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!
>, i.e. I believe P. In the stereotypical cases of belief, e.g. not selfdeception, repressed belief, or forgetfulness (Shoemaker 1996: 80), because it is believed, the content <
> is ‘‘available’’—I would say ‘accessible’—as ²⁶ So beliefs for the non-megalomaniacs among us are ‘‘incomplete’’, in a sense analogous to the incompleteness of a formally axiomatized system rich enough to contain formalized arithmetic. In earlier work I explored the parallel between the report of the existence of a true but formally unprovable G¨odel sentence and a Moore paradox sentence like (20); see Jay David Atlas, ‘‘What is Paradoxical about G. E. Moore’s Paradox?—Rationality, Sincerity, Implicature, and the Selflimitations of Belief ’’, November 1995, MS, Pomona College, Claremont, California. Needless to say, I do not think that a G¨odel sentence is assertorically anomalous, nor do I think that the sentence that reports its existence is asserted anomalously. The relationship between a Moore sentence and the report of the existence of a G¨odel sentence is of great interest, logically and linguistically.
Jay David Atlas
139
a premise in my theoretical and practical reasoning. When so, I consciously think it and, furthermore, ‘‘mentally assent’’ to it. (Mentally assenting is like verbally asserting it or answering a query about its truth by replying ‘yes’—only quieter.) But it is analytic (for Shoemaker) that mentally assenting to <
> is a conscious process; Shoemaker (1996: 79) actually writes ‘necessary’ instead of ‘analytic’, but I take the liberty. When doing so, one ‘‘manifests’’—we language mavens would have said ‘expresses’, though that is not without its problems—that one believes P. This is a causal model: one asserts P because one believes P (and < P> is being entertained), just as one has spots because one has measles. Having spots ‘‘expresses’’/‘‘manifests’’ measles (Sellars 1969/1996).) The next step in Shoemaker’s argument is one for which I can find no argument in his essay. He wants to go from ‘‘manifesting’’ that one believes P to ‘‘manifestly assenting’’ to one’s believing P (Shoemaker 1996: 82–3). For if this inferential move were acceptable, then one would be assenting to one’s believing P. The argument would now show that if one assents to < P> , one assents to one’s believing P. Since assenting is conscious (an analytic claim for Shoemaker), one is aware of assenting to believing P. Since assent is a kind of believing—a stipulation (Shoemaker 1996: 78), one is consciously believing that one believes P. By this argument, Shoemaker gets two large philosophical payoffs. First, he gets the claim that if one assents to < P> , one assents to < I believe P> . Second, he gets the claim that we have first-person access to our beliefs: (a) if one believes P, one will be able to access the content <
>; (b) if one is able to access the belief-content <
>, one will both believe (at least tacitly) that one is able to access the belief-content <
> and consciously believe that one is able to access the belief-content <
>. The question is how, in the crucial step, Shoemaker goes from manifesting that one believes P to ‘‘manifesting assent’’ that one believes P. There is a hint in the text (Shoemaker 1996: 82). According to Shoemaker, saying (sincerely) < I believe P> can manifest that one believes P. If one sincerely asserts < I believe P> , and one sincerely asserts it because (and only because?) one mentally assents to it, then one believes that one believes P, because it is analytic for Shoemaker that one mentally assents to believing P only if one believes that one believes P. And the deed is done; if one manifests believing P, one manifestly assents to believing P. So if one believes P, one believes that one believes P—privileged access (Shoemaker’s ‘‘self-intimation’’). The crux of the matter is Shoemaker’s claim that asserting (sincerely) < I believe > P manifests that one believes P. If ‘manifest’ is paraphrasable by (one sense of ) ‘express’, one has the claim < asserting (sincerely) < I believe P> expresses that one believes P> . But this claim is only plausible if ‘express’ is not taken in its causal, manifesting sense, e.g. the sense in which one’s asserting P expresses one’s believing P. It is plausible only in the semantic sense of ‘express’, that in which
140
What Reflexive Pronouns Tell Us
my assertion < I believe P> expresses the thought that I believe P. (See Sellars 1969/1996.)²⁷ In short, the apparent success of Shoemaker’s argument relies essentially on an equivocation between two senses of ‘express’, the causal (manifesting) sense and the semantic sense. Had Shoemaker’s argument succeeded, he would have had the following explanation of the oddity of the Moore sentence schema < P ∧ ¬ BI [P]> . On Shoemaker’s view we may derive from < BI (P ∧ ¬ BI [P])> , on the assumption that believing a conjunction entails believing each conjunct, < BI [P] ∧ BI [¬ BI [P]]> . But if it is a theorem of the logic of belief that if I believe P, then I believe that I believe P, one then infers < BI [BI [P]] ∧ BI [¬ BI [P]]> . By the converse conjunction principle to the previous one, one may infer < BI [BI [P] ∧ ¬ BI [P]]> . So I, the believer of the Moore sentence ‘It’s raining but I don’t believe it’, if my believing P entails my believing that I believe P, and if belief distributes over conjunction, believe an overt contradiction. What is essential to this explanation of the oddity of the original form of a Moore’s Paradox sentence by appeal to the existence of a contradictory beliefcontent is Shoemaker’s elaborate argument in support of the view that if I believe P, then I believe that I believe P. But that is just the argument that fails; it seems to succeed only because of an equivocation, a slide between two senses that motivates an illicit move from ‘‘manifesting’’ that I believe P to ‘‘manifestly assenting’’ to < I believe P> .²⁸ Shoemaker has assumed that by deducing a contradictory belief-content from believing the Moore sentence, he has shown why the uttering of the sentence assertively is ‘‘odd’’. This is a traditional line to take. To my knowledge it was first employed by the American logician C. H. Langford in 1942, in his contribution to the Schilpp volume on Moore ‘‘The Notion of Analysis in Moore’s Philosophy’’ (Langford 1942/1968: 333), where he writes: I want to cite an example which is due to A.M. MacIver and which is worth repeating on its own account.∗ Suppose someone to remark: ‘‘He thinks that he has been to Grantchester but he has not.’’ The person referred to may entertain this proposition as an hypothesis. But suppose he actually asserts the proposition: ‘‘I think that I have been to Grantchester but I have not.’’ This sounds self-contradictory, and the reason is that he will actually be saying that he thinks that he has been to Grantchester, whereas the but-clause in the indicative mood will signify or mean pragmatically# that he does not think so.∗∗ *See Analysis, Vol. 5 (1937–8), 43–50, and The Journal of Symbolic Logic, Vol. 3 (1938), 158. ²⁷ ‘Assertion’ has the traditional act/object ambiguity in English: assertion as an act of asserting, assertion as what is asserted. ²⁸ Hintikka’s (1962) seminal discussion of the logic of belief takes the Moore sentence to be ‘‘doxastically indefensible,’’ i.e. that one’s believing one’s Moore sentence is inconsistent with Hintikka’s modal S4 axioms for the belief-operator. Of course, the characteristic axiom for S4 is, in the case of the belief operator, just < Bp → BBp> .
