset from those belonging to the whole set. The definition above provides such a distinguishing feature and thus blocks the sliding from identity to eliminativism about entities. We are, however, dealing with brain states still, because mental states are a way of being of brain states, as appearances. The term “appearance” does not mean “deceive”; it refers to a way-of-being proper of brain states. It means Erscheinung, not Scheinung at all. Identity theory may be refuted, therefore, only by showing that no Erscheinung can be a property of physical systems. On the other hand, as noted at the beginning of this paper, claiming mental states are physical states is the only available way to save both mental causation and causal closure.
Marco Salucci
Notes . Cf. Place (1956), Feigl (1958), Smart (1959). . Cf. Lewis (1966, 1972). . But the best analysis of mental causation (with respect to the antireductionism issue too) is available in Kim (1998). . In the following, M stands for mental and P for physical. . Cf. Nagel (1974), Jackson (1982, 1986). . See Alter (1998), Bigelow and Pargetter (1990), Conee (1994), Churchland (1985), Dennett (1991), Foss (1989), Horgan (1984), Lewis (1983), Loar (1990), Lycan (1990, 1995, 1996), McMullen (1985), Mellor (1993), Nemirow (1980), Papineau (1993), Pereboom (1994), Raymont (1999), Salucci (2000, 2002), Stemmer (1989), Teller (1992), Tye (1986), van Gulick (1993). . But it is possible to build an antireductionist argument via Leibniz’ law taking privateness as an exclusive feature of mental states. Objections related to Leibniz’ law are early discussed in Place (1956) and Smart (1959); a classical study of Leibniz’ law is Grelling (1936). Of course, the effectiveness of arguments grounded on Leibniz’ law depends on the applicability of Leibniz’ law itself. First, Leibniz’ law is applicable to strict identity, whereas mind-body identity does not appear to be of this kind – in the very least for the reason that not every brain state is also a mental state. Second, arguments appealing to Leibniz’ law may be question begging: our knowledge of the brain is so scant that, for some properties, we do not know whether they are real mental properties or only such believed. In spite of its antireductionist stance, functionalism is allied with physicalism against the arguments grounded on Leibniz’ law (cf., e.g., Fodor 1968; Lycan 1972). No wonder that, since the rising of functionalism, those arguments have faded. . Cf. Salucci (1994: 243–244), now in Salucci (1997: 130); for the related notion of selfconnected system, see Churchland (1995). . Since 1988, Jackson thinks his own argument is not right. But he stresses it is very attractive. Cf. Jackson (1988). . The claim that all physical systems are objectively (i.e., intersubjectively) observable is question begging. . Smart’s remark (quoted above) on correlation and identity is here in point: one cannot correlate something with itself. . A classical discussion of this topic is in Anscombe (1957). . However brief, Place’s (1956: 30) remarks on “is” in the sense of a definition and “is” in the sense of a composition deal with such questions. . As it happens in general for identity, cf. Causey (1972). . Historical aspects of the mind-body problem are discussed by Salucci (1997). . As proposed, e.g., in Churchland (1986). Cf. also Churchland (1995). . As that discussed in Smart (1959) concerning Gosse’s anti-Darwinian theory.
The envious frog . Among the rare critical remarks about the multiple realization thesis cf. Bechtel and Mundale (1999), Churchland (1986), Enc (1983), Hooker (1981), Richardson (1979), Pineda (2002), Salucci (1996), Zangwill (1992). . Cf., e.g., Putnam (1973). . Cf. Salucci (1996: 78). . Remarks along the same line are available in Salucci (1996). Cf. also Hopkins [in print], Nunez and Freeman (1991), Lakoff and Nunez (2000), Peruzzi (2002). . Cf. on bats: Griffin (1958, 1962); on bees: Carricaburu (1977), Jander and Voss (1963), von Frisch (1955); on octopuses: Wells and Wells (1957), Wells (1959), Young (1983); on cuttlefishes: Boulet (1958, 1977); on frogs: Lettvin (1951); on cats: Hubel and Wiesel (1959, 1962); on primates: Gross (1972), Perrett (1985); on the sense of touch: Mountcastle (1957). I am referring to old studies to stress the fact they were already available while functionalism was spreading. . About the role played by acting and moving in perceiving skills cf. Foerster (1982), Viviani (1990), Viviani and Stucchi (1992). Much earlier, Condillac (1754) stressed the fundamental role of bodily motions in building perceptive representations. Poincaré (1903) studied the issue with regard to three-dimensional sight. Of course classical studies on the “sense-motory ring” are those of Piaget. Relationships between environment and organisms ought to be considered too, by focusing on the perceptive and cognitive skills an organism can have (cf. Gibson 1979; a criticism of Gibson is argued by Fodor & Phylyshyn 1981; for a defense, see Turvey 1981). What sense organs are sensitive to is determined by ecological resources: organisms can only detect environmental features favouring/preventing their survival (Lorenz 1973). Moreover, even the architecture of sense organs is influenced by general aspects of the physical environment: no cell could perceive infrared and ultraviolet radiation because of the harmfulness of these wave lengths. Atmosphere filters such radiations allowing the development of organisms fit to see visible light (Wald 1964). . Cf., Putnam (1973) and Fodor (1975). . Such a line seems to be consistent with Tye (2000). . E.g., by Searle (1992). Eliminativism is often viewed as aiming at ontological economy, whereas its main purpose is to eliminate theories of a given kind.
References Alter, T. (1998). A limited defence of the knowledge argument. Philosophical Studies, 90, 35–56. Anscombe, G. E. M. (1957). Intention. Oxford: Blackwell. Bechtel, W. & Mundale, J. (1999). Multiple realizability revisited: linking cognitive and neural states. Philosophy of Science, 66, 175–207. Bigelow, J. & Pargetter, R. (1990). Acquaintance with Qualia. Theoria, 61, 129–147. Boulet, P. C. (1958). Contribution à l’Étude Expérimentale de la Perception Visuelle du Movement chez la Perche et la Seiche. Paris: Mein Museum.
Marco Salucci
Boulet, P. C. (1977). Vision et comportaments chez les cephalopodes. Journal of Psychology, 1, 91–106. Brandt, R. B. (1960). Doubts about the identity theory. In S. Hook (Ed.), Dimensions of Minds (pp. 57–67). New York: New York University Press. Carricaburu, P. (1977). La vision des colours chez les insectes. Journal of Psychology, 1, 107– 127. Conee, E. (1994). Phenomenal knowledge. Australasian Journal of Philosophy, 35, 136–150. Causey, R. L. (1972). Attribute-identities in microreductionism. The Journal of Philosophy, 69, 407–422. Chalmers, D. (1996). The Conscious Mind. Oxford: Oxford University Press. Churchland, P. M. (1985). Reduction, qualia and the direct introspection of brain states. Journal of Philosophy, 82, 8–28. Churchland, P. M. (1989). Some reductive strategies in cognitive neurobiology. Mind, 95, 275–309. Churchland, P. M. (1995). The Engine of Reason: The Seat of the Soul. Cambridge, MA: MIT Press. Churchland, P. S. (1986). Neurophilosophy: Toward a Unified Science of the Mind-Brain. Cambridge, MA: MIT Press. Condillac, E. B. (1754). Traité des Sensations. In G. Le Roy (Ed.), Oeuvres Philosophiques de Condillac, Vol. 1 (pp. 219–314). Paris: PUF. Dennett, D. C. (1991). Consciousness Explained. Boston: Little Brown. Enc, B. (1983). In Defence of the identity theory. Journal of Philosophy, 80, 279–298. Feigl, H. (1958). The ‘mental’ and the ‘physical’. In H. Feigl & M. Scriven (Eds.), Concepts, Theories and the Mind Body Problem, Minnesota Studies in the Philosophy of Science, Vol. 2 (pp. 370-497). (New ed. with a Postscript, 1967. Minneapolis: University of Minnesota Press.) Minneapolis: University of Minnesota Press. Fodor, J. A. (1968). Psychological Explanation. New York: Random House. Fodor, J. A. (1975). The Language of Thought. New York: Crowell. Fodor J. A. & Pylyshyn, Z. W. (1981). How direct is visual perception? Some reflection on Gibson’s ‘ecological approach’. Cognition, 9, 136–196. Foerster, H. von (1982). Observing Systems. Seaside, CA: Intersystem Publications. Foss, J. (1989). On the logic of what it is like to be a conscious subject. Australasian Journal of Philosophy, 67, 305–320. Frisch, K. von (1955). Dancing Bees. London: Methuen. Gibson, J. J. (1979). The Ecological Approach to Perception. Boston: Houghton Mifflin. Grelling, H. (1936). Identitas indiscernibilium. Erkenntnis, 6, 252–259. Griffin, D. R. (1958). Listening in the Dark. New Haven, Connecticut: Yale University Press. Griffin, D. R. (1962). Echo-Ortung der Fledermäuse. Naturwissenschaften, 15, 169–173. Gross C. G., Rocha-Miranda, C. E., & Bender, D. B. (1972). Visual properties of neurons in infratemporal cortex of the macaque. Journal of Neurophysiology, 35, 96–111. Hooker, C. (1981). Towards a general theory of reduction. Dialogue, 20, 496–529. Hopkins, B. (in press). Understanding motor development: Insight from dynamical systems perspectives. In A. F. Klaverboer & A. Gramsbergen (Eds.) Handbook on Brain and Behaviour in Human Development. Dordrecht: Kluwer.
The envious frog
Horgan, T. (1984). Jackson on physical information and qualia. Philosophical Quarterly, 34, 147–183. Hubel, D. H. & Wiesel, T. N. (1959). Receptive fields of single neurons in the cat’s striate cortex. Journal of Physiology, 148, 574–591. Hubel, D. H. & Wiesel, T. N. (1962). Receptive fields, binocular interaction and functional architecture in cat’s visual cortex. Journal of Physiology, 160, 106–154. Jackson, F. (1982). Epiphenomenal qualia. Philosophical Quarterly, 32, 127–136. Jackson, F. (1986). What Mary didn’t know. The Journal of Philosophy, 83, 291–295. Jackson, F. (1988). Postscript on qualia. In F. Jackson (Ed.), Mind, Method and Conditionals (pp. 76–79). London: Routledge. Jander, R. & Voss, C. (1963). Die Bedeutung von Streifenmunstern für das Formensehen der roben Waldameise. Zeitschrift für Tier-Psychologie, 20, 1–9. Kim, J. (1996). Philosophy of Mind. Boulder, CO: Westview Press. Kim, J. (1998). Mind in a Physical World. Cambridge, MA: MIT Press. Lakoff, G. & Nunez, R. (2000). Where Mathematics Comes From: How the Embodied Mind Creates Mathematics. New York: Basic Books. Lettvin, J. Y., Maturana, M. R., McCulloch, W. S., & Pitts, H. (1951). What the frog’s eye tells the frog’s brain. Proceedings of the Institute of Radio Engineers, 47, 172–205. Lewis, D. K. (1966). An argument for the identity theory. The Journal of Philosophy, 63, 23–35. Lewis, D. K. (1972). Psychophysical and theoretical identifications. Australasian Journal of Philosophy, 50, 249–258. Lewis, D. K. (1983). Postscript to: Mad pain and Martian pain. In D. K. Lewis (Ed.), Philosophical Papers, Vol. 1 (pp. 130–132). Oxford: Oxford University Press. Loar, B. (1990). Phenomenal states. In J. Tomberlin (Ed.), Action Theory and Philosophy of Mind (pp. 81–118). Atascadero: Ridgeview. Lorenz, K. (1973). Die Rückseite des Spiegels. München: Piper Verlag. Lycan, W. J. (1972). Materialism and Leibniz’ law. The Monist, 56, 276–287. Lycan, W. (1990). What is the ‘subjectivity’ of the mental? In J. Tomberlin (Ed.), Action Theory and Philosophy of Mind (pp. 109–130). Atascadero: Ridgeview. Lycan, W. (1995). A Limited Defence of Phenomenal Information. In T. Metzinger (Ed.), Conscious Experience (pp. 243–258). Tucson: University of Arizona Press. Lycan, W. (1996). Consciousness and Experience. Cambridge, MA: MIT Press. McGinn, C. (1989). Can we solve the mind-body problem? Mind, 98, 349–366. Rep. In Block N., Flanagan O., & Guzeldere G. (Eds.), 1997, The Nature of Consciousness (pp. 529– 542). Cambridge, MA: MIT-Bradford Books. McMullen, C. (1985). Knowing ‘what it’s like’ and the essential indexical. Philosophical Studies, 48, 211–233. Mellor, D. (1993). Nothing like experience. Proceedings of the Aristotelian Society, 93, 1–16. Mountcastle, V. (1957). Modality and topographic properties of single neurons of cat’s somatic sensory cortex. Journal of Neurophysiology, 20, 408–434. Nagel, T. (1974). What is it like to be a bat? The Philosophical Review, 83, 435–450. Nemirow, L. (1980). Review of T. Nagel’s Mortal Questions. Philosophical Review, 89, 475– 476.
Marco Salucci
Nunez, R. & Freeman, W. J. (Eds.). (1991). Reclaiming Cognition. The Primacy of Action, Intention and Emotion. Thorverston: Imprint Academic. Papineau, D. (1993). Philosophical Naturalism. Oxford: Blackwell. Pereboom D. (1994). Bats, brain scientists, and the limitations of introspection. Philosophy and Phenomenological Research, 54, 315–329. Perrett, D. A., Smith, P. A. J., Potter, D. D., Austin, A. J., Head, A. S., Milner, A. D., & Jeeves, M. A. (1985). Visual cells in the temporal cortex sensitive to face view and gaze direction. Proceedings of Royal Society B, 233, 293–317. Peruzzi, A. (2002). Il contenuto della forma logica. In R. Lanfredini (Ed.), Forma e contenuto (pp. 211–222). Milan: LED. Pineda, D. (2002). The causal exclusion puzzle. European Journal of Philosophy, 10, 26–42. Place, U. T. (1956). Is consciousness a brain process? The British Journal of Psychology, 47, 44–50. (Rep. in Lycan, W. J. (Ed.), Mind and Cognition. A Reader, 1990. Cambridge, MA: Blackwell.) Poincaré, H. (1903). L’espace et ses trois dimensions. Revue de Métaphysique et de Morale, 2, 281–301, 407–429. Putnam, H. (1973). Philosophy and our mental life. In H. Putnam (Ed.). (1975). Mind, Language and Reality: Philosophical Papers, Vol. 2 (pp. 291–303). Cambridge, MA: Cambridge University Press. Raymont, P. (1999). The know-how response to Jackson’s knowledge argument. Journal of Philosophical Research, 24, 113–126. Richardson, R. C. (1979). Functionalism and reductionism. Philosophy of Science, 46, 533– 558. Salucci, M. (1994). Il Dibattito tra Funzionalismo e Materialismo nella Filosofia della Mente Anglosassone Contemporanea, Ph.D. Dissertation. Florence: University of Florence. Salucci, M. (1996). Materialismo e Funzionalismo nella Filosofia della Mente. Pisa: ETS. Salucci, M. (1997). Mente/Corpo. Firenze: La Nuova Italia. Salucci, M. (2000). La coscienza è riducibile a stati cerebrali? Iride, 30, 367–376. Salucci, M. (2002). L’argomento della conoscenza. In R. Lanfredini (Ed.), Forma e contenuto (pp. 33–50). Milan: LED. Searle, J. (1992). The Rediscovery of the Mind. Cambridge, MA: MIT Press. Smart, J. J. C. (1959). Sensations and brain processes. Philosophical Review, 68, 141–156. Stemmer, N. (1989). Physicalism and the argument from knowledge. Australasian Journal of Philosophy, 67, 84–91. Teller, D. (1992). A contemporary look at emergence. In A. Beckermann, H. Flohr, & J. Kim (Eds.), Emergence or Reduction? Prospects for Notreductive Physicalism (pp. 139–153). Berlin: De Gruyter. Turvey, M. T., Shaw, R. E., Reed, E. S., & Mace, W. M. (1981). Ecological laws of perceiving and acting: In reply to Fodor and Pylyshyn. Cognition, 9, 237–304. Tye M. (1986). The subjectivity qualities of experience. Mind, 95, 1–17. Tye, M. (2000). Consciousness, Colour and Content. Cambridge, MA: MIT Press. Van Gulick, R. (1993). Understanding the phenomenal mind. Are we all armadillos? In M. Davies & G. Humphries (Eds.), Consciousness (pp. 137–154). Oxford: Blackwell. Viviani, P. (1990). Principi di organizzazione nella coordinazione percetto-motoria. Sistemi Intelligenti, 2(2), 149–212.
The envious frog
Viviani, P. & Stucchi, N. (1992). Biological movements look uniform: Evidence of motorperceptual interactions. Journal of Experimental Psychology, 18, 603–623. Wald, G. (1964). The receptors for human colour vision. Science, 145, 1007–1017. Wells, M. J. (1959). Functional evidence for neurons fields representing the individual arms within the central nervous system of Octopus. Journal of Experimental Biology, 36, 501– 511. Wells, M. J. & Wells, J. (1957). The function of the brain of Octopus in tactile discrimination. Journal of Experimental Biology, 34, 131–142. Young, J. Z. (1983). The Distributed Tactile Memory System of Octopus. Proceedings of the Royal Society B, 218, 135–176. Zangwill, N. (1992). Variable reduction not proved. Philosophical Quarterly, 42, 214–218.
Chapter 7
Knowing what it is like and knowing how Luca Malatesti University of Stirling
Introduction Physicalism in philosophy of mind is the doctrine that mental states and processes, if they are something, are physical states and processes. Notoriously, Frank Jackson has attacked physicalism with the knowledge argument.1 This argument involves two main assumptions. The first one concerns Mary, a scientist who knows all there is to know about the physical nature of colours and colour vision lacking any previous colour experiences as she is confined in a black-and-white laboratory. The second assumption states that when Mary is released and sees a coloured object, let us say a red rose, she learns something she did not know before having that experience. Jackson concludes that there are non-physical facts concerning the occurrence of non-physical properties or qualia. This paper does not consider whether the knowledge argument is successful. Instead, I argue that the ability reply to the knowledge argument fails. The central assumption of this objection is that, by having colour experiences, Mary acquires a set of abilities rather than new beliefs as required by the knowledge argument. Against the ability reply, I maintain that on her release Mary acquires new beliefs about objects looking the same colour. As a preliminary, I show, against an important criticism of the knowledge argument, that we can make sense of what Mary knows about colour experience when she is in the black-and-white laboratory.
Luca Malatesti
.
Mental states in a physical world
In recent years, many philosophers of mind have promoted different versions of physicalism. These thinkers share the ontological hypothesis that the mind is part of the natural world studied by physics and the other natural sciences such as chemistry, biology and neuroscience. Their formulations of physicalism differ in dealing with two main interrelated issues. First, there are diverging views about the ontological relationship between mental and physical properties. The supporters of type identity theory have promoted the idea that types of mental states are identical to types of physical states.2 Others have endorsed the weaker claim that the particular instantiations, or tokens, of mental types are identical to physical tokens.3 Second, there are different positions about how the exhaustive scientific account of the mind should be related to the study of the physical world. Type identity theorists have claimed that psychology is completely reducible to the study of the physical properties of the brain. This means that all the explanations available in psychology can be couched in physical (neurological) terms. Others have argued that, although tokens of mental properties are identical to certain physical tokens, psychology is explanatorily autonomous and cannot be reduced to physics or neuroscience.4 Despite these differences, it appears that contemporary physicalism is based upon two assumptions. The first assumption is that mental states are causally responsible for physical changes that constitute our behaviour. In particular some physicalists have maintained that every mental state can be completely individuated in terms of a certain causal role.5 A causal role is given by a set of conditions that specify the stimuli that cause the mental state, and the behaviours and others mental states that are caused by this mental state. For example, pain can be regarded as the state, caused by certain dangerous stimuli, that causes certain behaviours of avoidance and certain mental states such as the desire to avoid such stimuli. The second assumption shared by many physicalists is what can be called the hypothesis of the causal completeness of physics.6 This is the idea that all physical effects are caused only by prior physical histories. From these two assumptions it follows that mental states are physical states. The knowledge argument is not meant to challenge the premise of the causal completeness of physics. In fact, Jackson maintains that qualia, besides being non-physical, are epiphenomenal properties of experience.7 While these properties or their instantiations can be caused by physical modifications of the brain, they cannot cause any physical change in the brain or in our body. The main conclusion of the knowledge argument is that a specification of a
Knowing what it is like and knowing how
causal role or a description referring to physical properties of the brain cannot accommodate conscious experiences. The quale of an experience, i.e. what it is like to have that mental state, is a feature that escapes the characterisations of mental states proposed by physicalists.
. Mary’s scientific knowledge A central assumption of the knowledge argument is that there is a type of knowledge concerning colour experiences that requires having these mental states (or closely related ones) as a necessary condition for its possession.8 I call the supposed knowledge that satisfies this condition knowledge of what it is like to have a conscious experience or knowledge of what it is like for short.9 Before examining what Mary supposedly comes to know on her release, it is relevant to consider what she knows while she is still in the laboratory. It has been maintained that if the knowledge argument is used against any complete (future or possible) scientific knowledge of colour vision, it fails simply because we cannot grasp what this science might be.10 The idea that on release Mary learns something she did not know before is based on our limited understanding of what a complete science can achieve. Moreover, we cannot exclude that having this knowledge will cause the appropriate mental states or experiences that supposedly are required for knowing what it is like.11 However, the argument might be used to investigate specific research projects of the type currently carried forward in the scientific study of colour vision. Psychophysics is one of the contemporary disciplines involved in the study of colour vision. The central task of this science is to determine the number of colours and their mutual relations in terms of subjects’ responses to measurable energy changes in the environment. The results of these empirical investigations are modelled by quality spaces. Points in the space stand for the colours objects look to have; distances between these points represent relations of similarity between these colours. The colour solid is an example of quality space that represents the ordering of experienced colour along three dimensions: hue, saturation and lightness or brightness (see Figure 1). Scientists and philosophers agree that the construction of quality spaces is central in the contemporary scientific attempts to describe colour experience.12 Quality spaces provide the fundamental data that neuroscience should explain. For instance, it is known that the dimensions of variations in the colour solid are generated by the activity of certain neural mechanisms known as opponent processors.13 So, if Mary’s scientific knowledge is of the same type we presently
Luca Malatesti White
Green Lightness
Yellow Saturation
Grey
Blue
Hue
Red
Black
Figure 1. The colour solid
have, we can assume that the colour quality space will play a main role in her understanding of colour experiences. Although the determination of the quality space for colour experience is far from being complete, it seems that we can use this model to understand what Mary knows before her release.14 While she is still in the laboratory, she thinks about sensory qualities presented in colour experience as positions in a colour space.15 For instance, Mary’s notion of red is that of the sensory quality that an object x looks to have for a normal observer in certain specified conditions and that is completely characterised by its position in a complete colour quality space:16 (RD) x is red, i.e. x is the quality that colour experience represents things looking red to have, if and only if x is more similar to y that is orange, than z that is blue, (and so on and so forth by considering all the relations of similarity represented in the colour space).
Although other colour terms are involved in this relational definition, like orange, blue, yellow, each of them can be eliminated and replaced by similar relational descriptions by means of the logical technique known as “Ramsification”.17 Thus we can assume that Mary, before her release, thinks about sensory
Knowing what it is like and knowing how
qualities of colour experience in terms of relational descriptions based on the complete colour space. In addition, she can also explain them in terms of brain mechanisms. Having made clear what Mary’s scientific knowledge is taken to be, it is time to turn back to the evaluation of the knowledge argument. It seems to be a requirement of Jackson’s argument that knowing what it is like is knowledge of facts. First, the occurrences of “knowledge” in the premises of the knowledge argument should refer to the same type of knowledge. In fact the argument is based on the following modus tollens. By having a colour experience Mary knows what it is like to have that mental state. This is something that she did not know while she was still in the laboratory. When she was there she knew all the physical facts. Therefore, she comes to know non-physical facts. Second, Mary’s complete scientific knowledge seems to be propositional, being knowledge of physical facts. Thus, willing to avoid a fallacy of equivocation, the upholders of the knowledge argument have to maintain that knowledge of what it is like to have a conscious experience is propositional.
. The ability reply David Lewis and Laurence Nemirow analyse Mary’s epistemic progress as a form of knowing how.18 According to a shared view in philosophy, this type of knowledge does not require any knowledge of facts. Abilities like knowing how to swim, catch a ball, or play musical instruments, constitute a type of knowledge. But it seems that possessing these abilities does not require any belief or relation with propositions that can be true or false. Having abilities is just to be capable of doing certain sorts of things in the appropriate way. Moreover, it seems that this type of knowledge is not about anything in the world. Although in order to exercise an ability certain conditions in the world have to be satisfied, it seems that someone who knows how to do something is not representing the world to be a certain way. Both replies seem to be based on two main theses. First, knowing what it is like is equated to certain abilities. Thus, according to Nemirow: Knowing what it is like is the same as knowing how to imagine having the experience. (Nemirow 1990: 495)
Moreover, David Lewis provides a more comprehensive list of abilities:
Luca Malatesti
Rather, knowing what it is like is the possession of abilities: abilities to recognize, abilities to imagine, abilities to predict one’s behaviour by imaginative experiments. (Lewis 1983: 131)
The second thesis is that having these abilities does not require any knowledge of facts or how the things are or, as Lewis puts it, the “elimination of nonphysical possibilities”. The ability reply provides an elegant model of the situation represented by the knowledge argument. It attempts to reconcile a physicalist outlook with the intuition that there is a type of knowledge essentially connected with having experiences.19 In addition, it is difficult to deny that, in many cases, by having experiences we acquire certain abilities. However, the identification of knowledge of what it is like with a set of abilities can be challenged.20 There are certain beliefs that Mary acquires when she has colour experiences that she cannot have while she is still in the laboratory. To see this, we have first to consider what might be the content of Mary’s supposed new knowledge.
. Resisting the ability reply When considering colour experiences, we can distinguish between the experiencing and its object. For example, if I see a tomato, the tomato, which is the thing seen, is different from my seeing it, which is an experience of this object. Given this distinction, philosophers have identified different senses in which we can think about the ascription of qualitative properties involved in colour experience. On the one hand, the qualitative feature is the colour ascribed to the object experienced. Thus, “being red”, for instance, is a property of the tomato.21 On the other hand, seeing the colours of objects might be explained in terms of properties that we ascribe to the experiencing itself. Thus it has been maintained that experiences have certain properties in virtue of which coloured things look in a certain way. Red tomatoes look red to us because our experience, our seeing them, has certain properties. It seems that Jackson intends to defend the existence of qualia as irreducible properties of colour experiences.22 Without considering the plausibility of the idea that knowledge of what it is like concerns properties of the experience, I will investigate whether Mary learns about the properties that coloured objects look to have. So, let us assume that Mary, before her release, has the possibility of studying, without having any colour experience, a patch, let us call it A, that looks red in certain conditions C to normal observers. By using the appropriate
Knowing what it is like and knowing how
instruments and investigating other subjects’ discriminatory responses Mary can determine that A looks red. In fact, she can determine the position that the sensory quality presented by A occupies in the qualitative space. Thus, Mary comes to believe that A looks red. In order to prove that Mary acquires certain new beliefs by having colour experiences, we have to devise two possible situations. In the first one, outside the laboratory, we show her, who is now a normal observer in conditions C, a red patch B. Now she is not allowed to study the patch by considering its physical properties or normal observers’ responses and brain states while seeing it. I argue that Mary is not able to know in these circumstances that A and B look the same colour. Her belief that the patch A looks red does not help. Her knowledge of the position of looking red in the system of relations of similarity cannot be applied to B. In seeing B, Mary has very limited relational information about how B looks. The only relations of similarity and difference she might actually detect are those between the way in which this patch looks and the background.23 It might be argued, however, that Mary might know which relational property is involved in her experience of B from her knowledge of its neural correlates. This requires that Mary can recognise just by seeing B that she is in a certain brain state. Some have found this assumption plausible. For example, Patricia Churchland has claimed that as an engineer can see the world according Newtonian physics, Mary can “see” her internal world via the utopian neuroscience.24 The conceptual framework provided by a mature neuroscience will provide a way to directly introspect brain states. So, for example, we will achieve ’direct, self-conscious introspection’ of such properties as spiking frequencies of neural aggregates when confronting perceptual stimuli. This idea has found some critics.25 But even if we concede it, how does Mary acquire the ability to “see” the experiences she is having as certain physical states? She has never had any previous colour experience, so it seems that she could not learn to relate her colour experiences to her brain state. Mary cannot come to believe that B presents the same relational property as A, therefore she cannot judge that they look the same colour. Let us consider a second situation in which Mary comes to know that A looks red when she is in the laboratory. After her release, we show her the patch A and then the patch B. I argue that now she can recognise that A and B look the same colour. Lewis has suggested that if Mary sees a red object, then she gains the ability to recognise an experience as of a red object. This is a recognitional ability concerning the experiencing. However, it seems plausible to assume that this ability implies also the ability to recognise by sight when objects look red.
Luca Malatesti
This ability concerns the object of the experience. By having the experience of A she is enabled to recognise that B looks the same colour as A. Therefore, it seems that she can also judge that A and B look the same colour. The final step is to show that the belief about looking the same colour that Mary acquires by having experiences is not one she might have in the laboratory in virtue of her scientific knowledge. Here she can believe that A and B look the same colour only if she believes that A and B present a sensory quality that has the same position in the quality space. But when on her release she sees A and B she can believe that they look the same colour without believing that B presents a sensory quality that satisfies a relational definition. The notion of looking the same colour that she has in the laboratory differs from the one that she acquires by having colour experiences. Important things have to be left unsaid here. The distinction between beliefs concerning looking the same colour according the scientific concepts and those based on having colour experience is epistemic. It is at the level of the concepts that someone has to posses in order to have these beliefs. Whether different facts should correspond to these different beliefs has to be investigated. Moreover, I can only notice, without investigating further, two differences between the beliefs involved in knowing what it is like assumed in Jackson’s argument and the belief that Mary acquires. According to Jackson, knowledge of what it is like involves beliefs concerning monadic facts about properties of experience. I have argued that having colour experiences is essential to acquire beliefs about the relation of looking the same colour between objects.
. Conclusion To sum up, we have seen that the version of knowledge argument based on a model of contemporary psychophysics requires that knowing what it is like is propositional knowledge of facts. The upholders of the ability reply have challenged this assumption by arguing that knowing what it is like is a form of knowing how. However, on her release, Mary acquires certain beliefs about the fact that objects looks the same colour. A lot remains to be said about the nature and content of these beliefs.
Knowing what it is like and knowing how
Notes . Jackson (1982), Jackson (1986). He has recently recanted, see Jackson (1998a) and Jackson (1998b). . This is the classical version of the type identity theory developed by Smart (1959), Feigl (1967) and Place (1956). . This position is advocated by functionalists, see for instance Fodor (1974), and by Donald Davidson, see Davidson (1970). . Fodor (1974). . A wide debate concerns the compatibility of causal analyses of mental states with type identity theory. For compatibilism, see Jackson, Pargetter, and Prior (1982). Anticompatibilism is defended in Tye (1983). . For a defence of these tenets, see Papineau (2002: 13–18). In the appendix of this book, Papineau offers a historical analysis of the role of these theses in contemporary physicalism. . Jackson (1982). . Although the knowledge argument considers perceptual experiences of coloured objects, we cannot exclude that imagining, remembering, or stimulating the visual cortex might provide knowledge of what it is like. What is relevant to Jackson’s argument is that having these mental states cannot be considered a requirement to possess complete scientific knowledge of colours and colour vision. . I avoid using the expression what an experience is like, that might suggest that the content of this knowledge is about the experience. This assumption might not play any central role in the knowledge argument. . Dennett (1991: 399–403), Churchland (1986: 331–334). . This line is suggested in Churchland (1989). See also Lewis (1990: 580–581). . This model of contemporary colour vision science emerges from the work of colour scientists, see, for example, Hurvich (1981). A detailed analysis of this model has been provided in Clark (1993) and Clark (2000). . See Hurvich (1981: 113–149). . The most detailed colour spaces concern experience of colours under specific visual conditions, just a part of the totality of colour experiences. In addition, the number of discriminatory judgements required to determine a complete colour space just by using statistical procedure, (see references in the next footnote), is at the moment beyond our technological possibilities. . Colour spaces, ideally, can be built from tables representing judgements of similarity and by applying certain logical and mathematical procedures that determine the number and order of the qualities along which subjects discriminate, see Clark (1993: 76–116, 210– 221). Given that none of the concepts involved in these procedures require having colour experiences, we can assume that Mary, when in the laboratory, has a complete grasp of the notion of qualitative space. . This construction is adapted by Clark (2000: 256–257).
Luca Malatesti . See Clark (2000: 256–257). . Nemirow (1980), Nemirow (1990), Lewis (1983) and Lewis (1990). . It is important to notice that some philosophers endorse this view on knowing what it is like although they reject physicalism, see Mellor (1993). . Other criticism have been advanced in Lycan (1995: 244–249), Loar (1990: 607–608) and Tye (2000: 11–15). . Sellars (1963: 93–94, 192–193) provides a seminal discussion of this distinction. . However in certain passages he says that Mary learns “something about the world”, Jackson (1986: 293). . Churchland (1986: 333). . Paul Churchland has promoted this view in Churchland (1985). . This claim has been criticised by physicalists, see Newton (1986), and by antiphysicalists, see Robinson (1993).
References Churchland, Patricia S. (1986). Neurophilosophy: Toward a Unified Science of the Mind-Brain. Cambridge, MA: MIT Press. Churchland, Paul (1985). Reduction qualia and the direct introspection of brain states. Journal of Philosophy, 82, 8–28. Churchland, Paul (1989). Knowing qualia: A reply to Jackson. In Paul Churchland, A Neurocomputational Perspective (pp. 67–76). Cambridge, MA: MIT Press. Reprinted In Paul Churchland & Patricia Churchland, On the Contrary: Critical Essays 1987–1997 (pp. 143–157). Cambridge, MA and London: MIT Press (1997). Clark, A. (1993). Sensory Qualities. Oxford: Clarendon Press. Clark, A. (2000). A Theory of Sentience. Oxford: Oxford University Press. Davidson, D. (1970). Mental events. In L. Foster & J. W. Swanson (Eds.), Experience and Theory. (pp. 79–91). Boston: Massachusetts University Press. Reprinted In D. Davidson (Ed.), Essays on Action and Events (2nd ed. 2001), (pp. 207–225). Oxford: Oxford University Press. Dennett, D. (1991). Consciousness Explained. London: Little & Brown. (Reprinted 1993. London: Penguin.) Feigl, H. (1967). The ‘mental’ and the ‘physical’. In H. Feigl, M. Scriven, & G. Maxwell (Eds.), Minnesota Studies in the Philosophy of Science. Vol. II: Concepts, Theories and the MindBody Problem (pp. 370-497). Minneapolis: University of Minnesota Press. Fodor, Jerry. (1974). Special sciences (or the disunity of science as a working hypothesis). Synthese, 28, 97–115. Reprinted In N. Block (Ed.). (1980). Readings in Philosophy of Psychology, Vol. 1 (pp. 120–133). Cambridge, MA: Harvard University Press. Hurvich, L. (1981). Color Vision. Sunderland, MA: Sinauer Associates Inc. Jackson, F. (1982). Epiphenomenal qualia. Philosophical Quarterly, 32, 127–36. Reprinted In W. Lycan (Ed.). (1990). Mind and Cognition (pp. 469–477). Oxford: Blackwell.
Knowing what it is like and knowing how
Jackson, F. (1982). Epiphenomenal qualia. Philosophical Quarterly, 32, 127–136. Reprinted In W. Lycan (Ed.). (1990). Mind and Cognition (pp. 469–477). Oxford: Blackwell. Jackson, F. (1986). What Mary didn’t know. Journal of Philosophy, 83, 291–295. Reprinted In N. Block, O. Flanagan, & G. Güzeldere (Eds.). (1997). The Nature of Consciousness (pp. 567–570). Cambridge, MA: MIT Press. Jackson, F. (1998a). From Metaphysics to Ethics: A Defense of Conceptual Analysis. Oxford: Claredon. Jackson, F. (1998b). Postscript on qualia. In F. Jackson (Ed.), Mind Method and Conditionals. Selected Essays (pp. 76–79). London: Routledge. Jackson, F., Pargetter, R., & Prior, E. W. (1982). Functionalism and type-type identity theories. Philosophical Studies, 42, 209–225. Lewis, D. (1983). Postscript to ‘Mad pain and Martian pain’. In D. Lewis (Ed.), Philosophical Papers, Vol. 1 (pp. 130–132). New York: Oxford University Press. Lewis, D. (1990). What experience teaches. In W. Lycan, (Ed.), Mind and Cognition (pp. 499– 519). Oxford: Blackwell. Reprinted In N. Block, O. Flanagan, & G. Güzeldere (Eds.). (1997). The Nature of Consciousness (pp. 580–595). Cambridge, MA: MIT Press. Loar, B. (1990). Phenomenal states. In J. Tomberlin (Ed.), Philosophical Perspectives, Vol. 4 (pp. 81–108). Atascadero: Ridgeview. Reprinted In N. Block, O. Flanagan, & G. Güzeldere (Eds.). (1997). The Nature of Consciousness (pp. 597–616). Cambridge, MA: MIT Press. Lycan, W. (1995). A limited defence of phenomenal information. In T. Metzinger (Ed.), Conscious Experience (pp. 243–258). Thorverton: Imprint Academic/Scöningh. Mellor, D. H. (1993). Nothing like experience. Proceedings of the Aristotelian Society, 93, 1– 16. Nemirow, L. (1980). Review of T. Nagel, Mortal Questions. Philosophical Review, 89, 475–476. Nemirow, L. (1990). Physicalism and the cognitive role of acquaintance. In W. Lycan (Ed.), Mind and Cognition (pp. 469–477). Oxford: Blackwell. Newton, N. (1986). Churchland on direct introspection of brain states. Analysis, 46, 97–102. Papineau, D. (2002). Thinking about Consciousness. Oxford: Clarendon Press. Place, U. T. (1956). Is consciousness a brain process? British Journal of Psychology, 47, 243– 255. Robinson, H. (1993). The anti-materialist strategy and the knowledge argument. In H. Robinson (Ed.), Objections to Physicalism (pp. 159–183). Oxford: Oxford University Press. Sellars, W. (1963). Science, Perception and Reality. London: Routledge & Kegan Paul. Smart, J. J. C. (1959). Sensations and brain processes. Reprinted In C. V. Borst (Ed.). (1970). (revised version) The Mind/Brain Identity Theory (pp. 52–66). London: Macmillan. Tye, M. (1983). Functional and type-physicalism. Philosophical Studies, 44, 161–174. Tye, M. (2000). Consciousness, Color and Content. Cambridge, MA and London: MIT Press.
Chapter 8
Human cognition An evolutionary perspective Ian Tattersall American Museum of Natural History
To a scientist, causality is a tricky thing. For almost invariably, the cause of the phenomenon being investigated turns out to be an effect of a prior process, back in a seemingly infinite recession. What’s more, in evolutionary science in particular, causation often seems ruled by contingency. If any cause must in some sense precede its effects, then those effects are at best not only retrospectively secondary, but they are prospectively contingent to whatever use Nature may make of them. Efforts to seek “chains” of causation thus run the risk of creating an artefact rather than elucidating a consistent process. The notion of “mind” carries equal hazards, especially since it is impossible for any organism possessing the cognitive qualities of normal Homo sapiens to experience (or even effectively to imagine) the mental states of other organisms. We can alter human mental states chemically, and we can observe the behaviors and responses of members of other species, but we cannot combine experiment with observation to create satisfactory subjective characterizations of alternative kinds of mind to our own. Yet it seems to be inbuilt into the human psyche to want to know just what it is that makes our ways of interacting with the rest of the world so unique, and by extension how that capacity was required. In this article I shall thus look at our fossil and archaeological records, the archives of our physical and behavioral evolution, with an eye to determining the extent to which they can be helpful in this quest.