Jay David Atlas
141
#(My note: Langford (1942/1968: 332) wrote: We may call what a man does not state, but intends his linguistic behavior to signify, his pragmatical meaning, and we may distinguish this from the sense of his words, which is the proposition expressed by them.$ [Langford’s note:$ This term has been used by Charles Morris in the same or a similar sense.]) **The course of Moore’s argument in ‘‘A Defence of Common Sense’’ will be clearer if this distinction between formal and pragmatical contradiction is carefully observed. For in saying that certain philosophers contradict themselves when they assert, in effect, ‘‘There have been many other human beings beside myself, and none of them, including myself, has ever known of the existence of other human beings,’’ Moore is not holding that a formal contradiction can be derived from the sense of those words, but only that the pragmatical meaning of such an assertion is incompatible with its literal meaning.
The peculiarity of Langford’s analysis is that he is diagnosing the ‘‘apparent self-contradiction’’ of a statement of the form < BI [P]&¬P> by the hypothesis that in saying < ¬P> the speaker ‘‘pragmatically means’’ < ¬B[P]> . There is no recognition from Langford of the need to discuss an apparent scope distinction between the wide-scope negative belief-proposition that Langford takes to be the speaker’s pragmatical meaning and the narrow-scope negative proposition B[¬P], which might be an equally good candidate for the pragmatical meaning of uttering < ¬P> . Without defense—and I find the claim dubious—Langford chooses the wide-scope negative formula as the speaker’s meaning in order to explain the apparent self-contradiction by an actual logical contradiction between the sense of ‘I believe I have been to Grantchester’ and the speaker’s alleged pragmatical meaning ‘‘I don’t believe I have’’ when asserting ‘I have not’.²⁹ In addition he thinks it is self-evident that such a logical inconsistency between these different sorts of ‘‘meaning’’ suffices to explain the apparent ‘‘oddness’’ of asserting the Moore sentence. But no theory providing the explanatory connection is offered at all. In a similar vein, George Lakoff (1975: 264–5) discusses Moore’s paradox in the form < P&B[¬P]> . Lakoff ’s explanation of the paradoxicality of the utterance is a version of a solution that appeals essentially to a speech act sincerity condition, to an incorrigibility thesis, and to rationality assumptions. In addition to the sincerity condition ‘‘if x sincerely asserts P, x believes P’’, Lakoff makes three assumptions: (i) that if one believes a conjunctive statement, one believes ²⁹ There is an appealing, but a mistaken, argument to support Langford’s reasoning. Suppose a` la Moore in asserting < ¬P> the speaker pragmatically means < BI [¬P]> . The consistency of belief requires the axiom < B[¬P] → ¬B[¬¬P]> , while Belief Double Negation yields < B[P] → B[¬¬P]> , from which there follows < B[¬P] → ¬B[P]> . If pragmatic meaning is preserved under logical consequence (modulo the axioms of rational belief ), then if the speaker pragmatically means < B[¬P]> , he also means < ¬B[P]> . Such consequences might be thought to be plausible if a speaker intends his utterings to signify the logical consequences—or perhaps more restrictedly the direct logical consequences (Atlas 1991: 137), modulo the axioms of rational belief—of what he intends his utterings to signify. The problem is, the preservation of pragmatic meaning under logical consequence is really quite implausible. So the easy argument for Langford’s position does not succeed.
142
What Reflexive Pronouns Tell Us
the individual conjuncts (belief conjunction-elimination < B[P&Q] → (B[P] &B[Q]) > ), an assumption also made by Shoemaker in his first argument, (ii) the incorrigibility of beliefs about mental states: that if one believes that one believes P, then one believes P (i.e. < BB[P] → B[P]> ), and the consistency assumption (iii) < B[P] → ¬B[¬P]> . He then can deduce the contradiction < B[¬P]&¬B[¬P]> from the assumption of a sincere, rational assertion of his Moore sentence. By explaining < P &B[ ¬P]> rather than Moore’s form < P & ¬ B[P]> , Lakoff avoids assuming Shoemaker’s self-intimation of belief (i.e. < B[P] → BB[P]> , the converse of the incorrigibility thesis). Lakoff (1975: 265) goes on to claim that the same principles will allow a deduction of contradiction from the assumption of a sincere assertion of the original Moore sentence form < P & ¬ B[P]> in order to explain the latter’s assertoric oddness, but so far as I can see this claim is incorrect. In order to deduce a contradiction, he needs the S4-like, self-intimation principle that, anti-Cartesianly, Lakoff thinks is questionable, though he does not explain why he thinks so (Lakoff 1975: 264). Langford and Lakoff explain the peculiarity of the Moore assertion by deductions of a logical contradiction of two sorts. In Langford’s case, precisely because the original assertion sounds self-contradictory to him, he claims that an inconsistency can be derived from the conjunction of the speaker’s unasserted pragmatical meaning and the asserted literal sense. In Lakoff’s case, because the assertion of the Moore sentence sounds odd to him, he claims that a sincere, rational assertion of the Moore sentence is logically inconsistent with the principles of rational belief and of speech-act theory. Yet it seems to me an open question whether the assertion of the Moore sentence does sound self-contradictory, and a closed question that even if asserting it seemed odd, the alleged reductio of the rational, sincere assertion of the Moore sentence could not explain the assertoric oddness. A difficulty with Langford’s explanation is that there are a denumerable number of self-contradictory statements that are perfectly felicitous, so that the logical inconsistency of the total signification of a statement cannot be sufficient for its assertoric oddness.³⁰ Half of the needed contradiction is not even asserted. In the case of Lakoff’s explanation, since neither a speaker’s insincerity nor the failure of the speaker’s beliefs to conform to Lakoff’s two axioms of rational belief and the incorrigibility thesis could be explanatory of the Moore statement being linguistically odd, it is bizarre to think that the impossibility of the rational, sincere assertion of a Moore sentence could explain the oddity of uttering a Moore sentence assertively ³⁰ For example, ‘0 = 1’, or even better ‘147/356 = 0.41292’, though contradictory, √ is not uttered especially not in a reductio ad absurdum proof, nor is the sentence ‘ 13 π = ∞ infelicitously, e−x2 dx’, an equation whose right-hand side is a definite Riemann integral, which is also −∞ a contradictory sentence but felicitously utterable, nor is the following, necessary falsehood, a non-theorem of the calculus, believed by European mathematicians for two hundred years to be a theorem and asserted as such, ‘Every real-valued function f: R → R that is continuous on the closed interval [a, b] is differentiable on the interval.’
Jay David Atlas
143
(as Moore (1993: 207) would put it), even if the impossibility of a sincere, ‘‘rational’’ assertion explains why either the utterance is insincere, or expresses a ‘‘non-rational’’ belief or a corrigible belief, or is not an assertion—oddities all, but none of insincerity, non-rationality, corrigibility, and an assertoric speech-act ‘‘misfire’’ is the oddity of uttering the Moore sentence assertively. None of those consequences would seem to explain the particular oddity of uttering the Moore sentence assertively. (As Paul Benacerraf once remarked, in reaction to this claim, an adequate account of the oddity must appeal to principles that every speaker/hearer can be expected to ‘‘know’’, in the sense in which we know our language and its uses. Not every speaker, surely, is natively a Port Royal logician.) The oddity remains even if the Moore utterance merely purports to be an assertion. Furthermore, ‘‘nonrationality’’ and insincerity of belief are properties of the asserter not of the utterance, so it is a category error to think that they can explain the linguistic oddity of the sentence uttered assertively. Arguments that appeal to self-contradiction or inconsistency with supposed rational principles of belief just do not explain the linguistic phenomena. What would explain it is a tale for another occasion. But if Lakoff finds privileged access dubious, Shoemaker and Hintikka find it self-evidently or necessarily true. I have already shown that Shoemaker’s arguments fail to support the privileged access thesis. Can the approach that I have been taking in this paper offer a definitive answer to the question whether the privileged access thesis is correct? V I I I . A LO G I C O - L I N G U I S T I C A RG U M E N T AG A I N S T P R I V I L E G E D AC C E S S ( “S E L F - I N T I M AT I O N ”) Why, apart from Cartesian prejudices about the transparency of the mind, should one have expected the privileged access thesis < BI [P] → BI [BI [P]]> to be correct? (25) If I believe it’s raining, then I believe that I believe it’s raining. The sentence in (25) has only two interpretations, the de se in the consequent of the conditional, in (26a), the de re in the consequent of the conditional, in (26b), and the de dicto in the antecedents. (26) a. I believe it’s raining ⇒ I believe that I myself believe it’s raining. b. I believe it’s raining ⇒ I believe myself to believe it’s raining. Neither interpretation makes the conditional true, either as an analytic entailment or a theorem of a logic of belief. The de re interpretation goes from the hypothesized truth of a non-first-person-content de dicto antecedent to an objectively first-person content, de re consequent. The de se interpretation goes from a non-first-person content de dicto antecedent to an essentially-first-personcontent de se consequent. Neither implication is semantically justified.³¹ The ³¹ As I mentioned in n. 12, those who cannot get the de re interpretation of the consequent in (26b) will at least get the de se interpretation in (26a), which suffices for my argument.
144
What Reflexive Pronouns Tell Us
English grammar of ‘believes’ sentences carries no commitment to first-person beliefs being ‘‘self-intimating’’ in Shoemaker’s sense and so to no distinctive first-person access to our beliefs. In fact, the falsity of (26) shows that the ordinary concept of belief, the one we express in our everyday talk, the one meant by the lexical item ‘believes’, explicitly eschews such a commitment to privileged access!This is what indirect reflexive pronouns tell us about belief: the ordinary, linguistically expressed notion of belief cannot be Cartesian. At his commentary on an early version of Shoemaker’s essay, at the Pacific Division of the American Philosophical Association in 1994, Rogers Albritton, reported by Shoemaker (1996: 81), in reaction to the idea that if I believe P then I believe that I believe P, besides pointing out the obvious infinite regress of first-person higher-order beliefs, remarked, of someone who says ‘I believe it’s raining’, ‘‘Where is it written that it takes two beliefs, or even one with two contents [‘‘P’’ and ‘‘I believe P’’], to confess that one believes it’s raining. Why can’t I just bare my soul in the matter?’’ Such an Albrittonian comment! The linguistic argument of this essay shows that Albritton’s intuition was correct.
REFERENCES Atlas, J. D. (1989), Philosophy without Ambiguity (Oxford: Clarendon Press). (1991), ‘‘Topic/Comment, Presupposition, Logical Form, and Focus Stress Implicatures: The Case of Focal Particles ‘Only’ and ‘Also’’ ’, Journal of Semantics, 8: 127–47. (2004), ‘‘Presupposition’’, in G. Ward and L. Horn (eds.), Encyclopedia of Pragmatics (Oxford: Blackwell), pp. 29–52. (2005), Logic, Meaning, and Conversation: Semantical Underdeterminacy, Implicature, and Their Interface (New York: Oxford University Press). and Levinson, S. (1981), ‘It-clefts, Informativeness, and Logical Form: Radical Pragmatics (Revised Standard Version)’, in P. Cole (ed.), Radical Pragmatics (New York: Academic Press), pp. 1–61. Bach, K. (1987), Thought and Reference (Oxford: Clarendon Press). Bo¨er, S., and Lycan, W. (1986), Knowing Who (Cambridge, Mass.: MIT Press). Braun, David (1998), ‘‘Understanding Belief Reports’’, Philosophical Review, 107: 555–95. (2002), ‘‘Cognitive Significance, Attitude Ascriptions, Ways of Believing’’, Philosophical Studies, 108: 65–81. Casta˜neda, H.-N. (1966), ‘ ‘‘He’’: A Study in the Logic of Self-Consciousness’, Ratio, 8: 130–57. (1989), Thinking, Language, and Experience (Minneapolis: University of Minnesota Press). Davison, A. (1975), ‘‘Indirect Speech Acts and What to Do With Them’’, in P. Cole and J. Morgan (eds.), Syntax and Semantics 3: Speech Acts (New York: Academic Press), pp. 143–85.
Jay David Atlas
145
Geach, Peter (1957), ‘‘On Beliefs about Oneself ’’, Analysis, 18: 23–4; repr. in P. Geach (1972), Logic Matters (Berkeley and Los Angeles: University of California Press), pp. 128–9. Hintikka, J. (1962), Knowledge and Belief (Ithaca, NY: Cornell University Press). Hunter, J. F. M. (1990), Wittgenstein on Words as Instruments: Lessons in Philosophical Psychology (Edinburgh: Edinburgh University Press). Kripke, S. (1980), Naming and Necessity (Cambridge, Mass.: Harvard University Press). Lakoff, G. (1975), ‘‘Pragmatics and Natural Logic’’, in E. L. Keenan (ed.), Formal Semantics of Natural Language (Cambridge: Cambridge University Press), pp. 253–86. Langford, C. H. (1942/1968), ‘‘The Notion of Analysis in Moore’s Philosophy’’, in P. A. Schlipp (ed.), The Philosophy of G.E. Moore, vol. i (LaSalle, Ill.: Open Court), pp. 321–42. Levinson, S. C. (1995), ‘‘Three Levels of Meaning’’, in F. R. Palmer (ed.), Grammar and Meaning (Cambridge: Cambridge University Press), pp. 90–115. Lewis, D. K. (1979a), ‘‘Attitudes De Dicto and De Se’’, Philosophical Review, 88/513–43; repr. in Lewis, Philosophical Papers, vol. i (New York: Oxford University Press), pp. 133–56. (1979b), ‘‘Scorekeeping in a Language Game’’, Journal of Philosophical Logic, 8: 339–59; repr. in Lewis, Philosophical Papers, vol. i (New York: Oxford University Press), pp. 233–49. Lycan, W. (1984), Logical Form of Natural Language (Cambridge, Mass.: MIT Press). Moore, G. E. (1993), Selected Writings (London: Routledge). Moran, R. (2001), Authority and Estrangement: An Essay on Self-Knowledge (Princeton: Princeton University Press). Murdoch, I. (1985), The Sovereignty of Good (London: ARK). Perry, J. (1977), ‘‘Frege on Demonstratives’’, Philosophical Review, 86: 474–97. Prior, A. N. (1967), ‘‘On Spurious Egocentricity’’, Philosophy, 42: 326–35. Putnam, H. (2002), The Collapse of the Fact/Value Dichotomy and Other Essays (Cambridge, Mass.: Harvard University Press). Quine, W. V. O. (1960), Word and Object (Cambridge, Mass.: MIT Press). Rosenthal, D. (1986), ‘‘Intentionality’’, in P. French, T. Uehling, and H. Wettstein (eds.), Midwest Studies in Philosophy, 10 (Minneapolis: University of Minnesota Press), pp. 151–84. (1993), ‘‘Thinking that One Thinks’’, in M. Davies and G. Humphreys (eds.), Consciousness (Oxford: Blackwell), pp. 197–223. Searle, J. (1975), ‘‘Indirect Speech Acts’’, in P. Cole and J. Morgan (eds.), Syntax and Semantics 3: Speech Acts (New York: Academic Press), pp. 59–82. (1983), Intentionality (Cambridge: Cambridge University Press). Sellars, W. (1969), ‘‘Language as Thought and as Communication’’, Philosophy and Phenomenological Research, 29; repr. in H. Geirsson and M. Losonsky (eds.), Readings in Language and Mind (Oxford: Blackwell, 1996), pp. 214–32. Shoemaker, S. (1996), ‘‘Moore’s Paradox and Self-knowledge’’, in S. Shoemaker, The First Person Perspective and Other Essays (Cambridge: Cambridge University Press), pp. 74–93. Wittgenstein, L. (1974), Letters to Russell, Keynes, and Moore (Oxford: Blackwell).