Ian Tattersall
.
Human evolution and cognition
What is it, exactly, that makes the cognitive processes of modern Homo sapiens unique in the living world? And how was that something acquired? These twin questions are among the most impenetrable of all those facing science. Yet at the same time they are possibly the most alluring, for our narcissistic species is unfailingly fascinated by the contemplation of itself and of the ways in which we human beings are distinguished from the rest of the living world. Certainly, we are part of that world, from which we have emerged precisely as every other species has done. Yet there is undeniably a gulf between us and every other living organism – including our closest living relatives, the great apes. And it is a gulf that lies above all in the ways in which we process information about the world, rather than in any of our undeniably striking physical characteristics. For although every living species is in some way physically and behaviorally distinctive, even as all are inevitably part of the biotic world, no other organism tries, as we do so hard, to distance itself from that world. And while this difference between us and the rest of nature, certainly as we learn to accept it, is at least in part a product of our perceptions, it is nonetheless a real one. There really is something uniquely – and disturbingly – distinctive in the way in which we modern Homo sapiens perceive and interact with the world around us. We recreate that world in our heads in order to explain it to ourselves, rather than simply reacting directly – in more or less complex ways – to the stimuli we receive from it. It was not always thus. The hominid family (the group containing Homo sapiens and all those now-extinct species that are more closely related to it than to the great apes and their fossil relatives) has roots that extend quite deep in time. The known human fossil record now stretches back to close to seven million years (7 myr) ago (Brunet et al. 2002), and contains around 20 distinct species (Figure 1; see also Tattersall & Schwartz 2000). Yet, as far as we are able to tell from an admittedly imperfect record, no hominid besides Homo sapiens has ever symbolically reconstructed the world in its mind in the way we do. Indeed, even the earliest fossil populations that anatomically resembled modern Homo sapiens apparently interacted with the world much in the way that their extinct predecessors had done, rather than in our own distinctive manner. So how did our unusual way of doing business emerge?
Human cognition Mya 0 H.sapiens H.neanderthalensis H.heidelbergensis
. Brain size and the evolutionary process Clearly, the answer to this question must lie somehow in the evolution of our brain, the underlying determinant of our behaviors. Conventional wisdom tells us that hominid brain size increased gradually over time (see, for example, Wolpoff 2000), reflecting an incremental improvement in hominid cognitive processes over hundreds of thousands of generations. This perception directly reflects the lingering dominance in paleoanthropological thought of the Evolutionary Synthesis (see Mayr 1986), the grand paradigm of evolutionary theory that has ruled in human evolutionary studies since about 1950. In brief, the Synthesis reduced virtually all evolutionary phenomena to the action of natural selection, slowly and consistently operating on the gene pools of lineages of organisms over vast spans of time. In essence, the focus under this construct is on the accumulation of tiny changes within a continuous reproductive chain extending over the eons from the first hominid to Homo sapiens – and what view, after all, could be more congenial to a field traditionally devoted to the pursuit of the Missing Link? Sadly or otherwise, it is now clear that this formulation, elegantly simple though it might have been, was inadequate to account for the evident complexities of the evolutionary process. In the early 1970s (Eldredge & Gould 1972; see also Eldredge 1985), researchers began to question whether the Synthesis could satisfactorily explain the patterns of change actually observed in the fossil record, pointing to the fact that stasis and discontinuity furnished at least as strong a signal in that record as did gradual change. The result is that in the ensuing three decades the realization has spread widely that natural selection operates at more levels than simply that of the individual organism, and that much more is involved in the overall evolutionary process than adaptation alone (see Gould 2000). In human paleontology (admittedly with the hominid fossil record then only a fraction of the size of that available now) Ernst Mayr (1950), one of the principal architects of the Synthesis, had contrived by the middle of the twentieth century to reduce the number of species of hominids known to three, all of them arranged in a linear progression culminating in Homo sapiens. Now, however, with a hugely enlarged fossil data base, it is possible to see quite clearly that – although we instinctively take it for granted, since this is the situation we know – it is in fact extremely unusual for Homo sapiens to be the lone hominid on the planet (Tattersall 2000). Indeed, incomplete as the record undoubtedly is, it furnishes abundant evidence throughout of hominid diversity. In northern Kenya, for example, in the period following about 2 myr ago, at least four
Human cognition
different hominid species (see Figure 1) shared not only the same continent, but the very same landscape. One of the factors that facilitated the acceptance of Mayr’s linear scenario was the undeniable reality that, the farther one goes back in time, the smaller hominid brains tend to become. In fact, this evident pattern is the strongest apparent evidence that anyone can actually quote for a pattern of linearity in human evolution. For brain sizes (when this information is preserved) are by their very nature quantifiable; and it turns out to be quite easy to join these fairly steadily enlarging (though spottily distributed) numbers up into a sequence through time, implying that change is inevitable and that the only matter of interest is the possibility of altering rates of change at different times (“character state velocities”: Eckhardt 2000). But is this “evidence” in reality so strong? If Mayr’s reductionist formulation were sustainable, then it might be. However, as I have noted, the actual pattern that is currently emerging from the enlarging hominid fossil record is very different. Instead of a steadily modifying chain extending across time, the signal is one of a diversity of hominid species throughout. The imperative thus becomes one of recognizing the species that exist within the morphological spectrum that our fossil precursors represent, for we can no longer see species simply as arbitrary segments of evolving lineages. The process of responding to this imperative, it has to be admitted, is at an embryonic stage (though see Tattersall & Schwartz 2000). But it is already quite obvious that the true number of known hominid species is already a large one, and that those species were morphologically very diverse. Yet more reason, indeed, to believe that the story of human evolution has been one of consistent evolutionary experimentation (with multiple species originations and extinctions), rather than one of withinlineage fine-tuning over the eons. And if we cannot read hominid fossils simply as links in a chain, it follows that there is a pattern out there to be found – a pattern that we cannot simply discover, but that requires analysis (Eldredge & Tattersall 1975). The central units of such analysis are the species themselves. Numerous hominid species have appeared, have competed in the ecological arena, and have gone extinct (with or without leaving descendant species). If we are to discern pattern in the human fossil record, then, it is essential that we be able to recognize those species with reasonable accuracy. This is not an easy task (Tattersall 1986, 1992). But it is an essential prerequisite to further studies, including any attempt at determining the pattern of hominid brain size increase over time. And at present several admissions must preface any effort to do so (Tattersall 1998). First, of course, it has to be admitted that we do not know
Ian Tattersall
the true number of hominid species out there in the fossil record (though we can probably make a good stab at determining a minimal number without severely distorting the phylogenetic pattern we perceive: see Tattersall 1986). Second, within-species brain size is notoriously variable (the brain sizes of behaviorally normal modern humans, for instance, run from under 1,000 to over 2000 ml); and even with a relatively good hominid fossil record we have no idea of the ranges of brain size variation characterizing even those extinct hominid species that we can agree on. Third, if our desire is to calibrate rates of change in brain size over time, we need to have reliable dating; and even where reasonably accurate dates exist for individual fossils, we have no idea of the overall time ranges (which probably varied widely) of the species they represent, within which they could potentially have given rise to descendent species. And, finally, we are very far from reaching anything approaching agreement on the phylogenetic relationships among those species in whose identities we can be reasonably confident. So what does the “average” increase in human brain size over the past several million years mean? Yes, go back to 3 myr ago and beyond, and hominid brains were in the ape size range – about a third the size of ours compared to body volume. At 1 myr ago hominid brains were, in very approximate terms, two-thirds the size of ours. And by about 200 thousand years (200 kyr ago), before the appearance of Homo sapiens, some hominid species, at least, had brains as big as our own. Unquestionably, larger-brained hominids (with myriad other derived characters as well, of course) eventually won out in the evolutionary stakes (though some big-brained species lost out, as well). Thus, overall, we can certainly detect a time-related trend. But what’s the pattern? The traditional tendency has been to join up brain sizes over time in a straight line, with the implicit assumption that slow, steady change linked them all. For the reasons I’ve just given, though, that’s hardly a practical option. And if it is correct, as it increasingly seems to be, that human evolution has been among other things a story of species competing with their close relatives as well as with other elements in the environment, it is at least as likely that a relatively small number of discrete enlargement events in different species was involved as that hominid brains (in diverse lineages) inexorably expanded generation by generation, come hell or high water. Big brains are metabolically expensive (Martin 1983), and there must certainly have been a strong countervailing advantage for them to have emerged as the norm. The conclusion seems unavoidable that this advantage must have lain ultimately in increased “intelligence” (whatever that is); but it is, minimally, as probable that more-intelligent hominid species outcompeted less-intelligent ones, as that larger-brained indi-
Human cognition
viduals simply reproduced more effectively in successive generations. This is most especially true of the dramatically fluctuating environmental and geographical circumstances of the Pleistocene, the “Ice Ages” epoch during which most hominid brain size increase took place. Well, if pattern in brain size increase over time is far from clear-cut, what about other brain attributes preserved in the fossil record? Holloway (2000) has summarized such evidence as can be discerned about brain reorganization from the morphology of fossil endocasts, which approximately reproduce the external morphology of the brain. And it turns out that such evidence is sketchy at best. Archaic bipeds (australopiths) in the 3.5–2.5 myr range seem to show some reduction (relative to the primitive hominoid condition) of the primary visual striate cortex, in conjunction with an increase in the posterior parietal association cortex. With the appearance of “early Homo” (an exceptionally motley group of fossils dated to around 2.5 to 1.9 myr ago), Holloway perceives some reorganization of the frontal lobes, plus an increase in cerebral asymmetries. This was apparently further accentuated somewhat later in time, with Homo erectus. By the time Homo neanderthalensis (his “archaic Homo sapiens”) comes along at about 200 kyr ago, Holloway finds “refinements in cortical organization to a modern Homo pattern” (p. 149). But that’s about it; there’s not a lot more to be said or even inferred on the basis of the existing data set.
. The behavioral record Fossil brains and braincases themselves thus do not get us very far in the quest for the origins of our extraordinary modern human cognition. This is hardly surprising in so far as, while we now know quite a lot about which brain regions are involved in which mental activities, we are utterly ignorant of how a mass of electrochemical signals in the brain is converted into what we experience as our consciousness. But it does mean that, if we are to pursue this line of inquiry further, we are forced to seek proxies for cognitive function. Such proxies are only to be found in the behavioral record left behind by our precursors. And with the exception of some stable-isotope studies (e.g. Sponheimer & Thorpe 2001) – which suggest that at least some populations of australopiths in the 3 myr range ate substantially more meat than is typically consumed by apes today – the behavioral record is more or less synonymous with the material archaeological record. This record begins with the invention of stone tools, about 2.5 myr ago. The earliest stone tools, known from several sites in eastern Africa, consist of
Ian Tattersall
simple sharp stone flakes knocked off one small riverbed cobble using another. Not very impressive, perhaps; but making even the crudest type of stone tool is a feat that, cognitively speaking, goes well beyond what any living ape has been able to achieve, even with intensive coaching (Schick & Toth 1993). And this invention must have had a profound effect on the lives of the small-brained creatures who made it. Cut-marks on the bones of animals found along with such tools show that the sharp flakes were used for butchering animal carcasses, and there is ample evidence that larger cobbles were used to break long bones to get at the marrow within as well as as hammers for flake production. Formerly these resources would have been effectively unavailable to small-bodied bipeds who were still dependent on the shelter of the trees. Beyond this, the refitting-together of complete cobbles from multiple flakes found at the same butchery site has shown that the makers of the earliest stone tools had considerable foresight, for they carried around suitable cobbles (a scarce resource) over considerable distances before making them into tools as needed. Again, we glimpse here a substantial advance over the cognitive capacities of living apes, and over those inferred for the ape/human common ancestor. How we would describe those advances in terms of the way these early hominids experienced the world around them is unclear; for one of the limitations of our own remarkable cognition is that it is impossible for us to experience, even imaginatively, any other cognitive state. One thing, though, is clear: that it is a fundamental error to assume that our hominid precursors were simply junior-league – and, by implication, inferior – versions of ourselves. This is something that it is very important to bear in mind in any account of human cognitive evolution; there are clearly many ways of being hominid, and ours is only one of them. Further, once we have stone tools we have evidence of the activities of hominids for whose cognitive processes we have no living model; and this is, of course, true for all subsequent hominid species prior to the emergence of behaviorally modern humans. It is highly unlikely that members of even the earliest, pre-toolmaking, hominid species behaved in a way approximating that of any living ape; and the problem of employing observable behavioral models becomes more intense as time passes, for, as I will suggest later, the pattern of events was not one of a simple increasing approximation to ourselves. In practical terms, of course, there is another difficulty. Hominids are not simply stone-tool-making machines, and stone toolmaking styles are at best an indirect reflection of the richly varied cognitive capacities and expressions of extinct kinds of hominid. Even technological expressions of other kinds may have differed substantially among ho-
Human cognition
minids who possessed similar stone tool kits, let alone other manifestations of their conscious states. Yet stone tools and site size and structure – or its lack – are in most cases virtually all we have from which to reconstruct the various cognitive conditions of our precursors. This having been said, on the preserved technological level the pattern is clear. It is one of long periods of relative stability, even of non-change, punctuated by the relatively sudden addition of new technologies. For while throughout the Old Stone Age (the Paleolithic) older-generation technologies tended to persist for long periods alongside the new, there is little evidence for a pattern of gradual change or development from one technology to another. This is hardly surprising, for it is the pattern we still see in technological development today. Major technologies tend to be based on new principles which are then elaborated upon in various ways; but the old does not immediately disappear with the introduction of the new, and it is rare that important new technologies are linear developments from old ones. This pattern was established early on, for simple flake tools continued to be made into comparatively recent times (after all, a sharp cutting implement always comes in useful), even as more complex stone tool types appeared. A million years after the first stone tools were made, their successor utensils were essentially the same, and it was not until a bit over 1.5 myr ago that a new type of stone tool appeared. Interestingly, while the first crude stone tool kit was made by physically archaic forms, later kits of equivalent simplicity continued to be made by the hominids with body structures much closer to our own who appeared on the scene sometime after about 2 myr ago. The new stone tool-type, introduced after the archaic hominids had essentially disappeared, was the “Acheulean” hand-axe, a much larger and more complex utensil carefully fashioned on both sides to a deliberate and symmetrical shape. For the first time, stone tool makers were arguably fashioning tools according to a particular “mental template” that clearly existed in their minds before production started, rather than simply aiming for an attribute (a simple cutting edge) regardless of the exact shape of the finished product. Whatever the case, the switch in technological styles presumably implies some kind of cognitive advance; but what exactly this might have been is far from clear, and in pondering this question we have to bear in mind how behavioral advances have to originate. Any technological innovation has to arise within a functioning population, for there is no place else it can do so. And any individual who invents a new technology cannot differ significantly in physical organization from his or her parents or offspring. A corollary of this is that we cannot usefully invoke the
Ian Tattersall
arrival of a new kind of hominid to “explain” the introduction of a new way of doing things – however convenient it would be if we could – despite the fact that in the remote past various technologies were almost certainly passed “sideways” from one kind of hominid to another. Yes, the invention of the hand-axe was a cognitive advance in the sense that it represented a new way of envisioning possibilities in the mind; but what this meant in terms of the physical apparatus underlying this cognitive process, and the way of viewing the world resulting from it, is far from clear. The overall pattern of highly sporadic innovation persists, however, at least as far as stone tool making is concerned. For it is only at about 300 kyr ago that we see the introduction of a radically new stoneworking technology. This was the “prepared-core” technique, whereby a stone ”core” was carefully fashioned until a single blow could detach an almost-finished stone tool with a continuous cutting surface around its periphery. Meanwhile, however, other important developments had taken place. Notable among these was the domestication of fire in hearths, for which strong evidence is first found at around 400 kyr ago (earlier potential signs of fire use are few and disputed: see Klein 1998; Tattersall & Schwartz 2000). The use of fire and cooking must certainly have made a fundamental difference to the lives of the hominids who controlled this technology, but we must be wary of ascribing to early hominid fire use all of the symbolic overtones that characterize the exploitation of fire by Homo sapiens today. A remarkable glimpse at life at around this time is also afforded by the recent discovery of a series of large and carefully-fashioned throwing spears at the site of Schoeningen, in Germany (Thieme 1997). Most archaeologists had considered that at this phase of human evolution sophisticated ambushhunting techniques had yet to be introduced. Wood preserves poorly if at all over more than a few hundred years, and in the absence of material evidence it was widely surmised that, if possessed at all by hominids in the half-millionyear-range, spears would have been of the thrusting type, involving dangerous up-close encounters with prey animals. Yet the 400 kyr-old spears miraculously preserved in a bog at Schoeningen are up to two meters and more long, and are clearly shaped like modern javelins, with their weight concentrated at the front. Yet another hint of substantial cognitive advance, but what, exactly? And, whatever it was, how long had it been in existence by Schoeningen times? Returning to the much better stone tool record, the best-documented practitioners of prepared-core stone tool making are without doubt the Neanderthals, a distinctive group of hominids with brains as large as our own, who inhabited Europe and western Asia from something over 200 kyr to a little un-
Human cognition
der 30 kyr ago. Despite its large brain size, however, Homo neanderthalensis was behaviorally as well as anatomically distinct from modern humans. There is no space here for a full discussion of the behavioral contrasts between the Neanderthals and ourselves, but the matter can be summarized succinctly by noting that, while in broad terms the Neanderthals did pretty much what their predecessors had done, if perhaps a little better, modern humans, in the guise of the invading Cro-Magnons who displaced them, were totally unprecedented behaviorally. Yes, the Neanderthals buried their dead (see Gargett 1989, for a dissenting view), though only occasionally, and then very simply; and a degree of social caring and support within the Neanderthal social unit is implied by the long survival at Shanidar, in Iraq, of an individual severely handicapped by a withered arm. But despite these echoes of what we would instinctively recognize as humanity, the Neanderthals showed effectively no evidence at all (at least until post-contact times) of symbolic activities of the kind that so richly characterized the lives of the Cro-Magnons. Yes, the occasional example of symbolic production (scratches on plaques, and so forth) has been reported from Middle Paleolithic (Neanderthal-equivalent) times. But the symbolic nature of virtually all such manifestations has been disputed at one time or another, and at best they are exceptions that prove the rule. It is quite possible that we might glimpse symbolism in the products of the occasional individual in early times; but what is important is not what individuals might privately or sporadically do, but what becomes common cultural currency within societies. Of course, many human societies have been recorded historically that had language and complex symbolic traditions, but that left behind them little in the way of a material record of the kind we might hope to detect in the archaeological record. But this simply emphasizes the significance of the symbolically dense record of the Cro-Magnons: although absence of evidence is certainly not evidence of absence, the aggressive presence of evidence for symbolic activity among the Cro-Magnons is little short of mind-boggling. Following their entry into Europe (from an unknown, but most likely African, place of origin) at about 40 kyr ago, the Cro-Magnons displayed virtually the entire panoply of symbolic behaviors that characterizes humans worldwide today. Well before 30 kyr ago, they were painting spectacular art on the walls of caves, producing exquisite carvings and etchings, making complex notations on bone, ivory and antler plaques, performing music on bone flutes with complex sound capabilities, decorating their bodies with elaborate adornments, burying the dead with sumptuous grave furnishings, and in general conducting a lifestyle drenched with symbolic overtones. In the technologi-
Ian Tattersall
cal realm, complexity of technique and the range of materials used in tool production increased dramatically; and local differentiation of technological traditions became the norm, in contrast to the simpler toolkits of the Neanderthals, which had remained fairly static over the entire expanse of time and space these hominids inhabited. By only a little under 30 kyr ago, the appearance of delicate-eyed bone needles announced the advent of couture, and ceramic figurines were being baked in simple but remarkably effective kilns. Clearly, however different the Cro-Magnons may have been from any modern humans in cultural details, they were unquestionably us. At the end of millions of years of human evolution during which cultural and technological – and by extension cognitive – innovation had been highly sporadic, with extensive periods of time characterized by business as usual, an entirely new cognitive phenomenon had at last emerged.