7 Moore’s Paradox and the Transparency of Belief Jonathan E. Adler and Bradley Armour-Garb
A number of philosophers have thought to extend Moore’s Paradox to concepts other than belief, e.g., I want Tina to go out with me, but I’m indifferent to whether she does. The scope of Moore’s Paradox is the problem of what concepts besides belief will generate a Moore’s (or Moorean) Paradox from the following schema: (MP@) I @ that (p, but it’s not the case that I @ that p), where @ is a schematic variable ranging over elements of a delimited class of propositional attitude verbs (e.g., desire, belief). In particular, we want to investigate the question, ‘Is the Moore’s Paradoxical nature of the paradigm, (MPi ) p, but I don’t believe that p, inherited from being instantiated in (MP@)?’ If so, what does this tell us about the nature of belief ? If we compare (MPi ) with another putative instance of (MP@), (MPii ) I desire that ( p, but I don’t desire that p), an immediate difference is evident—indeed, one that we take to hold generally: Only with (MPi ) do we eliminate an explicit place for an instance of ‘@’. So an affirmative answer to whether (MPi ) inherits its Moore’s Paradox nature from (MP@) depends upon—but not only upon—whether assertion expresses or implies belief. We examine this matter by reference to a speech-act account of illocutionary force and of the Unger-Williamson thesis¹—that, roughly, the norm of assertion is that one properly asserts that p only if one knows it. As we will show, reflection on this norm serves to elucidate the relevant notion of belief. Thanks to John N. Williams and an anonymous referee for comments. ¹ Peter Unger, Ignorance (Oxford: Oxford University Press, 1975). Timothy Williamson, ‘Knowing and Asserting’, Philosophical Review, 105/4 (1996), 489–523, repr. with minor additions in his Knowledge and Its Limits (Oxford: Oxford University Press, 2000), ch. 11.
Jonathan E. Adler and Bradley Armour-Garb
147
The question of whether (MPi ) fits under schema (MP@) leads us to critical reflection on the property we regard as central to resolving Moore’s Paradox—that belief is transparent to its content: To believe that p is for it to be the case (from one’s point of view) that: p.² With others, we construe Moore’s Paradox as essentially first-personal.³ Does the first-personal transparency of belief fail for other instances of @ that might be held to generate variants of Moore’s Paradox? It does not appear to hold for degrees of belief as with forms like: (MPiii ) I’m pretty sure that (p, but I am not inclined to believe that p). In order to address these questions, we return to the beginning: Is Moore’s Paradox paradoxical? That is, does it affirm a set of statements, each of which is highly credible, and yet which are incompatible? If it is paradoxical, what yields the paradox? If not, what does it tell us about the putative culprits that are implicated, belief and assertion?
1 . PA R A D OX I C A L I T Y A N D M O O R E ’ S PA R A D OX Moore’s Paradox is perplexing or puzzling, and the task is to explain how it might arise and how it is avoided. The puzzle is that a sentence that could be true is ‘heard’ as inconsistent, when asserted. Since the apparent inconsistency is among statements each of which has a plausible grounding, Moore’s Paradox can be presented in a paradox-like form. Colloquially, an instance of Moore’s Paradox, which takes off from a sentence of the form (1) p but I do not believe that p, appears to affirm one thing only to deny it, or to take it back. The suggestion of a contradiction when asserted (or even believed) does not lead us to reject a plausible assumption of the apparent paradox set, as is standard in diagnosing and treating a paradox; rather, it demands an explanation as to why it arises and why it is neither asserted nor believed.⁴ More specifically, the target question behind Moore’s Paradox is this: Why does assertion (or belief) transform a consistent, assertible sentence into a recognized (and unacceptable) inconsistency? ² Our account draws upon that in Jonathan E. Adler, Belief ’s Own Ethics (Cambridge, Mass.: MIT Press, 2002), ch. 7. ³ For insightful reflections on Moore’s Paradox from the first-person point of view see Sydney Shoemaker, The First-Person Perspective and Other Essays (Cambridge: Cambridge University Press, 1996), ‘Moore’s Paradox and Self-Knowledge’, ibid. 359–70, and Richard Moran, Authority and Estrangement: An Essay on Self-Knowledge (Princeton: Princeton University Press, 2001). ⁴ It bears noting, for what follows, that one author (Armour-Garb) of this paper allows for the acceptance of some contradictions. While he finds ‘dialetheism’ an attractive and plausible view, he recognizes the virtue of consistency and, thus, aims to investigate (and resolve) Moore’s Paradox with that norm in mind. For more on dialetheism, see Armour-Garb, Dividing the (Semantic) Paradoxes (Stanford: CSLI, forthcoming).