. The origin of modern human consciousness The fact that in cognitive terms the Cro-Magnons were already full-fledged modern humans when they entered Europe highlights the question of the geographical origins of the familiar modern human consciousness. The culture of the Cro-Magnons was distinguished above all by its richness of symbolic expression; but it was not the first to yield such expressions. Over the past half-million years or so the odd scratching on a piece of rock or bone has been interpreted as symbolic, but never undisputedly; and it is only in the African record that any artifact convincingly interpretable as symbolic has been reported from the period before about 50 kyr ago. McBrearty and Brooks (2000) have summarized the evidence for the early stirrings of “modernity” in Africa. This evidence extends beyond such obviously symbolic artifacts as engraved ostrich eggshells and gastropod shells pierced for bodily ornamentation, to such activities as flint mining and long-distance trade in materials. But the most striking finding so far in this domain is that of some recentlyreported ochre plaques from South Africa’s Blombos Cave. Bearing distinctive geometric incisions, these are dated to over 70 kyr ago. More indirect evidence from the southern tip of Africa includes an apparent symbolic division of living space noted at the shelter of Klasies River Mouth, up to 120 kyr ago (Deacon & Deacon 2000). Both of these innovations were associated archaeologically with Middle Stone Age tool assemblages roughly equivalent to what the Neanderthals were making at the same time in Europe, rather than with Late Stone Age traditions comparable to those of the Cro-Magnons.
Human cognition
Perhaps we should not be surprised at this. For if modern human cognition results from a generalized biological potential that can be expressed behaviorally in many different ways, there is no reason to believe that all of its possible consequences should have been discovered at once by our predecessors. Indeed, the entire recorded history of humankind has been one of the discovery of new things that can be done with our underlying potential; and we are even today discovering new ways of exploiting our remarkable cognitive capacities (see discussion by Tattersall 2002). These early southern African symbolic expressions are associated with hominids that were anatomically very similar (if not identical: see Schwartz & Tattersall 2003) to ourselves. In the Levant (which may in some respects be regarded as an ecogeographic extension of Africa; see Klein 1999), the first anatomically modern Homo sapiens from Israel’s Jebel Qafzeh cave are dated to something under 100 kyr ago (Valladas et al. 1988), and are associated with a Neanderthal-like Middle Paleolithic archaeological assemblage lacking any artifacts that are plausibly interpretable as symbolic. A plaque with geometric incisions has been reported from the Middle Paleolithic site of Quneitra on the Golan Heights at about 50 kyr ago. However, where this apparently symbolic piece fits into the story is difficult to determine, since it is clear that Homo neanderthalensis and Homo sapiens coexisted in the Levant for a long time, from over 100 kyr ago to a mere 40 kyr ago, or perhaps less. During this long period of coexistence Homo neanderthalensis and Homo sapiens appear to have made more or less identical stone tool kits (though different ranging patterns have been inferred: Lieberman & Shea 1994). This is very different from the sequence of events reflected in the European record, where it is clear that the Neanderthals disappeared totally within about 10 kyr of the Cro-Magnons’ first arrival. Perhaps it is significant in this connection that it was only after the invention in the Levant at about 47 kyr ago of an “Upper Paleolithic” toolkit comparable (and perhaps ancestral) to the Cro-Magnons’, that the Neanderthals finally disappeared from the region. This may suggest that a short-term cognitive “event” occurred around that time, with technological or cognitive consequences that finally transformed Levantine Homo sapiens into invincible competition for its Neanderthal neighbors. As we will see below, this is not necessarily incompatible with the view of an earlier stirring of symbolic activity in Africa itself. Any innovation, whether it be physical or cultural, necessarily has to arise within a limited population existing in a particular corner of the world. On the basis of the admittedly imperfect record we currently have, it seems most likely
Ian Tattersall
at present that this place was somewhere in the vast continent of Africa. But was the innovation in question here – essentially, it seems, the acquisition of symbolic ability – a biological innovation or simply a cultural one? Of course, at some level it has to be both. It is inconceivable, for example, that anyone could have trained one of the ancient “bipedal apes” that ultimately gave rise to us to be a proficient cave painter; there is clearly a biological basis for this and related abilities. And since the Neanderthals – and the earliest Levantine Homo sapiens – apparently lacked overtly symbolic behaviors, this basis must be something that transcends mere enlargement in brain size. It is a qualitative, not merely a quantitative, difference. Just what it is that transmutes electrochemical signals in the brain into what we experience as consciousness is, of course, unknown; and it is in any case beyond my particular expertise to speculate on the anatomical or physiochemical basis of human consciousness, although we may be certain that there is a structural modification of some kind at its core. In any case, we are limited by the fact that we cannot use modern instrumentation to peer into working ancient brains to determine differences in function between our brains and those of our closest non-symbolic relatives. And the living great apes are far too remote in common ancestry to us to be of more than limited use as investigative models. Thus at this point in our knowledge we have thin grounds for identifying the key factor(s) possessed by the human brain that account(s) for our unique reasoning abilities. However, whatever this key factor was, since any evolutionary novelty must already exist (for random genetic reasons) before natural selection can begin to work on it, it cannot, as we so often tend to assume, have been propelled into existence by natural selection. Natural selection is simply not a propulsive force, for in essence it works by elimination, not by creation. And the necessary neural constituent for symbolic thought must, obviously, have been in existence before it could be exploited for cognitive purposes. As a working hypothesis, then, it seems reasonable to view the acquisition of human consciousness as an emergent event, one that was unrelated to adaptation, but was rather due to a chance coincidence of elements. What those elements were, exactly, must for the time being remain a mystery, although the brain’s preexisting organization and its large size, both the results of a long and doubtless complex evolutionary history, must have been essential ingredients. Whatever the answer to this fundamental question, however, our immediate ancestor evidently possessed a brain that was already exapted for symbolic thought with the incorporation of a single small genetic change – plausibly the same change that led to the establishment of bony modern anatomy through-
Human cognition
out the skeleton. But even this is far from the whole story for, as we know, the first members of Homo sapiens – hominids indistinguishable osteologically from living people – behaved more or less in the way that Neanderthals did for tens of thousands of years. The acquisition of the biological substrate is thus not the whole story; for, to give rise to the cognitive potential that our species has been exploiting for the last 50–70 kyr, it must have been followed at some distance in time by an essentially cultural – or at least behavioral – innovation that “released” the underlying potential. This, it must be said, is hardly surprising. Indeed, a strong case can be made that all innovations must arise as exaptations; birds, for example, possessed feathers for millions of years before coopting them for purposes of flight. What, then, might the cultural releasing factor have been in the case of human symbolic cognition? The favorite candidate of those who have thought about this problem is the invention of language, for language is a human universal that is virtually synonymous with symbolic thought and I, at least, have a difficult time imagining such thought in its absence. This is not to say that all thought is linguistic, although it’s undeniable that all linguistic manipulations are symbolic. Human thoughts are processed by a brain that bears the marks of numerous evolutionary stages, from brain stem structures up through the neocortex; and it is important to recall that not all components of human ratiocination necessarily consist of symbolic combinations and recombinations. Our symbolic thinking processes are also intimately tied up with emotion and intuitive reasoning. The richness and depth of human thought undoubtedly results from this rather untidy combination of influences; but the addition of symbolic manipulation to the pre-existing components of the mix is what makes it possible to articulate questions to ourselves, and to find answers to them. I do not think it unfair to claim, for example, that the most parsimonious reading of the undeniably complex Neanderthal archaeological record is that these close relatives represented the ultimate in what could be achieved by intuitive reasoning alone (Tattersall 1998). And that the ability of these extinct hominids to survive in the presence of non-behaviorally modern humans, while later rapidly succumbing to the Cro-Magnons, was occasioned by the fact that the addition of symbolic aspects to the behavioral repertoire of Homo sapiens simply made the latter an unbeatable competitor. Which, for better or for worse, has steadily extended its domination of our planet ever since.
Ian Tattersall
. Conclusion This brief survey of the human behavioral record over the last few million years has, I hope, sufficed to show that human phylogeny has not been a simple history of perfecting adaptation over the eons. Instead, it has been a history of evolutionary experimentation, with new species regularly thrown by nature on to the evolutionary stage, contending with other hominids as well as with more remotely related competitors, and more often than not becoming extinct. A history of this kind is in line with what we know of the histories of most successful mammal groups, as well as with our modern understanding of how the evolutionary process functions. And it suggests that we cannot look to simplistic notions of “adaptation” to explain the acquisition of our remarkable cognitive function. This book is focused on various aspects of causality as it relates to the human mind, the alternative name we give to the symbolic cognitive state that, as far as we know, separates our own species from all other living forms on Earth. And the appraisal given here of the order of events that appears to have accompanied the origin of the human mind emphasizes that strictly reductionist explanations of this unique phenomenon are inadequate to account for it. Yes, any effect must by definition have a cause; but that cause need not be a directional one, and any particular cause need not necessarily have a single or inevitable effect. Random chance and pure hazard enter into the evolutionary process as much as into all other historical processes. Rather than pointing to a model of gradual fine-tuning as smaller and more primitive brains yielded generation by generation to larger and structurally modified ones within the slowly-modifying human lineage, the evolutionary sequence explored here indicates a long history of natural triage among diverse hominid taxa, all exploring different ways to express the hominid potential. Our own species simply happens to be the sole survivor of this long and eventful process; and we are presumably alone in the world today because we possess remarkable cognitive qualities that came about recently, in an exaptive and emergent fashion (albeit as an addition to a cognitive apparatus that had already been shaped by a very long accretionary history), rather than through long, gradual burnishing by natural selection. The human mind, in other words, did not evolve “for” anything. “Modular” notions of brain phylogeny are useful in the sense that they reflect the discrete and sporadic nature of successful evolutionary innovation. But it appears that the modern human capacity is more generalized, less mechanical, even less focused, than modular concepts suggest. It is for this reason that “evolutionary
Human cognition
psychology,” with its emphasis on the importance of individual genes in determining a vast range of human behaviors, is actually undermined by its reductionist appeal. Homo sapiens has been exploring the multifarious dimensions of this remarkable and emergent underlying capacity for seventy thousand years or more. And – if it permits itself – it will continue to do so indefinitely into the future.
References Brunet, M., Guy, F., Pilbeam, D., Mackaye, H. T., Likius, A., Ahounta, D., Beauvilain, A., Blondel, C., Clyde William, Bocherens, H., Boisserie, J.-N., de Bonis, L., Coppens, Y., Dejax, J., Denys, C., Duringer, P., Eisenmann, V., Fanone, G., Fronty, P., Geraads, D., Lehmann, T., Lihoreau, F., Louchart, A., Mahamat, A., Merceron, G., Mouchelin, G., Otero, O., Campomanes, P. P., Ponce de Leon, M. S., Rage, J.-C., Sapanet, M., Schuster, M., Sudre, J., Tassy, P., Valentin, X., Vignaud, P., Viriot, L., Zazzo, A., & Zollikofer, C. P. (2002). A new hominid from the Upper Miocene of Chad, Central Africa. Nature, 418, 145–151. Deacon, H. J. & Deacon, J. (1999). Human Beginnings in South Africa: Uncovering the Secrets of the Stone Age. Walnut Creek, CA: Altamira Press. Eckhardt, R. B. (2000). Human Paleobiology. Cambridge, UK: Cambridge University Press. Eldredge, N. (1985). Unfinished Synthesis: Biological Hierarchies and Modern Evolutionary Thought. New York: Oxford University Press. Eldredge, N. & Gould, S. J. (1972). Punctuated equilibria: An alternative to phyletic gradualism. In T. Schopf (Ed.), Models in Paleobiology (pp. 82–115). San Francisco: Freeman, Cooper & Co. Eldredge, N. & Tattersall, I. (1975). Evolutionary models, phylogenetic reconstruction, and another look at hominid phylogeny. In F. S. Szalay (Ed.), Approaches to Primate Biology (pp. 218–242). Basel: Karger. Gargett, R. H. (1989). Grave shortcomings: The evidence for Neanderthal burial. Current Anthropol., 30, 157–190. Gould, S. J. (2002). The Structure of Evolutionary Theory. Cambridge: Harvard University Press. Holloway, R. L. (2000). Brain. In E. Delson, I. Tattersall, J. A. Van Couvering, & A. S. Brooks (Eds.), Encyclopedia of Human Evolution and Prehistory (pp. 141–149). New York: Garland Publishing. Klein, R. G. (1999). The Human Career. Chicago: University of Chicago Press. Lieberman, D. E. & Shea, J. J. (1994). Behavioral differences between archaic and modern humans in the Levantine Mousterian. Am. Anthropol., 96, 300–332. Martin, R. D. (1982). Human Brain Evolution in an Ecological Context. 62nd James Arthur Lecture on The Evolution of the Human Brain. New York: American Museum of Natural History. Mayr, E. (1982). The Growth of Biological Thought : Diversity, Evolution, and Inheritance. Cambridge, MA: Belknap Press.
Ian Tattersall
McBrearty, S. & Brooks, A. S. (2000). The revolution that wasn’t: A new interpretation of the origin of modern human behavior. J. Hum. Evol., 39, 453–563. Schick, K. D. & Toth, N. P. (1993). Making Silent Stones Speak: Human Evolution and the Dawn of Technology. New York: Simon & Schuster. Schwartz, J. H. & Tattersall, I. (2003). The Human Fossil Record, Vol. 2. Craniodental Morphology of Genus Homo (Africa and Asia). New York: Wiley. Sponheimer, M. & Lee-Thorp, J. A. (1999). Isotopic evidence for the diet of an early hominid, Australopithecus africanus. Science, 283, 368–370. Tattersall, I. (1986). Species recognition in human paleontology. J. Hum. Evol., 15, 165–175. Tattersall, I. (1992). Species concepts and species identification in human evolution. J. Hum. Evol., 22, 341–349. Tattersall, I. (1998). The Origin of the Human Capacity. 68th James Arthur Lecture on the Evolution of the Human Brain. New York: American Museum of Natural History. Tattersall, I. (2000). Once we were not alone. Scientific American, 282, 56–62. Tattersall, I. (2002). The Monkey in the Mirror: Essays on the Science of Becoming Human. New York: Wiley. Tattersall, I. & Schwartz, J. H. (2000). Extinct Humans. Boulder, CO: Westview Press. Thieme, H. (1997). Lower Palaeolithic hunting spears from Germany. Nature, 385, 807–810. Valladas, H., Reyss, J. L., Joron, J. L., Valladas, G., Bar-Yosef, O., & Vandermeersch, B. (1988). Thermoluminescence dating of Mousterian proto-Cro-Magnon remains from Israel and the origin of modern man. Nature, 331, 614–616. Wolpoff, M. H. (2000). Paleoanthropology. New York: McGraw-Hill.
Chapter 9
Space, time and cognition From the standpoint of mathematics and natural science* Francis Bailly and Giuseppe Longo CNRS – Ecole normale supérieure et CREA, Ecole Polytechnique, Paris
Introduction This chapter offers a twofold epistemological analysis of the concepts of space and time: Part I frames them in the setting of contemporary physics, Part II deals with their role in biology and especially in the project of its mathematisation. Both investigations are closely connected with questions in cognitive science. The issues involved in the analysis of the foundations of mathematics and the natural sciences have profoundly affected approaches to human cognition and treatment of these foundational questions forms an indispensable preliminary to our whole understanding of the cognitive sciences. Contemporary physical theories have led to a steadily more pronounced geometrisation of physics, the counterpart of which has been a steadily more pronounced physicalisation of geometry. This is clearly illustrated in general relativity, where the geometrisation of gravitation (the trajectories of objects are described as geodesic curves in a Riemannian manifold) can equally well be interpreted as the physical realisation of a mathematical structure (the spacetime curvature is determined by the distribution of energy-momentum). This geometrisation is seen even more clearly in quantum field theory, where the introduction of non-abelian gauge fields to give an account of the dynamics of interacting fields has led to the development of an intrinsically non-commutative geometry (see Connes 1994). As for the epistemological status of space-time concepts, the mathematical specification of geometric notions can be seen as a process of the objectivisation of the forms of intuition of our phenomenal awareness. Indeed these very forms of intuition, just as much as the mathe-
Francis Bailly and Giuseppe Longo
matical specification of the structures of space and time, are to be investigated within the setting of specific contemporary physical theories. When we turn to the role of mathematics in biology, the constitutive role which mathematical concepts play in physics is in contrast to their prevailing conceptual status in biology. The various affordances and regularities which experience furnishes are transformed in physics into very rich mathematical structures – structures far richer than suggested by the ’symptoms’ through which our senses and/or physical instruments apprehend the physical world. Moreover, these mathematical concepts, rather than being merely descriptive, play a regulative role in constituting our concept of physical reality. One can say nothing of the subject matter of relativity, of quantum theory, or of the general theory of dynamical systems (the heart of theories of critical states and phase transitions) without mathematics. In biology, by way of contrast, one is struck by the enormous richness of structure with which living systems as given to us in phenomenal awareness are already endowed, and the fact that their theoretical formulation in terms of mathematical concepts suffices to model only certain aspects of that structure, and then in a manner which tends to fragment their organic unity and individuality and fails to do justice to their immersion in wider ecosystems. If we reflect on the role of mathematics in human cognition we are thereby led to re-examine its role in biology, since living systems are the starting point of all reflection on cognition. Nevertheless, despite these differences and granted the lesser extent of overall mathematisation in biology, one can recognise in many areas of biological research an apparent movement towards what may loosely be termed ‘geometrisation’. Questions involving our understanding of spatial concepts are posed not only in the study of macromolecular structures (e.g. the sequencing of DNA base pairs and the resulting expression of genetic effects, or the investigation of the spatial structures of proteins or prions) but also within developmental biology (in the study of the effects induced by spatial contiguity in embryogenesis for example) and in the study of organic function (e.g. the fractal geometries affecting the boundaries of the membrane surfaces engaged in the regulation of physiological functions ) and also in the study of population dynamics and its associated environmental context. Alongside these areas involving spatial understanding, the examination of temporal concepts is also strongly implicated in such areas as the study of the response times to external stimuli, the iterative character of internal biorythms, and in the study of synchronic and heterochronic patterns in evolutionary bi-
Space, time and cognition
ology, the outcome of which has been a recent formulation of synthetic theory of evolution itself. What connection can we trace between the roles of spatial concepts in physics and in the life sciences? The conceptual scaffolding of modern geometry is itself rooted in the conditions of possible actions and experiences which are a basic aspect of our presence in the world. It has at its foundation an inseparable intertwining between (i) our presence in the world as sentient creatures and centres of inter-subjective awareness (as suggested by Husserl), through symbolisation and abstraction, and (ii) the evolutionary leap to which this capacity for rational thought and creative imagination has led. Such a constitutive braid connects the phylogenesis of humans to their ontogenesis as cultural beings in history, via the stabilisation of inter-subjectivity through language. In this perspective we should also view the semiogenesis of conceptual constructions that arise in mathematics and physics. Without the initial spatiality of actions (especially gestures, with their intentionality) and the dimensionality of our primal imagination and cognition, we could never have arrived at the idea of a ... 10-dimensional manifold, in terms of which the theory of superstrings in quantum physics is elaborated. In Part 1 we analyse the notions of space and time as characterised by three types of physical theories: relativity, quantum theory and the theory of dynamical systems. In Part 2 and Part 3 each of the authors independently (Part 2: G. Longo; Part 3: F. Bailly) examines the same notions in connection to theoretical biology. We conclude by putting forward a tentative categorisation, in abstract conceptual-mathematical form, of the manner in which space and time operate as invariants in determining our forms of knowledge.