148
Moore’s Paradox: Transparency of Belief
In order to answer this target question, we should note the two forms of Moore’s Paradox: A. p but I do not believe that p and B. p but I believe that not p.⁵ For what follows, we focus on the A form (herein, ‘MP’), as we take it to be more basic than the B form, which form, as we establish below (§2), while contradictory, is not paradoxical. The task, then, is to investigate its maximally generalized form, in order to determine the scope of Moore’s Paradox. There is, we maintain, a general form of MP, viz. (2) p, but I do not @ that p for ‘@’ an attitude verb. The ‘I’ in (2) is necessary, for only then are we ensured the self-recognition that is needed for the proposition in question to be attached to the thinker/speaker in such a way as to generate the symmetric, but negated, content with the second conjunct. As already noted, the paradox is, essentially, first-personal.⁶ We maintain that no other attitude save belief will yield an MP as it stands (leaving aside for now ‘as if’ belief attitudes, such as supposing, assuming, etc.), as no other attitude both links that attitude with assertion and has the unique property of transparency, which links a belief to what is believed: To believe that p is to represent p as being the case. In order to make the case, we return, briefly, to the aforementioned link (we will develop the point further, in §3). Sincerely to assert that p is both to put that p forward as true and to express your belief that p. This is no accident: By sincerely asserting something, you express that you believe it; to believe it is for it to be represented to you, the believer, as true; accordingly, to assert something is to put it forward as believed and, thus, as believed to be true. This explains why an instance of the A form is not (and, indeed, ought not ever to be) asserted:⁷ In asserting a sentence of the A [MP] form, e.g., ⁵ Anthony S. Gillies, ‘A New Solution to Moore’s Paradox’, Philosophical Studies, 105/3 (2001), 237–50, discusses the B form, though his analysis does not seem to leave room for the contrast between it and the A form. ⁶ What is crucial is self-recognition and self-ascription. One can, of course, construct a version of MP that relies on the first-personal report without the personal pronoun, as one can refer to oneself by name. Of course, such cases can be problematic (think of Kripke’s Pierre), so we ignore them here (although, for more on such cases, see Dale Jacquette, ‘Identity, Intensionality, and Moore’s Paradox’, Synthese, 123/2 (2000), 279–92). Since both self-recognition and self-ascription are guaranteed by ascriptions (or thoughts) that make use of the first-personal pronoun, we will employ it, for what follows. ⁷ In so claiming, we accept Rosenthal’s (‘Moore’s Paradox and Crimmins’s Case’, Analysis, 62/2 (2002), 167–71) and H´ajek and Stoljar’s (‘Crimmins, Gonzales and Moore’, Analysis, 61/3 (2001), 208–13) treatment of Mark Crimmins’, case (‘I falsely believe that p’, Analysis, 52/3 (1192), 191), but maintain that the foregoing diagnoses it in a slightly different way.
Jonathan E. Adler and Bradley Armour-Garb
149
Jones is rich but I do not believe that Jones is rich, you express, in a single thought, that I believe that (Jones is rich, but I do not believe that Jones is rich). Thus, given that belief distributes over conjunction, and that all that is taking place is within a single, overt thought, we get I believe that Jones is rich and I believe that I do not believe that Jones is rich. This thought is the same as (i.e., is logically equivalent to) I believe that Jones is rich, I believe that I do not believe that Jones is rich. Recall that the transparency of belief holds that to believe that p is for it to be the case, for one’s point of view, that: p. From the first-person point of view, belief is factive, a key property of knowledge (if S knows that p then p).⁸ From the first-person point of view, then, the last statement implies I do not believe that Jones is rich. There is now an explicit contradiction with the previous statement, ‘I believe that Jones is rich.’ In simple and obvious steps, we then have derived a contradiction from the nature of assertion (and belief) alone, which is what is necessary for an explanation of why the assertion of Moore’s paradoxical sentence is heard as contradictory. Our analysis assumes a strong parallel between belief and assertion. Assertion expresses, not only implies, one’s belief; but it also the case that belief, like assertion, claims the truth of its content, even though that claim only becomes apparent, upon attending to one’s belief. Thus, one stands behind it as true, as the speaker stands behind his assertion as true.⁹ As no other attitude claims the truth of p, no other attitude is expressed by assertion, in which case no instance of (2), with ‘@’ a propositional attitude distinct from belief, will yield a contradiction with the first conjunct. With the other attitudes, we need to add the @ explicitly, as in (3) I desire [hope, guess, imagine] p, but I do not desire [hope, guess, imagine] p. But there is nothing puzzling here, since the starting assumption of the consistency of the sentence affirmed fails. (3) is, effectively, an instance of the explicitly contradictory form: C. @I p & ∼@I p. ⁸ Of course, we do not mean to imply that belief actually is factive. Rather, we mean that, in believing that p, the believer represents p as actually being the case. ⁹ We can then say that a speaker is entitled to assert that p (viz., that that p is assertible (for the speaker)) only if she is in a position that entitles her to claim that what she asserts is true.
150
Moore’s Paradox: Transparency of Belief
(3) and its ilk (for other attitudes) are contradictory sentences. So there is no discrepancy between the consistency of the sentence and the inconsistency ‘heard’ upon assertion. What of the B form, (4) I desire [hope, guess, imagine] p, but I desire [hope, guess, imagine] not p? The more general form here is D. @I p & @I ∼ p, with the ascribed propositions overtly contradictory. Sentences of the D form indicate the same attitude towards contradictory propositions, rather than an ascription of the same attitude toward a contradiction, and in the D form, we surrender the crucial implication of (4) that the agent grasps in a single first-personal thought his same attitude toward the contradictory propositions. But what of the attitude itself? Can we desire that p while simultaneously desiring its negation (in a single consciousness, as it were)? Assuming a preference for consistency, we can have no plan to implement, to act on, or to realize, both, as the result would be overtly self-defeating.¹⁰ But what of cases like this: I want Mary to come to the party (so I can meet her) but I don’t want [want not for] Mary to come to the party (because she’ll bring her unpleasant pal David)? Once the tacit restriction to a certain respect is factored in the contradiction dissipates. Or consider the following: (5) I desire to eat ice cream (because of the taste) but do not desire [or: want] to eat ice cream (because it is not good for me). This yields an inconsistency, if treated as of the C form. But, as is more plausible, it is believed as consistent—by speaker and hearer alike. That is, the sentence, when asserted, is ‘heard’ as inconsistent but is ‘treated’ (once interpreted) as consistent. So interpreted, it expresses a first-order desire to eat ice cream, and, as well, a second-order desire not to have that desire, a form of conflict that we discuss further below. Whether these are actually inconsistent or not, they are not, and could not be, cases of MP. For, even if contradictory, the contradiction arises from the sentence, rather than from the assertion of the sentence.¹¹ By contrast, in the ¹⁰ Cf. Michael Bratman, Intentions, Plans, and Practical Reason (Cambridge, Mass.: Harvard University Press, 1987). ¹¹ It might seem that the analogy with (5) (and others like it) could be sustained by simply prefacing assertions of the (MPii ) form with the operator, ‘I believe that’. In fact, however, to assert ‘I believe that p’ is to express a hesitancy to assert that p as all-out believed. It is, rather, to put forward a qualified, or a guarded, assertion that p, which is to convey a conversationally induced weakening of the assertion of that p, rather than an unqualified expression of belief (consider, for example, ‘I believe that George is in the office but do not hold me to it’).
Jonathan E. Adler and Bradley Armour-Garb
151
case of belief, the contradictory nature of MP follows only from the MP sentence (in the A form) as asserted.¹² This enables us to distinguish the first form of MP from the second: An instance of the B form implies the inconsistency of what is believed, rather than inconsistencies of whether something is believed, as occurs in an instance of the A form.