Part 1. An introduction to the space and time of modern physics . Taking leave of Laplace The physics of the nineteenth century carries the imprint of Laplace. His achievements in mathematics, physics and philosophy marked the moment at which the development in the direction of modern physics, initiated by Galileo, Descartes and Newton, reached maturity. Laplacean mechanics is organised in the framework of an absolute background space with Cartesian co-ordinates in which the motion of bodies is governed by the laws of Galileo and Newton. The perfection of this mechanica universalis is completely expressed through eternal mathematical laws. To cite Laplace: An omniscient being with perfect knowl-
Francis Bailly and Giuseppe Longo
edge of the state of the world at a given instant could predict its entire future evolution with perfect precision. But what counts for even more in Laplace’s work, for us earthbound and imperfect beings, is not this divine, and unachievable, knowledge but the approximate analysis of (possibly perturbed) systems. If one knows the state of a physical system to a certain degree of approximation, one can in general determine its evolution to an approximation of the same order of magnitude. In this sense, according to Laplace, mathematics rules the world and permits the prediction of its future state, by a finite and complete system of differential equations. In fact the analysis of the perturbations of planetary orbits was one of the chief impulses driving the development of nineteenth century mathematical physics. As for causality, in Laplace’s approach, determinism implies predictability. The development of twentieth century physics has taken a quite different direction. Relativity, quantum theory and general dynamical systems have led to an entirely distinct set of concepts and inspired a quite different philosophy of science from that which prevailed in the nineteenth, in particular as for causality. We cannot say the same of the mainstream in the cognitive sciences. Turing, in his seminal article founding the strong AI program and setting out the functionalist account of cognition, made the explicit hypothesis underlying his generalised discrete-state machine (the “Turing machine”): by its discrete nature, full predictability is possible, in the sense of Laplacean determinism (Turing 1950). The Laplacean idea of a finite and complete set of rules is thus consciously placed at the heart of the game of simulation (envisaged in the Turing test) through which he set out to demonstrate that the functioning of the brain was equivalent to that of a Turing machine.1 In fact the notion of a deterministic program, as it emerged in the work of the logicians of the 1930s (the theory of computability was developed by Curry, Church, Turing, Kleene and others in the years 1930–1936) is inherently Laplacean, as clearly spelled out by Turing. That is to say, it implies complete predictability of the states of a computer running a program (see Longo 2003a). From this ideal model, which stems from the logical calculi of formal deduction rather than from physics, the Laplacean paradigm of the brain as a Turing machine running a program has been transferred to the study of cognition in the biological setting. It is of crucial relevance to the project we are pursuing here that the abstract description of a Turing machine is in no way dependent on our understanding it as a spatial structure. The “Cartesian” dimension of its material being has no influence whatever on its expressive powers. Moreover,
Space, time and cognition
its internal clock records a sequence of discrete states in an absolute Newtonian time. It was explicitly invented and behaves as a logical machine, not a physical mechanism (see Turing 1950; Longo 2003a). By contrast, the analysis of space and time and their dimensionality is at the heart of any analysis of physical phenomena. In relation to any claim that living systems and their mental activities can be “reduced” to physics we ought to ask: to which physical theory? Which physical laws have to be employed in the analysis of biological and cognitive phenomena? Functionalism is the still prevailing approach to cognition and biology (the “genome is a program” paradigm, for example) and implicitly refers to a Laplacean causal regime. . Three types of physical theory: Relativity, quantum physics and the theory of critical transitions in the behaviour of dynamical systems
Relativity Relativistic theories introduce a 4-dimensional spacetime in which conservation laws and relativistic causal principles are described in terms of invariants with respect to the relativity group of the theory. In special relativity (SR), the objects of the theory and the space-time structure are given together with their invariance properties under the group of rotations and translations in this space (the Lorentz-Poincaré group). In general relativity (GR), physical space is described as a Riemannian manifold of all possible locations together with its dimensionality and symmetry properties. The metric coefficients are the gravitational potentials just as the local curvature of the Riemannian manifold is the energy-momentum. Thus geometry constitutes the invariants we name as “objects” and “physical laws”. It is not just that physical concepts acquire meaning within the framework of a mathematical space – the latter actually prescribes a thoroughly structuralised notion of objecthood and objectivity as invariants of geometrical structures. In metric spaces, which carry the record of and themselves serve to record the cohesion of and between objects (the stability of physical laws and the conservation of energy and momentum), symmetries and geodesics shape the physical content of the theory. Noether’s theorem describes these physical invariants in terms of space-time symmetry groups. Energy conservation for example is closely tied to invariance under the symmetry group of temporal translations, just as the geodesic curves furnish the trajectories along which quantities are conserved (inasmuch as they are stable minimal paths). See Bailly et al. (1999) and Bailly (2002).
Francis Bailly and Giuseppe Longo
The underlying unity of SR (which unites electricity and magnetism) and GR (which unites gravitation and cosmology) is reflected in the fact that SR may be considered a particular limit of GR. Once again we see geometry providing the framework for actually constituting new structural invariants and unifying them in the same space inasmuch as the stable properties of physical systems with that structure arise in connection with new groups of spatiotemporal transformations. But there is also another path in the direction of increasing mathematical abstraction: the generative role of mathematical ideas provides the basis for grasping the sense of new physical concepts, indeed constitutes it. Take, for example, the physical applications one can find for the compactified (numerical) real line: one takes the infinite real line and transforms it into a circle, by adding one point (which “represents” infinity). On that basis one passes from 4 dimensions (3 of space plus 1 of time) to 5, but this fifth dimension is derived mathematically from the Lagrangean action associated with a field which is both electromagnetic (hence classical, i.e., non-quantum) and gravitational (involving the unification of the Maxwell and Einstein equations). The physical properties carried by this new dimension of space are compactified – the fifth dimension is folded over on itself in the form of a circle: Kaluza-Klein theory (see Lichnerowicz 1955). The geodesic principles and the symmetries are conserved. The observables of the theory have not changed, because the fifth dimension of this spatial structure is below the threshold of observability – it is a pure consequence of the conceptually generative capacity of the mathematical formalism. At the same time this new dimension contributes to explanatory power, for it allows us to unify the structure of theories which were formerly quite distinct, while exactly preserving the invariants (energy-momentum etc.) which were at the heart of the two approaches. It is mathematical geometry which provides us with this new physically intelligible space; and, through this geometrisation of physics, mathematics plays a role of extraordinary explanatory power. In fact it supplies the models of space and time furnishing the framework for physical phenomena and gives them meaning, Kaku (1994). The required mathematical ideas are not laid up in advance in a Platonic heaven, but are rather constituted within the interface between ourselves and the world which they serve to organise conceptually. Recall the role of Riemannian geometry in organising the framework of relativistic physics.2 Relativity indeed furnishes one of the most beautiful examples of this mathematical constitution of phenomena: the most stable and coherent part of our conceptual apparatus – mathematics – provides the framework for a structuralised con-
Space, time and cognition
ception of objects, space and time which undergoes reciprocal adjustment as it encounters that source of friction (the world) which is continually suggesting/imposing new regularities to be incorporate in the structure, and drawing our active conceptual construction toward some models or deflecting it from others.
Quantum physics Relativistic theories present space-time as external to physical objects, aiming to understand the latter as singularities of a field, and their evolution as controlled by geodesics. In this case, their phenomenal appearance amounts to nothing more than the mathematical stability of the invariants attached to these geodesics. Quantum mechanics on the other hand adds to this external frame of reference (Minkowski space) an internal frame of reference expressed in terms of quantum amplitudes and their invariants. This internal frame of reference is essential because the atomicity implicit in quantum theory is a matter not, as in classical atomism, of smallest possible bodies in space, but rather an atomicity of the processes determining the evolution of the field (because the dimension of Planck constant is that of an action, i.e., energy multiplied by time). It is thus the variation of energy in time which is discretised in quantum theory and not the structure of matter or of space-time. Space and time remain continuous, as in relativity,3 and this remains true, in certain respects, of quantum fields, although they behave in a different manner from classical fields. However, the mathematical unification of the theory of quantum fields with that of the gravitational field is far from being accomplished. Our understanding of global or external spaces is profoundly bound up with that of local or internal ones: particles, as much as fields, display counterintuitive non-local effects. Like St. Anthony, it seems quantons4 can be present simultaneously at widely separated locations. This behaviour is not magic: matter fields are not local – they are not reducible to space-time singularities as in GR. Furthermore, matter includes fermionic fields. On this point, debate centres on the relationship between internal and external spaces – and the debate is very lively, notwithstanding the Einstein-Podolsky-Rosen Paradox which had appeared to demonstrate the opposition between GR and the physics of quanta. Briefly, quantum mechanics, which in first approximation had appeared to bring no essential new element to the determination of our theoretical notions of external space, has nevertheless introduced a novel (and counter-intuitive) perspective on our notion of locality. On the one hand, the physical laws of quantum mechanics remain local in the sense that the evolution of a system between measurements is generated by partial differential
Francis Bailly and Giuseppe Longo
equations. On the other hand, the characteristics of the probability amplitudes associated with the state vectors (complex numbers, the superposition principle) engender a non-separability in the properties of quantum systems which is bound up with measurement and corroborated by experiment (Bell inequalities and the Aspect experiment concerning quantons which have interacted in the past).5 Despite the absence of theoretical unification, there are mathematical invariants which carry over from the local to the global frame of reference and vice versa. For instance a global shift in the frame of reference does not alter the electric charge: certain measurements are locally and globally invariant (in the theory of gauge fields) and the fields themselves are associated with local gauge invariants. Super-symmetric theories best tend to illustrate the connection between internal and external spaces. In these theories one can adjoin further dimensions to the four of relativistic space-time, in the manner of the KaluzaKlein compactification of space, with the aim of preserving, as far as possible, the space-time symmetries; recent theories of quantum cosmology have sought to unify the theories mentioned here, in a tentative yet very audacious manner, at the level of the Big Bang by representing space as a six-dimensional manifold in which four dimensions would expand (the four-dimensions of the observed universe) while the compactification of the other dimensions provides for the way in which the properties of matter (fermionic fields) and interactions (bosonic fields) are structured. We should also mention the possible role of the non-commutativity of quantum measurements (the complementarity of position and momentum): a fundamental difference from classical and relativistic approaches. A very promising framework for unification has been proposed via a geometric approach (by Alain Connes, in particular). The idea at bottom consist in reconstructing topology and differential geometry by introducing a noncommutative algebra of measurements (the Heisenberg algebra) in place of the usual commutative algebras, see Connes (1994). Once again, the geometric (re-)construction of space has the effect of making (quantum) phenomena intelligible.
Dynamical systems and their critical behaviour The physical theories of the type we next consider are concerned with dynamical systems which, for some values of the control parameters (e.g., temperature), display discontinuous or divergent evolution (phase transitions such as the freezing of liquids), progressive transition from ordered to disordered states (as in paramagnetism and ferromagnetism) and qualitative change in their dy-
Space, time and cognition
namical regimes (such as bifurcations of phase-space trajectories or transitions from cyclic to chaotic behaviour). They may be regarded collectively as theories of phase transitions. In approaching the question of causality by the status of space and time in these theories, we must distinguish between two classes. Firstly the class of theories concerning systems which possess a high number of degrees of freedom – the phase space is therefore very large (as in thermodynamics and statistical mechanics). It was in relation to this class of theories that problems relating to temporal reversibility and irreversibility were first posed. The second class of theories is concerned with non-linear dynamical systems which can be treated only in terms of a small number of degrees of freedom, and the properties of whose dynamics (bifurcations, existence of strange attractors etc.) are associated precisely with the nonlinearity (whether treated within the framework of continuous differential equations or via discrete iterative procedures.) These systems also pose questions of reversible or irreversible behaviour, but in slightly different terms from those in the first class. In both cases, and in contrast with the situation prevailing in relativistic and quantum theories (where we find ourselves in a fairly regular universe), here our attention is more on the singularities than the regularities of the systems in question.6 Both these classes of theories mark an apparent return to more classical conceptions of space and time than those encountered in connection with relativity or quantum theory. In particular, the introduction of spaces with a large number of dimensions (such as the phase space of statistical mechanics) does not involve their fulfilling the sort of constitutive role assigned to space-time structures in relativity or quantum theory. Nevertheless these two classes of theories have also given rise to new approaches to physics, this time relating to aspects which are, on the one hand, in relation to space, markedly morphological and global; and on the other hand, in relation to time, markedly evolving and directional; and this marks the causal relations. Yet, these systems are characterised by numerous other traits. One is the role they frequently assign to the global aspects. If one takes the “most simple” dynamical system, three bodies together with their associated gravitational fields, the very unity of the system prevents its being analysed in Laplacean terms. One cannot know/predict the position and momentum of each body without at the same time analysing the same parameters for the others. They are correlated through their mutual gravitational fields so as to constitute a sort of holon: a global configuration which, evolving in time, determines the behaviour of each of its elements. A step-by-step analysis – that is to say analysis of the behaviour first of one body, then two... or the approximation of
Francis Bailly and Giuseppe Longo
that behaviour via Fourier analysis – is simply not possible here. This is what robs the system of the kind of completely predictable behaviour conceived by Laplace. What wrecks this Laplacean predictability is that in sufficiently complex dynamical systems (in the three bodies problem rather than the two) divergences are present (i.e., discontinuities related to control parameters). The nonlinearity of the mathematical representation reflects the intrinsic unity of such systems. The dramatic change, as for knowledge and causal regime, is due to the fact that determination, under a finite set of equations or inference rules, does not imply predictability. In fact dynamical systems are often assigned their proper time in a “peremptory” fashion. Insofar as they exhibit phase transitions, by the bifurcations (particularly that of forms in space) as well as their transitions from cyclic to chaotic regimes, these “impose” directionality on the states of the system, differently from other physical theories. Their time is orchestrated by phase transitions and, irreversibly, by bifurcations and transitions to chaotic behaviour. The essentially irreversible character of time for these systems marks a definite contrast with the picture of time in relativity (where it is intrinsically reversible and its flow is regarded as an epiphenomenon), and it seems to provide a concept of time appropriate to living beings (strongly affected by thermodynamic phenomena amongst others). The irreversibility of time characteristic of such “critical” systems is connected with their unpredictability and their chaotic behaviour. . Some remarks We have examined aspects of the geometrisation of modern physics. The mathematics of space and time moulds a framework for the understanding and organisation of phenomena and the unification of different “levels” of their structure. The epistemological and mathematical aspects of space and time turn out to be profoundly bound up with one another in a manner which plays a pivotal role in shaping scientific enquiry, in particular in providing for the unification of physical theories. We have briefly mentioned the (pre-quantum) unification of electromagnetism (governed by the Lorentz-Poincaré group) and gravitation (governed by the group of diffeomorphisms of GR). More recent theories introduce new symmetries (super-symmetries or symmetries of spacetime structure in a generalised sense, associated with the notion of super-space) allowing the articulation within a common framework of the external and internal spaces of quantum systems. From an epistemological standpoint, the unifying aspect of these theories is that they lead to the construction of unfa-
Space, time and cognition
miliar spaces whose physical relevance is then corroborated by experimental investigation. More recently still, a non-commutative geometry has been forced on us by quantum measurements and we have hence been led to propose geometric structures even further removed from the ones directly suggested by the sensible world. Geometry provides a mathematical framework organising the practical as well as the theoretical aspects of our spatial experience. Our access to space as expressed in the most developed physical theories is based on measuring instruments very far removed from our naive sensations and hence necessarily follows a route to the (re-) construction of our notion of space very different from what these might suggest. The curvature of the light is detectable only through sophisticated astrophysical measurements, it is not apparent to our raw intuition. The geometry of the universe rests on a geodesic structure quite unfamiliar from the viewpoint of sensory experience. The nonlocality of quantum phenomena follows from microphysical measurements quite inaccessible at the level of our physiology. It is even possible that our geometry itself will take the form of mathematical structures in which the classical notion of a point is no longer basic (e.g., the theory of superstrings or twistor theory). Notice besides this that the generalisation – via homotheties – to all physical scales and dimensions, of all Euclidean properties drawn from our sensory experience is quite arbitrary, see Longo (2003). Straight lines and dimensionless points do not exist (or “exist” in only the same sense as any other mathematical construct or abstraction). They can be replaced by other abstractions which may turn out to hang together better with experimental evidence and with new tools of measurement. One last word about theories of dynamical systems, near to or undergoing critical change. The treatment of space-time these theories suggest (centred on phase space and the transition of their dynamics from stable to chaotic behaviour) introduces new elements important also for other theories, above all in connection with certain recent cosmological theories (models of the phase transition associated with the Big Bang, singularities and cosmic strings, for example) and also in connection with the relations between local and global structure. This class of theories forms the key bridge between physics and the life sciences – and also (if we may skate fearlessly over a great many intermediate levels of organisation) with their great associated critical “epiphenomena” – namely, cognitive phenomena.
Francis Bailly and Giuseppe Longo
Part 2. From physics to biology: Space and time in the “field” of living systems . The time of life As a preliminary, I want to analyse the particular features of time characteristic of living systems. Temporal irreversibility is at the heart of the study of dynamical systems exhibiting critical behaviour, but it is also characteristic of living systems. At every stage phylogenesis and ontogenesis are marked by ’bifurcations’ and by the emergence of unpredictable phenomena and structures which resemble those observed in critically sensitive dynamical systems, which thus subsume biological phenomena. Moreover, living systems contain a great many subsystems which display this kind of critically sensitive behaviour – dynamical and thermodynamical. These contribute not only to the temporal irreversibility of the system but also to a kind of unity which is apparent in the kind of dynamical systems we touched on above in connection with the threebody problem. Poincaré’s three bodies, in exhibiting an example of this kind of unity, form a primitive Gestalt associated with a purely gravitational interaction. Two bodies exhibit a quite different dynamical behaviour, stable and predictable. It could even be said that what comes into play in the three-body dynamical regime is a kind of emergent behaviour, a unity of non-stratifiable relationships: one cannot analyse first the position, then the velocity, of each body step by step, independent of the unity of the system they form. In a recent email exchange, F. Bailly remarks: The spatial and temporal (and spatiotemporal) terms do not appear to possess the same significance or play the same role within the two principal approaches (“geometric” vs “algebraic-formal” which you have distinguished). In the “geometric” approach, space is the correlate of geometry itself, it intervenes at the perceptual level. Time is the time of genesis of structures, the recording medium of their process of constitution. In the “algebraic-formal” approach, by contrast, spatiality is the echo of an abstract linguistic inscription, of formal symbols, while temporality seems to be principally a matter of sequential functioning, of the execution of algorithmic calculation.