2 . M O O R E ’ S PA R A D OX A N D VA R I E T I E S O F B E L I E F Thus far, we have marked a difference between belief and other attitudes, in support of our claim that no attitude but belief can (or: will) yield a MP, and by identifying both the features and the links that give rise to it. But ‘belief’ is a blanket term that covers divergent attitudes of different degrees of strength. As we will discuss below (§3), we treat belief as full belief, as only full belief possesses the needed transparency. Now, we consider a sharp contrast of belief with the other attitudes, which contrast emerges from a consideration of iterations of second-order attitudes. Consider the notorious datum that you can desire to smoke a cigarette but not desire that you desire to do so, i.e., that attending to your desire does not invariably yield a (second-order) desire, and so it goes for other attitudes (hoping, liking, etc.).¹³ Here we have a credible instantiation of (MP@), with ‘desire’ for @, i.e., (6) I desire that (I smoke this cigarette, but also that I do not desire that I smoke this cigarette). There is nothing contradictory about (6), whether asserted or thought, even if it is awkwardly expressed.¹⁴ Also, and corroboratively, there is no difference between (6) and its second-person attribution: (7) He desires that (he smoke that cigarette, but also that he does not have that desire). ¹² Here we have a clear difference of the (B) form. In the (B) form an asserted instance would be, effectively, to assert a contradiction, e.g., Jones is rich but I believe that Jones is not rich. Treating ‘I believe’ here not as a qualifier (a weakener) of the assertion, but (in its standard way) as transparent, the assertion reduces to Jones is rich and Jones is not rich. Of course, to infer such a contradiction would be a confusion, given the pragmatically implied weakening in the assertion of a self ascription of belief (cf. n. 11, above). ¹³ For a discussion of related issues, see John N. Williams, ‘Wittgenstein, Moorean Absurdity and Its Disappearance from Speech’, Synthese, 149/1 (2006), 225–54, §11. ¹⁴ In fact, it is not at all unusual for one to desire to have the opposed first-order desire, which is (roughly) an analogue of the B form, i.e., (6*) I desire that (I smoke this cigarette and that I have the desire not to smoke it).
Moore’s Paradox: Transparency of Belief
152
What is, trivially, not possible is only for the second-order desire to succeed, and for the first-order desire to remain. But if you believe that the #3 train stops at Franklin Avenue, and if you attend to that belief, then (unless you revise your original belief) you believe that you believe it, which can be captured by the following abstract form, (BI) B(p) (and attending to your belief that p) B(B(p)).¹⁵,¹⁶ What is involved is not a further endorsement of p, but is, rather, a non-optional judgment that one does so believe. That is, the recognition that it is true that one has the belief that p is thereby to believe that one believes that p. What then of degrees of belief—of uncertain, non-full, or partial belief ? There are a number of ways to think of these. We begin with (8) Probably p, but it’s not probable that p. This is just an instance of the explicitly contradictory D. If we move the negation in, i.e., to (9) Probably p, but probably not p, then we seem to have another instance of (8), given the negation law for probability: For the probability of p to be less than .5 is for the probability of ∼p to be greater than. 5. However, ‘not probable’ can be used as a really full negation, where one has no degree of belief in a proposition at all (Of course, this is forbidden in standard treatments).¹⁷ The following seem MPs as well, and so degrees of belief stand with belief, rather than with the other attitudes in this respect: (10) p, but it’s not probable that p, understood as something like (11) p, but the probability of p is less than half, which may (but need not) be heard as something like (12) p turned out to be true, despite its being improbable, ¹⁵ The same goes for (BE) B(B(p)) (and attending to your believing it) B(p) and (B∼E) B(∼B(p)) (and attending to your believing it) ∼B(p), as we hope is obvious. We should note that we employ the turnstile because, as is standardly used, it captures both truth preservation and inferential entitlement, both of which we apply to the present cases. Because constitutive of belief is only the judgment or claim that p, we think that these iterative belief principles do not run afoul of Williamson’s powerful criticisms of the assumption of luminosity (Knowledge and Its Limits, ch. 4). But we cannot address the issue here. ¹⁶ Given (BI), considerations from n. 4, and (BE), and assuming that we are attending to our beliefs (NB, even believing that we believe), we get, and so are entitled to infer (from the attending) (BE/O) B( p) B(B( p)). ¹⁷ So, for example, we have no degree of belief that Hamlet blinked an even number of times on the day of his sixteenth birthday, nor do we have a degree of belief in its negation.
Jonathan E. Adler and Bradley Armour-Garb
153
and which, if so heard, is not only consistent (as asserted), but is, in fact, familiar. What then of epistemically qualified assertion, as in (e.g.) (13) I’m pretty sure that (p, but I don’t believe that p [highly]) or (14) I’m pretty sure that (p, but it’s doubtful that p)? (13) either itself, or as asserted, is not just consistent, but possibly an implication like the following: ‘I’m [just] pretty sure that Jones is in Alaska, so, of course, I do not [fully] believe it.’ Likewise, (14) seems intelligibly understood as e.g. ‘I’m [just] pretty sure that Jones is in Alaska, though I have acquired reasons to doubt that he is.’ Or, better, with ‘because’ substituted for ‘though’. However, assertions as expressions of all-out belief (or acceptance as true) must represent themselves as satisfying a total relevant available evidence condition. The body of evidence grounding each conjunct must be the same body of evidence that purports to be the total relevant evidence. Not only is this evidence all that the agent has available, but in accepting that proposition the agent implies this is all the evidence necessary to establish it as true. Qualified assertions like (13) or (14) need not have the latter implication. They remain open to inquiry, to the discovery of further relevant evidence, and so to an alteration in the degree of qualification. They are failed assertions because the fundamental goal of assertion is to transmit truths about the world, rather than about one’s attitude toward it (except, of course, in the odd cases in which these are of interest to the hearer). So, for example, in asking after the whereabouts of Jones, what you want to know is where Jones is, not where the speaker thinks that he is. Accordingly, it does not satisfy your interest in being so informed, to learn only that the speaker is pretty sure. Also, notice that these examples fall in place with the other propositional attitudes, in having to explicitly represent the attitude in the sentence, and unlike either (full) belief or the basic case, A.¹⁸ But why is this so? We maintain that the reason turns on certain features of full belief, which have ramifications for other (familiar and neighboring) paradoxes.
3 . B E L I E F A N D T R A N S PA R E N C Y A central difference between full belief and partial, or qualified, belief (in addition to other attitudes) is that only the former is transparent to its content. Recall: For you fully to believe that p is for it to be true that p (from your first-personal ¹⁸ As a consequence, the logic of the statement is the same as the logic of the assertion—markedly unlike cases under (A).
154
Moore’s Paradox: Transparency of Belief
point of view).¹⁹ Thus, full belief is not essentially any strength of attitude—you ‘disquote’ from your attitude. Now consider any strength of belief-like attitude in an MP version: E. p, but I’m not very sure that p.²⁰ Although, in this stark form, the E form assertion seems puzzling, and so calls for the speaker to enter some explanation, such assertions (or their kin) are assertible, when construed literally. In order to see this, consider (15) Suzy loves me, but I’m having doubts about it. Nervous Ned explains: ‘All the signs are that Suzy loves me, and we get along great. But that hunk Tony just walked by, and I thought he might be attracted to Suzy.’ Assuming that Nervous Ned remains competent to judge his love-relations, his expressions of doubt allow for the assertibility of his (psuedo-) MP, and its literal acceptance: Both conjuncts of (15) can be true (i.e., true-from-Ned’s-point-of-view), and assertible, without paradox. We think that this result holds quite generally. It undermines any attempt to read full or all-out belief as too weak to imply the truth of what is believed (from the first-person point of view).²¹ For if belief is any strength of attitude, it should exclude doubt, as in the standard Peircean view.²²
4 . T H E N O R M O F A S S E RT I O N A N D B E L I EV I N G W H AT YO U K N OW Recall that the Unger-Williamson thesis maintains that, in asserting that p, one represents oneself as knowing that p. Along a different vein, Moore claims that ‘by asserting p positively, you imply, though you don’t assert, that you know that p.’²³ We are sympathetic to the claim that (sincerely) asserting that p represents oneself as knowing that p and, following Williamson, will assume ‘the Knowledge Rule’ that tracks it, i.e., the norm according to which you must assert that p only if you know it. Rather than arguing for the Knowledge Rule, we shall employ it, together with reflection on full belief and assertion. ¹⁹ By contrast, partial belief ‘marks’ a qualified, or guarded, assertion, rather than a bald assertion of that p itself. The uptake of partial belief is that it transforms putatively paradoxical utterances into non-paradoxical, even familiar, ones. Thus, consider a paradoxical (or incoherent) assertion of the form, ‘p but it is not very probable that p’. ²⁰ For a related variant of MP and a different diagnosis, see Byeong Lee, ‘Moore’s Paradox and Self-Ascribed Belief ’, Erkenntnis, 55/3 (2001), 359–70. ²¹ Here, and in what follows, we use ‘full’ and ‘all-out’ interchangeably. ²² Cf. Charles, Peirce ‘The Fixation of Belief ’, in his Essays in the Philosophy of Science, ed. V. Thomas (New York: The Bobbs-Merrill Co., 1957), 3–30. ²³ The Commonplace Book of G. E. Moore 1919–1953, ed. Casimir Lewy (London: George Allen & Unwin Ltd; New York: Macmillan Co., 1962), 125.
Jonathan E. Adler and Bradley Armour-Garb
155
The Knowledge Rule states: (KR) One must: Assert that p only if one knows that p, where ‘must’ expresses a kind of obligation and where the rule thus yields that only knowledge warrants assertion. As Moore, Unger, and Williamson note,²⁴ we seem to get just a variant of MP, with ‘believe’ replaced by ‘know’. Given (KR), this is easily explained: If only knowledge warrants assertion then to have warrant to assert the conjunction, (MP*) p but I do not know that p, is for the asserter to know that (p and that he does not know that p). Assuming that one can know a conjunction only if one knows each conjunct then, if one has warrant to assert an instance of (MP*) then one knows the first conjunct—that p. But one knows a conjunction only if it is true, which, via disquotation and the aforementioned norm, yields that one does not know that p. Thus, (KR), and the assumption that the speaker has warrant to assert an instance of (MP*), yields a contradiction, thereby supporting the paradoxicality (and, so, the unassertability) of (MP*).²⁵ The knowledge rule forbids the combination: one asserts that p when one does not know that p. Thus, if (KR) is in place, we are obliged to assert that p only if we know it; and, if we do not know it, we are obliged not to assert it. Of course, we violate (KR) all of the time, as Williamson is well aware. But the norm motivates an observation that will play a role in what follows: To make an assertion is to confer a responsibility (on oneself) for the truth of its content; to satisfy the rule of assertion, by having the requisite knowledge, is to discharge that responsibility, by epistemically ensuring the truth of the content.²⁶ Granting the obligation (which we do), it does, however, raise a question: Even if we must assert that p only if we know it, under what conditions do we take ourselves to be entitled to assert it? To be sure, one might reply that we are entitled to assert that p when we have warrant (and, thus, authority) to do so, and that we have warrant to assert it only when we know it. But an infinite regress ²⁴ Ibid. 277; Unger, Ignorance, 256–60; and Williamson, Knowledge and Its Limits, ch. 11. ²⁵ Beyond noting this key argument for (KR), the Unger-Williamson thesis, and a cautionary word, we shall just assume them. The cautionary word is this. Someone might naturally and sincerely (and perhaps dejectedly) assert, about the outcome of a fair lottery in which he holds a single ticket, (15∗ ) I’ll lose. In so asserting, the speaker surely does not imply that he knows that he will lose. However, in this case, it is presumed as mutually evident that a lottery is a gamble and that its outcome (like many kinds of future events) cannot be known in advance. So there is a mutual accommodation to allow for the assertion of (15∗ ), rather than for what is, strictly speaking, correct, viz., (15∗∗ ) I’m almost certain I’ll lose. ²⁶ Williamson ‘Knowing and Asserting’, 521–2.
156
Moore’s Paradox: Transparency of Belief
threatens, if our entitlement to assert depends on a confirmation of its warrant, which notions, of course, are not identical. This, however, is a mistake, although an important one. The knowledge rule, qua norm, is taken to be constitutive of the speech act of assertion. As such, it tells us the conditions under which we have the authority to perform that act. But it does not follow from (KR) that if we do not know that which we have just purported to have asserted (because it is false, say) then we have failed to assert it. We have so asserted, but we oughtn’t to have done so: We violated the rules, rather than failing to perform the speech act. But this does raise a question: How, and under what conditions, do we take ourselves to be entitled to assert? The answer, it seems, is clear, assuming (KR):²⁷ As we take ourselves to be entitled to assert that p only when our assertion conforms to (KR), we take ourselves to be entitled to so assert when we take ourselves to know that p, i.e., we take ourselves to be entitled to assert that which we believe that we know.²⁸,²⁹ Recall (BI)—that if one attends to her belief that p then she believes that she believes it. Now suppose that the thinker attends to her belief and that what she believes is presented to her as fully believed and, thus, as transparent. In that case, if she attends to her belief, so that its content appears to her transparently, then, we maintain, she believes herself to know that content, in which case she takes herself to be entitled to assert it, as it is presented to her as known, even if, in fact, it is false, she does not know it, etc. So, what is fully believed is thereby taken by the believer to be known; and, insofar it is so taken, it is taken to be assertible, in accordance with the Knowledge Rule.³⁰ If this is right then we have a ‘test’ for determining both full belief and the appearance (to a speaker) of assertability: When one attends to her belief and when the content of that belief is presented, to the believer, as transparent, then ²⁷ Williamson (ibid., passim) makes clear that he takes the knowledge rule to be the actual norm of assertion—the norm that we actually employ and that guides our practices of asserting. Accordingly, and assuming the norm, we take it to play a role in elucidating the actual concept of full belief. ²⁸ Williamson (ibid. 512–13) argues against a norm that bears a resemblance to our present proposal. The norm he rejects is the BK rule, (BKR) One must: assert that p only if one believes that one knows that p. It is important to note that the present proposal does follow Williamson in rejecting (BKR). The present proposal is to treat an element of (BKR) as related to (KR), by setting out a condition under which we take ourselves to be entitled to assert: We take ourselves to be entitled to assert that p when we have a full belief that p, and we have a full belief that p when, having attended to our belief, we come to believe that we know it. ²⁹ Notice, in addition, that, if correct, this is related support for the claim that full belief is not essentially any strength of attitude. ³⁰ One might object that, while it may well be true that we take ourselves to be entitled to assert something, when we take ourselves to know it, it is wildly implausible to assume that, when we attend to a belief, that belief is then represented to us as known, in a non-metaphorical way. This objection seems to be based on a mistake, for, when one attends to, or is guided by, a full belief that p, what she attends to, or is guided by, is that p, rather than one’s attitude toward it.