This remark refers to the distinction, which in other writings I have drawn in the context of the foundations of mathematics, between principles of construction (in particular those with a geometrical aspect) and principles of proof (formal principles of logic), see Longo (1999, 2002). Mathematics is built up on the basis of both types of principles. The philosophical fixation, implicit in the analytic tradition, with logicism and formalism has tended to exclude
Space, time and cognition
or sideline the first of these. The “linguistic turn” has given us extraordinarily rich logical/formal machinery (and literally machinery in the form of digital computers) but it has also endorsed the myth of the complete mechanisation of mathematics, indeed of any form of knowledge. I have argued that incompleteness theorems in formal systems are due to the gulf between these two types of principles. Conceptual constructions based on spacetime regularities possess an autonomy, an essential independence in relation to purely formal descriptions, in a sense made exact by mathematical logic (through the work of Hilbert and his school). Unfortunately physicists are prone to label any “mathematical treatment” of a subject as an instance of “formalisation”. For logicians these are quite distinct notions: there is the Gödelian (and other forms of) incompleteness in between, see Longo (2002), Bailly, Longo (2003). The distinction of principles of geometric construction from algebraic-formal principles of proof is in my view one of the crucial factors which underlies the constitutive role of space-time concepts and geometry in the analysis of cognition. In the conception of time as the medium of algebraic manipulation and formal calculation, as seen in the sequential running of a computer program, one recognises an important fruit of the formalist view of the foundations of mathematics. The 1930s marriage of Hilbertian formalism, together with the problems it addressed (the completeness and decidability of formal systems) and, on the other hand, a mechanistic positivism, was at the origin of the attempt to treat human rationality in terms of a mechanism which indefatigably executes formal algorithms. But this forgetting of space, which also greatly influenced the characteristic mathematical approach to time (as the medium of the genesis of structure) led to the severe reduction of the analysis of human cognition and, which is a greater distortion, of animal cognition – humans can use logic and formal calculi as supplementary cognitive aids which permit a biased grasp of at least a part of what is involved in understanding, but it is just this part which is least accessible to other living beings. What marks an interesting historical reversal of this trend is that today we cannot study or seek to construct computers without taking account in a new way of considerations involving space and time. The geometric aspects of the structure of computers enters into the study of distributed, concurrent and asynchronous processing, which areas pose spatio-temporal problems of a kind completely foreign to the theory of Turing machines, the theory which dominated the study of computability from the 1930s to the 1980s and the very interesting mathematical aspects of which for a long time formed my own
Francis Bailly and Giuseppe Longo
principal field of study. In these new areas the main problem concerns the time of structural genesis and the constitution process. This is a kind of time which involves space and which thus poses a new set of problems for computer science as well as for physics. Is this a further aspect of the new role of geometry in the study of cognition? Should we think of the time of cognitive processing as an inherently distributed time? Finally, where is the living system which does not exist other than in space and time? Take the dynamical self-organisation of ecosystems for example. Their genesis is above all a genesis of structure, from protein folding to the morphogenesis of an elephant; and their time is the history of a process of constitution. Dynamic irreversibility, Gestalt, systemic unity and cohesiveness – what happens to all these are aspects of living systems which act in space and time? . Three forms of time In the foregoing remarks, we have the outline of two ways in which we can regard phenomenal time as constituted – phenomenal, because it is jointly construed by us-and-the-world: it is a constitutive element in our forms of knowledge of a Reality-out-there, but one which must be endowed of structure to become intelligible. This time is at once a real and a rational time, remarkably, but not absolutely objective. It is the co-construction of the knowing subject and the world, as rooted in the regularities which we detect in the world – regularities which are out there but the explanation and the (scientific) objectivity of which are constituted in intersubjectivity – an intersubjectivity with a history. Let us now examine these two forms of phenomenal time more closely, with the aim of suggesting a third. The first form, algebraic-formal, is that of clock mechanisms – the same clocks which the Enlightenment regarded as a possible model for the operations of the understanding in general – and which later became the time of a (discrete state) Turing machine (see Longo 2003a, for the “discrete vs continuous” issue in computational models of mind). A Turing machine tells time by the movement of scanning/reading its tape – to the left or right – tick-tock – like an absolute Newtonian clock. Nothing happens between one movement and another (to the left or right as the tape is scanned) nor can anything be said of their duration: these movements are the measure of time itself. This notion recalls the time of myth inscribed in Homer. To recall an analogy suggested by Bernard Teissier: during the Trojan War, time is marked by the sorties of Achilles from his tent. Achilles leaves his tent, something (the War)
Space, time and cognition
happens; he re-enters his tent, everything stops – time stops. Achilles’ motions provide the (only) scansion of time. Troy and the Trojan war (in the sense relevant here) lie outside the (space of the) world – they exist in the realm of myth. Their internal time contributes to an extraordinary poetic effect. Turing and Homer are as one in this respect: the time characteristic of 1930s formalism is the time of algebraic-formal construction – the absolute time of a formal mechanism lying outside space, the time of “calculation-in-itself ” is the time of one step after another in a void. In fact this time is secreted by the actions of a Turing machine viewed purely as a clock. But Greek Thought proposed another representation of time as Kronos, son of Ouranus. Kronos (derived from chaos and devouring his children) is “true”, physical, time – the “paddle” of the real world. This version of physical time fits well with the analysis of dynamical systems displaying critically sensitive behaviour (e.g., characterised by phase transitions). It is a time in a space – the space of the geometry of dynamical systems, a time recorded by their bifurcations, by their irreversible transitions from stable to chaotic regimes. Indeed it is the time of “the genesis of structure”, of constitutive process, because a bifurcation, or a catastrophe, can depend on the entire history of a system, and not only its state description at a given instant. To represent time as a linear continuum, the line of the real numbers, is very convenient; in many contexts one can choose no better model. But I here take up the reasons for the dissatisfaction with it which Hermann Weyl expressed in Das Kontinuum (1918). Its “points” cannot be isolated in the manner of points on a spatial line because the present blends into, and indeed has no meaning except in conjunction with, the past and the future. While giving substantial contributions to the mathematical setting of relativity, Weyl recognised the limitations of relativity theory to represent time as an epiphenomenon, given that time is equipped with the same structure as the spatial continuum. Moreover, reversible time, due to the equations of relativistic physics, has nothing to do with phenomenal time as a mixture of experienced and rational time. The time of dynamical systems theory and theories of critical states seems much better adapted to capture irreversibility than that of relativity theory (and perhaps better than that of quantum physics). Moreover, the time of catastrophe theory can be given no meaning other than in space (in this respect it is like the time of relativity theory): firstly, bifurcations and chaotic behaviour require space for their manifestation; secondly, there is no such thing as the time of a single isolated dynamical system displaying linearity in its bifurcations. No such system exists. The genesis of structures proceeds in parallel, through
Francis Bailly and Giuseppe Longo
interaction of a plurality of structures (sub- and super-systems) in a spatial setting.7 There are exceptions to the immersion of this second form of time in space. One could say, for instance, that the grammatical structure of natural languages, and other aspects of their structure possess a history and an existence in time without making reference to space. But language is an intrinsically intersubjective phenomenon – it is a plurality of speakers, situated and acting in space, which makes language possible. There is no language of an isolated speaker, language is always spoken within a cultural ecosystem, which is often in friction with other cultures. As the temporality of physical systems is associated with the genesis of structures in space through the interaction between systems which are both dynamical and distributed, the synchronisation of such systems becomes a central problem (though one can have asynchronous physical interactions of course). Already inn relativity it shows up in the exchange of signals between differently accelerated systems. In computer science, this problem is partly bound up with the analysis of concurrence between processing units distributed in space. Both the time of Turing machines and that of Achilles’ sorties requiescant in pace. Today we have a more “structured” time – that of a plurality of dynamic, distributed and concurrent (or more generally interacting) systems with their own local times, demanding synchronisation where required. But if there is no time apart from this synchronicity, the same holds for asynchronicity, because it is already inherent in any “real” interaction between systems in a not purely local universe. We are today in a position to propose a notion of time better adapted to our scientific understanding of the physical world: one enriched by the consideration of relativistic phenomena and (irreversible) dynamical systems. This time is essentially relational in character. Just as the absolute space of Newtonian physics no longer seems to make sense, so the absolute clock of the Turing machine, isolated in an empty universe, no longer seems to define an adequate representation of time. They would be akin to the standard metre of Sevres, isolated in an empty universe: in that universe there is no distance, just the metre. But, there is yet a third form of time to be discussed and it is one appropriate for biology. The time in question is a phenomenal time, superposing experienced and rational aspects; it is constituted jointly by ourselves and the world, in the very acts of our intentional experiencing the world. That it manifests resistance to our attempts to grasp it is essential to its understanding. It is not to be thought of as “already there”, yet it is not something arbitrary – because
Space, time and cognition
the regularities which supply us with clues and suggest how to speak about this time are certainly there; it is we however who choose how to regard them. In biology, matters are effectively more complex than in relation to physics, and one is obliged to move away from the idea that our brain (or any living organism) is a logical device or a programmable machine. First of all the “unity” and the “characteristic” time scales of living systems is related to the autonomy of the biological clocks of which F. Bailly gives a detailed account in the following section. This autonomy is even more striking than that of the mechanisms acting as clocks in the case of physics, because of the way in which a living organism strikes us as a unified individual. In physics, the present and future states of a system, and of the world as a set of dynamically interacting systems, depend “only” on past states. But the situation in the case of organic systems involves even more interactivity than that. On the one hand, there are autonomous clocks appropriate to the individual system – its metabolic rate, its various biorhythms (heartbeat, respiration etc). These are constants over ranges extending in some cases beyond entire species, even covering an entire phylum (the mammals for example). Evidently these clocks are far from being isolated systems – they regulate the functions of organisms in interaction with their environment; indeed their raison d’être is to constrain and regulate that interaction. On the other hand there is the phenomenal time of action within space on the part of this same living system – action characterised by aims and purposes, not least that of survival. Before discussing this however, let me review at least two further factors involved in the study of time in biology: 1. the local time of each individual living being, its internal clock(s), which is re-established after any action affecting them (any action within the limits the organism can tolerate). Its clocks indeed exist precisely to permit and to regulate such interactions; 2. a global time in which possible bifurcations in the system dynamics are determined, according to anticipatory capacities of the organism, by choices on which the possible future states (survival) of the system within its environment depend. “Intentionality” is thus characteristic of biological time and it extends far below the threshold of consciousness, as is seen in the behaviour of single-celled organisms which move in one direction or another to preserve their metabolic activity. This movement is one of the most elemental forms of choice: constituting bifurcations between possible directions (paths) of the system in phase
Francis Bailly and Giuseppe Longo
space. In the case of human beings this choice is made on the basis of explicit awareness and conscious anticipation of the future. It thus depends on the range of possible future states considered – Pauri (1999) makes the same point. It is thus a “contingent intentionality”, related to contingent goals of the kind characteristic of different organisms. There is no organism or species without one implicit goal, that of surviving. But this finality is not metaphysical, rather it is immanent and contingent. If it were otherwise, neither the individual nor the species could long survive. It is essential to the preservation of living systems, from a single cell to multicellular organisms, as they are capable of future-oriented actions. Intentionality in the Husserlian sense of the term, involves an envisaging, a mental act consciously directed towards a target. Here it has a broader meaning: it is thus the end result of a network of interactions which plays a constitutive role in phylogenesis. Pachoud (1999) also suggests enlarging the Husserlian notion of intentionality in order to revitalise the phenomenological program. Let us take an example from the study of primates. This example falls midway on the scale between the actions of an amoeba in its metabolic responses and the conscious intentional behaviour of a fully socialised human individual, or even the collective purposeful activity of an entire social group. This is the example: when we switch our attention from one point in our field of vision to another by a saccade (a rapid eye movement), the receptor field of the neurons in our parietal cortex is displaced suddenly, before the ocular saccade, in the direction in which we are thereby looking, Berthoz (1997: 224). In other words the brain, in order to follow the trajectory of an object, or to escape the claws of a predator whose intentions it has “understood”, displaces the receptor field of its neurons and anticipates the consequences of that displacement. This is only one example amongst many which can be given of the role of anticipatory action of the future characteristic of living systems. I consider it of great interest because it is a form of intentional future-oriented behaviour below the threshold of consciousness in animals, but very close to conscious movement. In fact, it seems that the glance actually produces a change in the biochemical (and hence the physical) state of the neurons, in the act of anticipating the future. This new state is imprinted on their structure – the new state in which they are then found does not depend only on their present and past states, but also on their anticipated future state. In what follows, F. Bailly will develop a suggestive analogy between local curvature in the Riemannian spaces of relativity theory and the locality of the internal time scales of living systems. A constant non-zero curvature provides for a local spatial scale linked to the (local) metric exactly as metabolic or car-
Space, time and cognition
diac rhythms appear to provide a time scale, more or less regular but local, observed in the individual system but common to the species or wider phylum. In contrast with the absolute locality of the metric of curved Riemannian space, local biological clocks are embedded in a wider ecosystem, and their contingent finality is not what it would be in the case of an isolated organism; rather they contribute to ensuring the stable existence of the organism in a changing milieu. They synchronise it with similar systems and maintain it when in interaction with dissimilar ones. Whereas constant local curvature furnishes an invariant, local, metric element, independently of what goes on in the rest of the world, the internal clocks of living systems play a role in interaction. They aid in the establishment of a common time scale and they allow for the regulation and synchronisation of other clocks within an ecosystem. . Dynamics of the self-constitution of living systems Any individual organism, or any species, defines what may be termed a zone of “extended criticality” (F. Bailly), which appears to be a feature impossible in physics, where “critical” states are generally unstable singularities. In this zone of extended criticality numerical invariants characterise the time scales of the autonomous system and reorganise the “unity” of the system in relation to heteronomy. When one examines a species embedded in an ecosystem much of the conceptual framework taken over from physics appears inadequate. Although the theory of dynamical systems has furnished some effective mathematical tools for biology, the study of a living system with the methods developed in mathematical physics has conceived the evolution of the system as taking place within a “frozen” field of force, or at any rate within a network of fields of force given at the outset. That is to say, the phase space does not change in the course of evolution. A marble rolling in a cup is a simple classical system. Its field of forces: gravity, the geometric shape of the cup, the frictional resistance – all already in place at the outset. The analysis of the ensuing oscillations follows very straightforwardly. In the case of more complex dynamical systems the mathematical analysis of their behaviour may make reference to so many different forces that the majority of systems turns out to be intrinsically unpredictable. However, qualitative analysis allows us some remarkable insights into their possible evolution (the existence of singularities, bifurcations, attractors and so forth), even in the absence of complete predictability. In the case of a living system a further factor is involved: the field of forces acting on the system is itself constituted in
Francis Bailly and Giuseppe Longo
the course of the evolution of the system. In analysing that evolution one may have to pass from one phase space to a completely different one. Take a species within an ecosystem. Doubtless its interactions with the physical aspects of its ecosystem are determined by forces which relate to those aspects (e.g., gravitation, the physics and chemistry of the atmosphere or of seawater) but within an ecosystem one finds also other living beings. They react on the species in question. In fact species co-constitute themselves in conjunction with one another. They may eat one another for example. And these other species were not necessarily present in the ecosystem before the one being studied, nor are they fixed and frozen entities. Their existence and evolution may itself depend on that of the species under consideration. Living systems in their interaction do not form a given field of physical forces – no minimum principle, no geodesic principle predetermines their evolution. For modern evolution (and we have for the present no better theory) they rather become more or less compatible with a situation which living systems themselves will have co-constituted and co-modified, rather than with one given in advance. Neo-Darwinian evolutionary theory refers to the combinatory explosion of life “in all possible directions”. That is to say, no overall pattern of development in the system is predetermined, still less predictable, except in the case of small laboratory populations (e.g., of bacteria) under very controlled conditions. But, in general, evolutionary behaviour is compatible only with (and could not exist without) the situation which it itself contributes to determining. Novelty arises on the basis of a given situation (which includes a genetic make-up) but also via the establishment of new patterns of interaction, the significance of which cannot be understood prior to their constitution. S. J. Gould mentions, for example, the tremendous role of “latent potentials” – illustrated by the double articulation of the jaws of certain reptiles 200 million years ago, which became the inner ear of birds and mammals. There was no a priori reason why things should have gone this way – no physical field of force and no genetic endowment on the part of reptiles imposed this development – it was made possible in the context of (indeed was co-constituted by) an ecosystem. It would have been impossible to predict. The only reason is a posteriori. We find ourselves further than ever from Laplace and there lies the scientific (mathematical) challenge. Thus novel possibilities modify the field of forces set up by the living ecosystem. It is as if the cup in which the marble was set rolling assumed a shape (even a variety of shapes) from amongst all the physically possible ones, whilst the marble was in motion. But it is even more striking than that, for the marble too becomes extremely malleable whilst at the same time seeking
Space, time and cognition
to safeguard its unity and autonomy, just as all living individuals and species endeavour to do. Briefly, the biological “field” is co-constituted in time. In this respect it is something over and above physical fields; it depends on the latter of course, but is not reducible to them; at any rate we are a very long way from being able to produce such a reduction. The unification of biology with, rather than its reduction to, physics remains a principal aim. But it may be that this looked-for unification will come about from a quite different theoretical direction. It may be that an account of quantum phenomena will emerge within the framework of a general account of systems, including anticipatory capabilities.8 In this connection one will need to enrich the very concepts of “causal determination”, “system” etc. Our aim at this juncture is a conceptual analysis which pinpoints the parallels and divergences between new mathematical models of space. What can be meant by a shift/enrichment of our concepts of “causal determination” and “system”? Let me illustrate it by means of a dialogue at a distance between Galileo and Kepler. Kepler, a mathematician and astronomer of extraordinary gifts, did not disdain the task of compiling almanacs and casting horoscopes and mingled his talents in order to make his living. It was thus he came to think that the moon had an influence on the character of women, and also on the pattern of the tides. Galileo, a man of science through and through, did not agree with these ideas. The first of the problems, however important, had nothing in his eyes to do with physics, and as for the tides, to claim that a distant body like the moon could be implicated in their cause seemed to him to smack of magic and astrology. Sooner than admit this, he set out, in his Dialogo sopra i massimi sistemi to explain tidal motion in physical terms, for the tides were clearly physical phenomena: the tides are the result of inertial forces acting within the framework of Galilean relativity. The combined forces acting on the earth – its rotation around its axis and its orbital motion around the sun – are the cause of the inertial motion of its waters. Galileo’s theory of inertia and the relativity of motion marked the debut of modern physics. But his theory of tides took no account of countervailing empirical data: Galileo’s reasoning would lead one to expect a 24 hour cycle in the tides. His error was one of methodology; an error which with mild abuse of language one could label “physicalism”: a (misconceived) reduction located at the heart of physics itself. By “physicalism” what I intend here is not so much the position that, “in the final analysis”, all the phenomena are of a physical nature and supervene in principle on the physical description of the world by means of (a final) physical theory, but rather the reduction of phenomena to
Francis Bailly and Giuseppe Longo
one given physical theory, constructed on the basis of a priori considerations around a restricted and well-defined range of phenomena other than those which are the target of reductive explanation. To speak in a modern idiom, Galileo’s difficulty was that he lacked the field concept. (What was more serious, he could give no account of what it was his theory lacked or suggest any measurements which could be carried out to test it as it stood.) Granted, he would have had to cover a great deal of ground to arrive at the concept of a field and its accompanying mathematical representation – ground including Newtonian gravity. The modern notion of field did not reduce supra-lunar phenomena to sub-lunar Galilean motion: it proposed instead new mathematical concepts and a novel synthesis. The difficulties involved in the analysis of living systems (and the methodological youth of biology) suggest that in the life sciences (perhaps) and in the cognitive sciences (surely) to we are at a stage analogous to that seen in the Kepler-Galileo debate. Anyone who observes that the range of biological phenomena displays aspects which elude description in terms of current physical theory risks being branded an obscurantist and accused of believing in magic. The situation is not helped by the fact that one does indeed encounter terminology of a magicalpoetical flavour in some writings on this subject. Confronted with this position, some tough-minded commentators cling to the notion of a deterministic program (in the sense of Laplace and Turing) and see it encoded into the brain, as the hardware on which the program, or rather a whole set of interlocking programs, is run. Others take up the issue of quantum non-locality, locating its manifestation at the level of the microtubules of neurons and claiming that this will turn out to form the reductive basis of consciousness. Others again turn to the study of dynamical systems and take this as the framework for modelling the evolution of neural networks and the plasticity of their behaviours. Clearly there are very important differences between these approaches. The first of them nowadays comes within a hairsbreadth of being a swindle. It has long been clear that we see less and less evidence in current physics of the kind of determinism embraced by Laplace and Turing, and even less in biology. This is not to deny the importance of both Laplace and Turing for rational mechanics and information sciences respectively. The second approach sets out a challenge to be taken up, but is currently lacking in experimental evidence, or in linkages between the scales of the structures and systems involved in the hypothesis: between the activity of neurons, which are very large scale structures, and that of the quanta, intermediary levels of description are altogether lacking. The third approach is founded on a
Space, time and cognition
strong body of evidence concerning the workings of the brain – the observed reinforcement of synaptic connections and, more generally, the effectiveness of the dynamical systems framework for the treatment of any interactive system. Here progress has been remarkable, yet the reduction is performed towards a specific physical theory: no novel conceptual unity is proposed. In these three approaches we also see the change of the notion of “determinacy”. For Laplace (and for the sequential programming of computers) any deterministic system is completely predictable.9 Within dynamic systems theory, determinism does not necessarily imply predictability. Quantum physics introduces a further and deep-going modification of the concept, via its dual, the notion of intrinsic (non-epistemic) indeterminacy. In less than two centuries our notions of what it is that determines what, and our notion of what is a system evolving in time, have undergone a profound shift. But we still have no equivalent general notions in biology (see Rosen 1991). We cannot say in what manner DNA determines the ontogenesis of living systems, nor in what way the state of a nervous system determines its later states. In an attempt to tackle these issues with the concepts of present mathematical physics, researchers have entered the conceptual kitchen, so to speak, and are busily drawing up a menu based on the recipes and cooking utensils they have already mastered, a menu drafted, where possible, in collaboration with the biologists. But to make better progress we stand in need of a robust notion of biological field – which is still lacking. . Morphogenesis Let us now turn again to the the notion of space appropriate to the study of living systems. One of the areas in which we see the richest use of geometrical concepts in the study of living beings and their associated ecosystems is in the study of morphogenesis, in which I include the study of the evolution of the forms of living beings and the influence of form on the structure of life in general. “The stability of living forms is geometric in character” (Thom 1972: 171). The topological complexity of a form is for Thom the locus of its “meaning” and of its organisation. Thom assigns an almost exclusive explanatory role to topology: the topological evolution of the form of a living individual provides the explanation for its biochemistry, rather than the other way around, Thom (1972: 175). The form in question contains information in two ways: it determines an equivalence class of topological forms under the action of a group of transformations; and it also supplies a measure of the computational complex-
Francis Bailly and Giuseppe Longo
ity of a system via the number and evolution of its singularities. Here we can glimpse the idea of a “morphogenetic field” which fashions living systems, in the course of their phylogenesis as well as their ontogenesis. Global structure and operations of a global character occupy centre stage in Thom’s view. In the embryo, he emphasises, we already have the global pattern of the organism, from which the specialisation of organs and their function follows. As it has been said (Jean 1994: 270): plants form cells, not cells plants. But just what is this “morphogenetic field”? This expression could lead us astray if we think in terms of the physical fields. The morphogenetic field must be thought of as in some sense containing all the known physical fields at once, together with new fields characteristic of co-constituted organisms. In particular, each field – physical, biological or cognitive, acts at a certain level of organisation, conceptually independent of others: the phenomenal level and its conceptual structuring by our forms of scientific knowledge are completely distinct. However the individual organism achieves de facto integration of this plurality of levels: its unity results from this integration of physical, biological and cognitive levels. These different levels of structure and organisation, analysed by quite different scientific methods and concepts, interact with each other via spatial and temporal linkages. Each level displays plasticity with respect to the others. However no current physical theory supplies the concepts needed to describe these forms of mutual action, control and constraint, operating between the different levels (the ascending and descending linkages between them at all levels of the system and its biological, chemical and physical components and subsystems). Cybernetics, the first theory of control automation, has certainly furnished remarkable models of the linkages involved in self-regulation. But these models have been located specifically at the physical level, and are constrained by the range of the theoretical tools they employ, whereas living systems establish linkages between conceptually wholly distinct levels of description. Thom’s analysis, subsequently enriched by the work of many other researchers, is also directed at the physical aspects of the topological plasticity of living forms, including those aspects induced by their “virtual” interactions. His work deals with an extremely informative physis of living systems, but still a physis. While, on the one side, it views the plasticity seen in the evolution of living forms as constrained by the dynamic fields operating in their morphogenesis, on the other hand, topological evolution is regarded as developing within a physical schema which takes no account of such phenomena as latent potentials or the combinatory explosion of life in all compatible directions. But,
Space, time and cognition
compatible with what? We do not mean compatible with the forces acting on a system at a given instant, but also with those it will experience. This poses a (mathematical) problem which is at the heart of the developmental plasticity of living forms. As has been understood by those who have contributed to the most fully worked-out aspects of the theory of morphogenesis, namely phyllotaxis, it is possible to induce forms very similar to those seen in phyllotaxis by means of superconducting currents imposed on a magnetic field, see Jean (1994: 264). For instance, the Fibonacci sequence, which is observed very frequently throughout the vegetable kingdom, can be reproduced by this method on any mesh of “soft objects” under repulsive forces and strong deformations (1994: 265). In this sense, such an analysis does indeed consider living forms with respect to their being as purely physical systems – that is to say as bodies subject to the influence of physical fields. But although an important and necessary investigation, this is not exhaustive as an analysis of the forms of living systems. Morphogenesis also has an important role to play in helping biology break out of the stranglehold of “genetic chauvinism”. The latter in the writing of some authors takes the form of a near maniacal expression of the LaplaceTuring vision of an absolutely deterministic causality, legislated in advance by the initial configuration of the system’s components. This vision of a closed future is strongly rooted in currents relating genetics and socio-biology (and unhappily congenial to religious believes in predestination). In contrast to such a picture, H. Atlan replied: “the program of a living organism is everywhere except in its genes”: certainly the patterns and the forms seen in phyllotaxis are not entirely in the genes. They are also in the structure of space and time and of physical matter and energy. The genes do not contain all the information on the symmetries which are set up in a system in interaction with its environment, such as are observed in crystals and minerals, Jean (1994: 266). The so-called “program” for the development of an organism is to be found in the interface between its phylogenetic record (its genetic legacy) and its physical and biological environment (its ecosystem). An example of the greatest importance is provided by the brain, which in the course of ontogenesis manifests a developmental pattern which is both Darwinian and Lamarckian. The immense number of possible connections between its neurons (each one of around 100 billion neurons has up to 10.000 synaptic connections, maybe more) could not be (or at any rate very little of it could be) encoded in the genes. Of the numerous connections established very rapidly during the growth of the foetus or the new-born child, most disappear
Francis Bailly and Giuseppe Longo
through selection effects, Edelman (1987). On the other hand, throughout the course of our entire life, stimuli lead the brain to establish new connections and reinforce existing ones, jettisoning and replacing existing connections as it does so, selecting certain neurons and leaving others to die off. Cerebral plasticity, at all levels, is at the heart of the continuity between phylogenesis and ontogenesis, and is what permits continuing individual identity: “the structure of the nervous system carries the material traces of its individual history”, Prochiantz (1997).10 . Information and geometric structure In an epoch of free-floating bits, the picture of information (even of intelligence) as purely a sequence of bits enjoys great currency. The digital encoding of information is of great effectiveness for certain purposes: once encoded, such information can be safeguarded and transmitted with unrivalled accuracy and speed. No method is superior to that of bit-storage in the construction of digital computers and the networks they form, which are now in the course of transforming our world. Moreover a number of notable mathematical results of the 1930s demonstrated that all discrete encodings and their effective treatment are equivalent. Kleene, Turing, Church et al. demonstrated the equivalence of (very) different formalisations of “computability”: the numerical functions calculated by using the systems of Herbrand, Curry, Gödel, Church, Kleene and Turing were the same. By means of an astonishing philosophical sleight of hand, trading on the surprising and technically difficult nature of these results, and influenced by the surrounding intellectual climate of formalist and positivist ideas, the claim was later made that any physical form in which information is processed, and thus any biological form of information processing or any form of intelligence, can be encoded in any such formal system, thus it can be encoded in the form of the strings of 1s and 0s used in the memory stores of digital computers – see Longo (2003a) for more on some parodies of Church-Turing thesis. A quite different way in which information can be thought of as structured, one involving geometric principles, is through equivalence classes of continuous deformations. These provide for the transfer and processing of much of the information essential to the make-up of living beings and, more generally, of physical systems. Continuous, differentiable or isometric transformations and the regularities they preserve or fail to preserve may help to structure and make intelligible living phenomena, as can be seen not least in the geometric structure of DNA or of proteins and their evolution. To these trans-
Space, time and cognition
formations the discrete and quantitative structure of bits of information serves as an addition; bits behave as singularities and thus as a possible measure of the topological complexity of the geometric structure. Information has both a qualitative and quantitative nature. The concentration on only its quantitative, digital, nature has become a severe limitation when information is assigned an explanatory role. A frequent reaction is: yes, granted the role of these kinds of transformation and this kind of continuity, nonetheless, in the last instance, the geometric structures involved are reducible to very minimal discrete units. Such a reaction hides many problems. Firstly, there is a problem about complexity: suppose one tries to describe, by a string of 1s and 0s, for example the three-dimensional structure of proteins exchanged in post-synaptic cascades, plus the biochemical flux in the brain fluid of an animal and the convection currents which accompany it. One faces extraordinary difficulties of principle as well as of practice. Physical and mathematical principles prevent our modelling this continuous and tri-dimensional information in discrete and linear form. The discrete-bits-representation becomes demonstrably intractable. Briefly, information in digital form, even when encoded in our tiniest microprocessors, would cover an area larger than the whole surface of the earth and there would thus be problems arising from relativistic effects obstructing its synchronisation. Moreover, if one entertains the idea of the possible discretisation of all spatio-temporal magnitudes, accuracy of approximation would it yield? The smallest living phenomena comprise dynamical systems (thermodynamic systems, systems with critical points). But this does not imply that a discrete mesh laid down a priori will be sufficient for their analysis. This is because sensitivity to initial conditions typically generates far-reaching consequences at or above the threshold of discernibility, triggered by a variation below that of the fineness of the measure. But what kind of discreteness are we really talking about here? On the assumption we can push the encoding right down into the microphysical realm (so as not to have to cover the whole earth with processors) it seems the discreteness in question will come from quantum physics and will arise at the scale of the Planck length. One then encounters a fallacy – well explained in Bitbol (2000) – of the same stamp as that involved in the case of the formalist and mechanist reduction of mathematics to formal manipulation (processing) of discrete symbolic inputs. The reference to well-defined and ultimate discrete level of “material points” is the conditio sine qua non of Laplacean mechanics. No such appeal is possible in quantum theory, because it is a theory of continuous (quantum) fields, where Planck’s constant, the only possible referent
Francis Bailly and Giuseppe Longo
for these notions of ultimate discretisation, has the dimension of an action – a fibration orthogonal to the continuum of space-time. Moreover quantum indeterminacy and the epistemological debate which has raged around it has involved assigning a role to the knowing subject of a profoundly anti-mechanist kind. In approaching this topic armed with digitised version of microphysical structure one is lending support to the myth of Laplacean mechanism, as a matter of the bit-strings programmed by formal laws of thought. We cannot renounce the mathematics of continua. The deformation of geometric structure can reproduce information in analogue form. And the analogy involved is intentional – one chooses what to represent or reproduce in analogous form, one selects those aspects of the original form which are to undergo processing or simulation. The choice of analogy is the outcome of a controlling vision or an aim, conceptually appropriate to living systems, of a kind which is missing in physics, where the phenomenal arrow of time is oriented without backwards linkages; see, nonetheless, Novello (2001). The fact that the reproduction and transformation of information via geometrical forms is analogue in character and may better accommodate intentionality, appears to be just what is crucial for biological representation. The eventual greater instability over time of geometrical forms by comparison with binary bit-strings is actually an enriching factor, because it corresponds to the possibilities of evolutionary change. The analysis of the geometric structuring of living systems (particularly of the brain) permits us to grasp a factor essential in information: selective analogical simulation carries with it evolutionary possibilities. By contrast with this, the perfect stability of bit-by-bit information processing renders such an elaboration impossible (whether this is a practical impossibility or one of principle is unclear). To conclude: in studying living phenomena, from the most elementary systems all the way up to cognitive agents, it is not so much a matter of denying the important role played by formal and mechanical aspects (“bits” are key singularities in relation to information) but rather of enriching this analysis through the phenomenal richness of geometric structures and their effectiveness in information processing. Once again, formalism and mechanistic physicalism are seen to be not a variety of scientific reduction of the kind we should expect to meet in scientific practice, but rather a philosophical monomania, which has lost touch with the plurality of forms taken by our knowledge of and interaction with the world. Research on both morphogenesis and architecture of dynamical neural networks (see e.g. Hertz 1991; Amari & Nagaoka 2000), despite its incompleteness (arising from the limitations of a physicist’s, though neither formalist nor
Space, time and cognition
mechanist, standpoint), at the very least suggests the richness of the geometrical structures implicated in any account of the organisation of living systems. . Globality and circularity in space and time One of the great difficulties for a mathematical analysis of living systems and their evolution lies in the aspects of circular co-constitution. To the dialectic tension of the individual and the ecosystem we must add that between the present and the future. As already pointed out, and contrary to certain theories of mind, I assign a very elementary sense to the notion of intentionality. The intentionality of our knowledge and our will is the ultimate and non-compositional epiphenomenon, the journey’s end of an intentionality characteristic of all living beings. It is the sense of intentionality illustrated by an amoeba moving in one spatial direction or another to preserve or ameliorate its metabolism. This kind of motion implies the unity of a living system, an individual with its membrane, so crucial to isolate it as a biological unity and the essential condition of autopoiesis, Varela (1989). Intentionality suggests an analysis of time which incorporates the description of structural loops: the self-defining structuring of the ecosystem, where anticipations of possible future situations contribute to determining the present evolution and its bifurcations. Interactions within the ecosystem take place in 3-space and time: that is to say, the local times or internal rhythms of the individual organism as well as the time of its spatial interactions – the unity of an ecosystem relies on spatio-temporal cicles within it. Mathematical methods appropriate for some kinds of structural loops have already been proposed. Mathematical Logic, for example, has suggested impredicative definitions and non-well-founded sets, amongst others, see Barwise (1996), Longo (2000). An impredicatively defined set contains elements, parts, the definition of which depends on the set itself (the local depends on the global). In fact topology very often employs impredicative notions (an intersection of sets containing the sets to be defined, etc.). In a certain sense, impredicative definitions are “formally unstable”, in a manner reminiscent of the way a dynamical system is unstable: its global structure dynamically determines components which in their turn serve to constitute it. It is not clear whether these approaches can tell us anything about the unity of living systems, because that unity clearly goes a long way beyond the forms of circularity they capture. However they do provide conceptual hints, for the properties which can be expressed and the functions which can be computed
Francis Bailly and Giuseppe Longo
in such formal languages greatly outnumber those seen in their predicative or stratified versions, not to mention the gain in simplicity (see Girard et al. 1989). Their representation in the setting of category theory brings a richer structural framework in which closure properties play an important role. (See Asperti & Longo 1991, for an application of the notion of “internal category” to the semantics of impredicativity.) Non-stratified or impredicative systems have received relatively little attention in the study of the foundations of mathematics, because of the hegemony of foundational/mechanistic trends taking the stratification of world structure as the only reliable source of explanation. It is in this way that the predicativistic approach long neglected or discarded tools which, by contrast, bring us closer to a “mathematics of the real world” (from “complex” dynamical systems to systems forming organic unities). The obstacles posed to the development of such a mathematics are profound. In the first place the need of “expressive and constitutive circularity” of the kind already seen has to be enriched and put to better use. The reasons why biologists have resurrect “teleonomic” arguments may provide a clue. The notion of telos is always close to hand in the description of living systems (see the “contingent finality” above). The prevailing mechanistic outlook gets rid of these teleonomic traits. By contrast, the analysis of the mutual dependence between states and aims needs to be integrated into the mathematical framework within which we formulate the description of living systems.
Part 3. Spatio-temporal determinacy and biology . Biological aspects The question of space has played a very important, even a foundational role in biology: one which has not always received due appraisal. Take for example the concept of milieu intêrieur (internal environment) introduced by Claude Bernard, which allowed an essential topological separation between the interior and exterior of an organism. Consider also the question of chirality in biology, highlighted by Pasteur. In the wake of his experiments on the tartrates and the manner in which their biological activity differed depending on whether they coiled to the left or the right, he stated unhesitatingly: “Life, as it is manifested to us, is a function of the asymmetry of the universe and a consequence of it”.
Space, time and cognition
Indeed Pasteur anticipated both developments in his own field of scientific inquiry and, mutatis mutandis, what later came about in physics with the discovery of the asymmetry of matter and anti-matter, which cosmology now views as the precondition or the existence of the universe and of the actual material structures we see all around us. Biological structures are subject to organising processes leading to the emergence of complex forms, such as those studied in developmental biology; furthermore, they display physiological functions which sustain the part/whole mutual dependence which mediate their integration as organisms and regulate the linkages between the different levels of organisation typical of organic existence. These facts clearly have a connection with theories of the critical behaviour of dynamical systems, such as that seen in phase transitions. It was not by chance that the first mathematical models of biological systems appealed to and borrowed from those of thermodynamics, in particular models of cascade effects in bifurcations of thermodynamical systems (see Nicolis 1986; Nicolis et al. 1989), followed by models of emergence of selforganised critical behaviour (see Haken 1978; Kauffman 1993; Varela 1989), and application of fractal geometry (see Mandelbrot 1982; Bailly et al. 1989; Bouligand 1989) and chaotic regimes (Babloyanz et al. 1993; Auger et al. 1989; Demongeot et al. 1989) to an organic context. Alongside these developments, it had been clear that the character and genesis of processes of formation could in many cases be modelled using the elementary theory of catastrophes (Thom 1977) and, more generally, singularity theory. What clearly shows up in the analysis of selfregulation and homeostasis (but also in the analysis of pathology and death) is what may be termed the “extended criticality” i.e. the enduring sensitivity to critical parameters of systems in that situation – a situation which is limited in spatial and temporal extent, but which nonetheless is extended (see Bailly 1991). As a comparison, recall that in the framework of quantum theory, energy and time are conjugate variables. But an asymmetry nevertheless holds between them. While energy is a well-defined observable of the quantum system, associated with a Hamiltonian operator, time appears only as a parameter: one seemingly less essential and less well incorporated into the theory. In biology we seem to have the inverse: it is the time characteristic of biological systems (an iterative time which regulates biological clocks and internal rhythms) which seems to be the essential observable; whereas energy (the size or weight of an organism for example) appears simply as a parameter (an accidental parameter at that). In this sense one might even say that biology, relative to the energy/time conjugacy is quasi-dual to quantum mechanics.
Francis Bailly and Giuseppe Longo
. Space: Laws of scaling and of critical behaviour. The geometry of biological functions Since the pioneering work of D’Arcy Thompson (D’Arcy Thompson 1961), recent studies (Peters 1983; Schmidt-Nielsen 1984; West et al. 1997) have shown that numerous macroscopic biological characteristics (as distinct from genetic traits at the biomolecular level) are expressed at the same scale across the range of entire species, indeed across genera, taxa and in some cases the entire animal kingdom. This scale-invariant parameter picks out the organism by its mass W or in some cases by its volume V. Furthermore, characteristic time scales for organisms (lifetimes, gestation periods, heart rates and respiration) all seem to obey a scaling law. They are typically in a ratio of one fourth of the mass (T ~ W1/4 ). Just as these frequencies scale as –1:4 of W, metabolic rates typically scale in a ratio of 3:4 of W and many other properties display similar scaling. Such scaling laws call to mind the behaviour of dynamical systems where critical transition in regime is associated with fractional exponents of some key parameters. What differentiates one group of organisms from another is simply the value of the ratios seen in the expression of these scaling relations. These remain the same across numerous species and even across much wider biological groupings. Perhaps the most spectacular example is that of lifespan, which is in the same ratio to body mass. Other kinds of scaling laws – allometries – link geometric properties of organs (as distinct from organisms) across numerous species, or within a single organism at different stages of its development. Here, however, our principal point is bound up with the display of fractal geometry in certain organs, engaged directly in the maintenance of physiological functions, such as respiration, circulation and digestion. This fractal geometry appears to be the objective trace of a change in the level of “organisation” and of top-down regulation of the parts by the whole, in conjunction with the bottom-up integration of the parts within the whole. The fractal geometry in question falls into two distinct kinds. On the one hand the examples seen in the interfacing membranes of the organism. The metric dimensions of these are between 2 and 3. Examples are the membranes of the lungs, the brain and the intestines. The other class is formed by branching networks, such as the bronchial tubes, or the vascular and nervous systems. Here the metric dimensions of the extremities may be greater than 2. These fractal geometries permit the reconciliation of opposed constraints associated with spatial properties. On the one hand, because the organs involved are engaged in the regulation of exchanges such as respiratory or cardiac function,
Space, time and cognition
their effectiveness and their corresponding size must be maximised in order to support and to fine-tune these exchanges; on the other hand the fact they are incorporated into an organism containing many parts means that their bulk must be minimised to ensure their overall viability. To the extent that organs are clearly individuated and alloted wholly to certain specialised functions, they must present a certain homogeneity throughout their spatial extent. These various constraints are clearly antagonistic and only the fractal character of the geometry of the organs in question allows them to be reconciled. Another aspect which raises interesting questions is the intrinsic threedimensionality of living systems. If one enquires into the abstract possibility of developing biology in dimensions other than three, one recognises that the choice of three dimensions again allows the reconciliation of antagonistic constraints. On the one hand it is required that the organism present sufficient local differentiation to permit different concurrent functions across its whole structure; on the other hand it needs to be the site of sufficient internal connectivity to co-ordinate the activity of all its parts. In a space of only two dimensions, if the differentiation were sufficient, the connectivity and coordination between the different parts could not be established because the required connections of the components would intersect so greatly as to disrupt their separated functioning. On the other hand, in a space of four dimensions, the degree of possible connectivity is clearly greatly enhanced, but it is known that in four dimensions mean field theories11 become applicable and the constraints of local differentiation become insufficient to allow for the establishment of systems stratified into different levels of organisation. Development of the system in three dimensions serves to reconcile these constraints at the cost of producing fractal geometries (their emergence reflects the existence of certain dynamical attractors). All these considerations concern the internal space of biological systems – they have no bearing on the dimensionality or topology of external space. Following on from the earlier presentation of the notion of space in physics in terms of fibrations (carrying the internal symmetries of the system) over a base-space (the external space-time) one might consider the existence, in similar terms, of internal “spaces” associated with different levels of biological organisation. These should be distinguished from the different levels of scale structure in physical systems. What is distinctive in the biological context is the way these levels are connected with the regulation of the lower by the higher level of organisation and with the manner in which the different levels of structure act in constraining the formation of the integrated wholes which together they constitute.
Francis Bailly and Giuseppe Longo
. Three types of time To a first approximation, in describing the actual state of an organism (or a population) one can consider two types of temporality, jointly implicated in its survival. The first type, which carries echoes of time as seen in classical physics, is associated with the stimulus-answer coupling between an organism and its environment. It is manifested chiefly by relaxation processes (in the quasi-canonical form e–t/τ and exponential combinations thereof).12 The second (of a very different nature) is associated with internal clocks which administer the biorhythms of a living system and ensure its continued functioning (Glass et al. 1988; Reneberg 1989). It takes the form eiwt and its combinations. But the most important aspect of biological time is perhaps irreducible to these distinct forms: the internal “temporality” of organisms is iterative rather than historical. The measure of duration in this internal time is no longer a dimensional magnitude, as in physics, but rather a pure number registering the iterations already effected and those still remaining for an organism which experiences a finite number of these within a range fixed in advance, depending on its class. Thus, all the mammals, from mice to elephants or whales, form one such class. This is characterised by the number of heartbeats per average lifetime (around 109 for mammals) or the number of corresponding breaths (around 2.5 ×108 ). The variation in these frequencies between species is traceable to a single parameter – the body mass of the average adult. This striking trait is directly connected to the scaling law mentioned earlier.13 The importance of this aspect of biological time is emphasised by recent attempts to re-think the principal features of evolutionary theory in terms of the “living clocks” approach (see Chaline 1999). By interpreting evolutionary transformations in terms of their synchronic and diachronic effects, this approach concerns both the developmental level of individual organisms and the evolution of species. But as G. Longo has proposed in Section 2, it seems that in biology it is necessary to take into account a third type of temporality, connected with what he terms “contingent finality”. This expression means a degree of (nonreflexive) intentionality which may help to explain the evolutionary and adaptive aspects of living systems. It is an anticipatory form of temporality, linking the current state of the organism and the future state of its environment, to which the organism contributes by its behaviour; and it is a form specific to biology, termed as “teleonomic” by Monod, which arises in connection with a coupling between the rhythms registered by inner biological clocks of the sys-
Space, time and cognition
tem and the stimuli-responses the system undergoes while interacting with its environment. (This coupling may introduce delay effects.) Unlike the two previous dimensions, this “third dimension” corresponds to aspects of biological systems not seen in physics, in that it concerns the variety of integration-oriented factors involved in determining the present state of the system with those involved in determining its future state. Perhaps this is another reason why, as G. Longo has underlined, one cannot define the notion of a trajectory traced out in the course of the evolution of a biological system in the manner one does for phase spaces in physics. There is no scope in the biological case for the application of a geodesic principle which extracts and determines one trajectory (that obeying an extremal principle) from amongst all the virtual possibilities. It seems the logic of biological systems operates in a quite different fashion, in a manner designed to display a Bauplan selected by external criteria. As Gould claimed in his account of the organisms found in the Burgess Shale where it seems all the virtual possibilities – every possible pattern of development – saw the light of day. Here it seems we are dealing with criteria operating not so as to secure the emergence of a single possible form of the system (as with minimum principles in physics) but rather so as to secure the elimination of impossible forms so as to produce a maximal variety of system structure within a limiting “envelope of possibility”.14 Without seeking to formulate premature conclusions, we can nevertheless draw some lessons for our understanding of our notions of space and time, in the light of recent developments in theoretical biology. The distinction between internal and external space is connected with the distinction between the autonomy of an organism (the homeostatic stabilisation of its functioning and its identity) and its heteronomy (its dependence on and adaptation to its environment), see Bailly (1998). Equally, the internal/external articulation of a space physically determined in structure and another space, determined by the functionalities of the organism and its complex morphology, leads directly to issues of the relationship between the genetic programming of an organism and the epigenetic factors involving its interaction with its environment in the course of its typical development. The essential new element distinguishing the situation in biology from that in physics lies in the fact that in biology the articulation of this internal/external distinction applies also to time and in a crucial way, involving the relationship between the different types of temporality: one of them akin to that in physics and with the character of a dimension, the other specifically biological, iterative and expressed in pure numbers, which seems to play
Francis Bailly and Giuseppe Longo
a quasi-constitutive role with regard to our concept of a biological system, in that it supplies the basis for the characterisation of the invariants operating in the definition of equivalence classes in the biological setting (the invariants of mammalian biorhythms illustrate this well). We could go a little further in tracing this comparison with physics, and seek to locate the distinguishing feature of biological systems at an even more fundamental level, that of the dimensionality of its elements (in a topological sense). With the rise of string theory, ultimate entities have gone from being punctate to being linear. In biology an inverse, but curiously convergent, development occurs. It seems that what is regarded as lying at the most fundamental level of the organisation has undergone a change from being something which occupies a volume to being something linear. What appears fundamental to the genetic programming and the resultant life forms, is held to be a linear sequence of macromolecules, aligned in an order determined by the base pairs of DNA which constitute the genetic endowment and to a large degree govern the development and functioning of the organism. The biological activity of these macromolecules depends very strongly on their 3D spatial structure (as is shown by the activity of prions) but it is nonetheless remarkable that the linear chains of macromolecules have come to assume such importance, and that the spatial structure of their enfolding appears to be so largely dependent on their linear sequences, which effectively control the interactions regulating that enfolding. The manner in which the concepts of space and time are treated is, once more, fundamental to the constitution of biology as a science. Beyond the analysis of perception, any epistemological account of the status of such concepts – to the extent it is based on the results of natural science and aims to be objective – cannot ignore their role in the framework of theoretical biology. . Epistemological and mathematical aspects Now, it may be instructive to run over the epistemological ramifications of space and time by means of an analysis “transversal” to the theoretical frames of reference considered so far. Three pairs of concepts appear important for such an analysis. Firstly, local vs global concepts in relation to space; second, iterative vs processual aspects in the understanding of time (with an aside examining how both these pairs of concepts are intimately bound up with the topic of causality). Third, regular vs singular in connection with our system of representation and reference.
Space, time and cognition
In fact, by considering physical and biological aspects of space and time, one cannot evade the epistemological issues involved in their purely formal definition. Spatial and temporal concepts are “abstract” concepts in two different senses of that term. The first relates to the process of abstraction which takes its cue from common features of the theoretical treatment of space and time and the second, concomitant, sense relates to their being formal, quasi-a priori notions which come to be imposed as the result of that process of abstraction, as an intrinsic component of our notion of objectivity itself.
I. LOCAL vs GLOBAL In passing from relativistic to quantum theories and then to general dynamical systems theory, and finally to biology, we recognise a shift in the relative importance and pertinence of the local/global opposition (broadly speaking, a shift from the former to the latter). Despite the stress on a global interactive point of view seen in Mach’s Principle, general relativity completely preserves the principle of locality inasmuch as it is essentially and exhaustively expressed through partial differential equations. In this respect, it lends itself to an interpretation in terms of local causes propagating within the light cone. Quantum physics can equally well be presented as a theory of local interactions and their propagation by state vectors. The Schrödinger Equation and that of Dirac are just as much partial differential equations as those of Einstein. Hence it could be taken to involve a notion of causality of the same apparent kind. But quantum measurements on the one hand, and non-separability on the other stand in the way of a completely local interpretation of the theory. Classical causality is affected by the fact that measurement leads to intrinsically probabilistic results while non-separability disrupts any purely local representation of the propagation of effects.15 The case of theories of the critical states and dynamical systems takes us a step further: here non-locality plays a twofold role. Firstly the fact that interactions can now take place at long distance leads to correlations becoming infinite. Local variations and effects lose their relevance both for analysis and measurement in favour of the global behaviour of the system. This even reaches the point that our notion of what counts as an object needs to be redefined. Furthermore this global behaviour is itself governed by critical exponents and scaling laws which are in no sense local (since they are dependent on the dimensionality of the embedding space and on an order parameter). A concomitant of this situation is that the usual notion of causality (even when there exists a linear correlation between cause and effect – small causes giving rise to small effects) is undermined (in critically sensitive systems, in-
Francis Bailly and Giuseppe Longo
finite effects can arise from finite causes which lead to discontinuities in the evolution). Curie’s Principle (that symmetry of causes is mirrored in symmetry of effects) is thus called in question (at least for systems displaying singularities and discontinuities in their behaviour) by the symmetry breaking which accompanies phase transitions. In biology, locality seems pertinent chiefly to the description of underlying physico-chemical processes, while the definition of biological systems and their manner of functioning involves global concepts associated with the fundamental non-separability of living systems and their complexity. This global level of structural organisation becomes decisive for the representation of processes of regulation and integration which stabilise the functioning of a biological system. To this there corresponds a more complex notion of causality involving an entangled and interactive hierarchy and its associated ‘agonistic or antagonistic’ effects. In brief, the notion of local causality cannot be called in question without the global notion being affected as well (and the global notion is associated with “contingent finality”). This opens the way to a distinction – one meaningless in physics – between the normal and the pathological. A locally pathological mode of functioning can co-exist with the preservation and global functioning of an organism.
II. ITERATIVE vs PROCESSUAL Relativistic theories, with their characteristic metric structure, display an almost completely spatialised type of temporality. They introduce the concept of an event as a “marker flag” in a generalised space. Only physical causality – the fact that interactions between point-events cannot propagate outside the light cone, or to reformulate that requirement from a mathematical standpoint, the fact that the signature of the metric is fixed – serves to introduce a distinction between spatial and temporal dimensions. Conceptually, this distinction is intimately bound up with the fact that from the viewpoint of the symmetries of the system, Noether’s theorem classifies time as a conjugate variable with respect to energy (or conversely, views energy as the conjugate of time), just as the spatial variables are conjugate to the components of the momentum. But essentially, relativity, via the group of general covariant transformations, treats time as a notion of the same kind as space. Contrasting with this situation, in quantum theory time is treated as a simple parameter. Its status as the conjugate of energy is preserved (as is seen in the Heisenberg indeterminacy relations) but it does not appear as an observable of the theory. Moreover it seems that certain phenomena – those connected with quantum state transitions or even measurement (setting to one side the deco-
Space, time and cognition
herence approach) – do not easily lend themselves to an interpretation in terms of temporal concepts of the kind connected with our experience of passage and duration. One sees something similar in the apparently instantaneous connections associated with the behaviour of non-separable quantum systems. This further stresses how greatly the conceptualisation of causality is bound up with that of time. In theories of dynamical systems, time regains more classical characteristics, but ones made apparent in connection with different aspects of phenomena than in the classical case. Besides being the time of events (as state transitions), it plays further roles, distinct from the role it plays as a parameter: in the definition of stability, in the definition of irreversibility, in the characterisation of attractors (asymptotic behaviour, as seen in fractal geometries, the Lyapunov exponents, etc); and when a bifurcation takes place, it can establish a cycle and thus acquire an iterative character. In biology, temporality displays two quite distinct aspects: the external time, i.e., the relaxation time of stimulus and response, of functional adaptation to an exterior environment, and the iterative time of pure numbers associated with internal biorhythms involved in the regulation of physiological functions. The corresponding notion of causality, adaptive and intentional in character, seems closely connected with the mutual articulation of these two aspects.
III. REGULAR vs SINGULAR Relativistic space-time is “regular”, continuous and differentiable, and singularities (whether of the Schwarzschild or the initial manifold) play a quasiincidental role which assumes central importance only in certain astrophysical and cosmological contexts. The situation is different in quantum physics, where the regularity of certain spaces is associated with the discretisation of others and where space-time structures can be envisaged as fractal at very small scales. The regular/singular couple carries the traces of the old debate about the interpretations of the theory, namely in terms of fields or in terms of particles. In theories of “critical” systems, the interest in singularities is accentuated. They are associated with increase in complexity, and also with the particular consequences of nonlinear dynamics (e.g., for what concerns solitons and their propagation). In fact, these theories are essentially singular since critical situations all involve singularities (divergence, discontinuity, bifurcations). The mathematics of singularities (singular measures, catastrophes) plays a predominant role in modelling the behaviour which gives rise to complexity. Neverthe-
Francis Bailly and Giuseppe Longo
less, it appears that the outcome of these features is in fact a new form of regularity, one located at a more general level of analysis, revealed in laws of scaling and leading to a universal classification embracing very different systems, which nonetheless manifest identical behaviour with respect to the singularities in their dynamical evolution. The critical transitions in these systems are typically restricted to a very narrow range, even to a single point in phase space, a single value of the control parameter, and on either side of this very narrow critical zone regular behaviour of the system once again becomes dominant. Precisely this last aspect seems to contrast with the position prevailing in biology. Organisms and ecosystems can survive and maintain themselves within a range of values of critical parameters – an extended zone of criticality. Exit from this zone implies the death of the organism: its underlying physico-chemical structure can no longer sustain biological functions. Any biological system behaves in a manner characterised by a dense distribution of critical points in the space of control parameters, and not a discrete or isolated one. Homeostasis then, corresponds to a sort of structural stability of the trajectories, relative to the attractor-basins of the dynamics. Space and time, especially as they feature in the framework of modern physics, are neither objects nor categories. To recall Kant’s formulation (Kant 1986) they are “a priori forms of sensible intuition”, and as such the preconditions of any possible experience. In the light of the most deep-going analysis our current physical theories allow us to make of them, they seem to reflect the mathematical structure of a group and a semi-group, respectively (see Bailly 1999). Indeed, the mathematical properties postulated for space, inasmuch as it is the medium and support of displacements in general, necessarily connect with and exemplify the group structure. Given the tight connection between the group structure and its associated equivalence relations, an abstract, epistemically basic frame of reference emerges which acts as a kind of pole of attraction for any representation of objectivity, namely the frame: <space, group structure, equivalence relation>. Similarly in respect to time, the property of possessing an orientation – “time’s arrow” as the index of change – is reflected in the abstract structure of a semi-group. That structure can be put in correspondence with an order relation. This leads us to the recognition of a second epistemically basic frame of reference: . To repeat, for the avoidance of all confusion: the space and time which feature in these frames are not so much an aspect of entities with an intrinsic nature, but rather of the conceptual grid presupposed by any natural science: they are conditions of possibility rather than of concrete actuality. If this ap-
Space, time and cognition
proach is correct, these two “poles” leave their mark in our notions of permanence and change, stability and evolution, identity and differentiation. They delimit the “field” of preconditions for any natural science, to the extent that the phenomena studied therein are manifested in a spatio-temporal setting. . Closing remarks This Tableau Général is still incomplete (notably in the life sciences). But what clearly stands out is that contemporary theories across a whole range of scientific domains involve conceptions of space and time that have not yet been fully stabilised or clarified (although in physics super-symmetric string theories are seen as leading to unification). Will we continue to distinguish space radically from time despite the merging of their status in the setting of relativity theory? Shall we continue to refer to a single notion of space in view of the variety of topological and other structural properties (compactification, noncommutativity, internalisation, fractal dimension) envisaged for it in current physical theory? Or in the face of the complexification and fractalisation of the forms of space arising in biology? Likewise shall we retain the representation of time as a unique parameter in physics or as intrinsically irreversible? And how will the time of physics turn out to be connected with that of biology? By jettisoning its formalisation in terms of isolated bifurcations perhaps, in order to take into account the synchronic and diachronic effects in biological systems? One reason for this rather confused state of affairs (albeit one closer to the often counter-intuitive nature of reality than our spontaneous perceptions can bring us) is that the very epistemological status of space and time remains relatively problematic despite the formal categorisation introduced in the foregoing. One additional difficulty has been introduced by the fact that since recent developments in physics (mirroring what has long been the situation in biology) we now have to take account of spaces external and internal to the systems under investigation. We here encounter a distinction long made by philosophers – and notably by Kant – but this time in a form lying on the side of the objects themselves, whereas for Kant it was conceived as lying on the side of the epistemic subject. It is the distinction between space as the form of external and time as the form of internal sense (Kant 1986). Such a distinction with respect to objects was unacceptable to the Kant of the Critique, for having broken with an ontological characterisation of the objects of natural sciences, any such internality was denied and location was considered something totally external. However,
Francis Bailly and Giuseppe Longo
towards the end of his life, in re-examining Newton’s Principia, Kant could not refrain, on the evidence of the Opus Postumum (Kant 1986a) from thinking through this point afresh, in particular in relation to the question of energy. He renewed his investigation of certain aspects of Leibniz’ thought on this subject which he had largely avoided beforehand and he would perhaps have found in the latter-day evidence for a spatialisation (and temporalisation?) internal to the objects of physics and biology, if not the answer to his puzzle, at least a spur to new investigations. Any attempt to make the internal/external dichotomy correspond in a straightforward way to the distinction between space and time is infected with artificiality – the more so if the distinction is seen as lying within reality itself rather than the knowing subject. Nevertheless, taking into account the Kantian view of space and time as the very conditions for the constitution of objectivity, it does not seem too extreme to speak of them, rather than of subjective “forms of sensible intuition”, of objective “forms of sensible manifestation”. Such forms could themselves be connected with the concepts of externality and internality. In the case of “the external”, one essentially considers the phenomenal manifestation of relations between objects (interactions and corresponding measurements). In the case of “the internal” one considers, rather, constraints concurrent with the phenomenal manifestation of the continuing identity of an object – or better, of its identification. Notice that the distinction between internal and external aspects is one of the conceptually distinguishing features of biology. But it can be expected to be an element in our conception of the objects of theoretical physics as well, since it now has an objective, mathematically expressed counterpart in the distinction between the external space-time (the base space) and internal spaces (the fibres over the base) which is now a fully developed aspect of the mathematical formalism of key areas of physics. Briefly (and ironically, in the light of certain epistemological tendencies which have sought the reduction of biology to physics), it is on the side of biology that one now looks for the conceptual clarification allowing the development of a more comprehensive abstract framework for the understanding of physical phenomena (see Rosen 1991). We have thus arrived at a kind of conceptual “re-normalisation”. The external space-time of physics can be seen as a manifestation of the couple <space/relation> and as providing a ‘base’ relative to which the corresponding couple