Jonathan E. Adler and Bradley Armour-Garb
157
she, in effect, believes that she knows it and so takes herself entitled to assert it, as it is what she fully believes. 5 . F U L L B E L I E F A N D FA L L I B I L I T Y Try as we might, and with only the best of intentions, we are liable to assert or believe things that are false and are liable to disbelieve things that are true. That is, just as the preface-assertion (‘at least one statement in this book is false’) is likely true, though not assertible, the same is true of (what we call) the fallibilist-assertion viz., (FA) I believe at least one false proposition. (For now, we put aside the pragmatic weakening implied by the ‘I believe’, and treat this as an accurate report in thought.)³¹ Since we have such an extensive and wide-ranging set of beliefs, it would be hubris not to be sure that some of them are false (we are, after all, not omniscient). So (FA) is likely—perhaps, inevitably—true, and, if it is believed, our full set of beliefs is inconsistent. But, if (FA) is asserted, we can run an argument that parallels the Preface Paradox, with the same disastrous results. We review it briefly, with especial attention to the standard presentation of the Preface Paradox, with which we begin.³²
5.1 The Preface Paradox In a version of the Preface Paradox applicable to all, Smith believes each of an enormous and diverse set of propositions: Bp1 , Bp2 , . . . , Bpn . Call the set of propositions that Smith believes ‘Smith’s corpus’. As Smith is minimally fallibilist, he thinks that some of the members of the his corpus are erroneous, i.e., B∼(p1 & p2 &, . . . & pn ). Call the proposition that he believes, the preface-proposition. What Smith takes to be erroneous—what his fallibilism appears to commit him to—is not one or another of the members of the corpus; rather, it is that he thinks of the full range of members of the corpus as likely to contain an error. His fallibilist basis for his inference to the preface-proposition is second-order—a reflection on his believing (the members of) the corpus, and not on the content—the particulars —of what he believes. ³¹ See nn. 11 and 12. ³² D. C. Makinson, ‘The Paradox of the Preface’, Analysis, 25/6 (1964), 205–7.
158
Moore’s Paradox: Transparency of Belief
The paradox depends on assuming that the person not only fully believes each pi , but the fallibility-proposition, as well. If so, the person’s beliefs are inconsistent. However, it is standard to claim that, whereas inconsistent beliefs could be rational to maintain, only an outright belief in a contradiction is forbidden: B[(p1 & p2, &, . . .& pn ) & ∼(p1 & p2 &, . . .& pn )]. We can avoid this last step if we reject a number of assumptions—most prominently, a conjunction rule (to believe each of a set of propositions is to believe their conjunction). However, this route is not open to us because, given the transparency of belief, we can ‘detach’ from the ‘B’ operator, and then we do get the out-and-out contradiction: (p1 & p2, &, . . .& pn ) & ∼(p1 & p2 &, . . .& pn ) by logical equivalence (of p1 ,. . .,pn and p1 &. . .&pn ). Instead of rejecting the conjunction rule, we reject the fallibility proposition—no one all-out believes that one of her beliefs is false.³³ In a natural framework of assertion, the person asserts each sentence that expresses these propositions—including his denial of their joint truth (viz., the fallibility-proposition). In the context of assertion, the conjunction principle holds: There is no significant contrast between a conjunction of assertions and an assertion of a conjunction.³⁴ Now while we do find people saying things like ‘Blah-blah-blah, though of course I’m fallible, I could be mistaken,’ if we take this expression in a woodenly literal way, rather than as a rhetorical device, we seem to have an overt expression of belief that is unassertible because blatantly inconsistent: The speaker claims a set of propositions (simultaneously) true, and then denies it.³⁵ Of course, as a practical matter, we are not held to all our assertions over time. We forget, and we change our mind without in each case explicitly having to withdraw a distant assertion. The connection between the Preface Paradox and Moore’s Paradox is even tighter, given that disbelief implies a lack of belief.³⁶ But, in addition, a problem ³³ Like Moore’s Paradox, as well as (FA), the Preface Paradox is essentially first-personal. To see this, consider the fact that all or most of us can (and do) all-out believe that at least one of another cognizer’s beliefs will be false. And this is so even though we may not know which ones are the to-be-rejected culprits. Something similar applies, when we take a third person attitude towards ourselves qua believers. When we do, we find that we can say, for example, that at least some of our beliefs are likely false. Thanks to an anonymous referee, who pressed us to address this issue. ³⁴ Michael Dummett, Frege: Philosophy of Language (2nd edn. Cambridge, Mass.: Harvard University Press, 1981), 336. ³⁵ We assume, of course, that, in asserting, a speaker represents himself as recognizing explicit relations like negation. ³⁶ So that the (B) version implies the (A) version.
Jonathan E. Adler and Bradley Armour-Garb
159
emerges: Whereas the unassertibility of Moore’s Paradox corresponds to an incoherence of thought, the Preface Paradox develops a line of coherent thought. Thus, many follow Makinson in taking the right response to the Preface Paradox to be that one can rationally be involved in (recognizable) inconsistency.³⁷ Accordingly, in order consistently to resolve the Preface Paradox, one must resolve it in such a way as to explain the coherence of the preface-assertion without its assertability.
5.2 A Fallibilist Paradox³⁸ As fallibilists, we recognize that it is overwhelmingly likely that some of our beliefs are false. Thus, like the preface-assertion we are faced with the fallibilist-assertion (FA), and, with it, the contradiction in belief that appears to ensue. This is the Fallibilist Paradox. In both paradoxes, we seem faced with a dilemma: We have derived a contradictory belief from a collection of beliefs each of which we accept.³⁹
5.3 Resolving the Preface and the Fallibilist Paradoxes Our resolution consists of two related parts, already suggested and implicit in our account of Moore’s Paradox: First, the fallibility-proposition is not all-out believed and, second, the justification for the fallibility-proposition, and the proper expression of fallibility, is second-order: It is to cast doubt on our perfection as believers, not to cast doubt on what we believe (viz., the first-order propositions).⁴⁰ There is no inconsistency in (fully) believing each of a set of propositions, and having a high degree of belief that not all of them are true—indeed, something like the latter follows from the conjunction principle for probability, under very ³⁷ Cf., e.g. Graham Priest, ‘What is So Bad about Contradictions?’, Journal of Philosophy, 95/8 (1998), 410–26. It bears pointing out that dialetheists, while they believe that some contradictions are both true and assertible, do not accept that all contradictions are true and assertible. Thus, accepting dialetheism by no means entails that dialetheists will (must, should) take the relevant instance of (e.g.) a Moore-style sentence to be genuinely inconsistent. Indeed, dialetheists, like their consistentist kins, subscribe to a consistency-governed norm; they simply maintain that there are reasons for accepting some contradictions. For the details, see Graham Priest, J. C. Beall, and Bradley Armour-Garb (eds.) The Law of Non-Contradiction: New Philosophical Essays (Oxford: Oxford University Press, 2004). ³⁸ The indefinite article is used in recognition that there is already a‘fallibility paradox’ (cf. Keith Lehrer and Kihyeon Kim, ‘The Fallibility Paradox’, Philosophy and Phenomenological Research, 50, suppl. (1990) 99–107). Ours is distinct from theirs, but indebted to their presentation (Oxford: The Clarendon Press, 2004). ³⁹ Although the presentation of the paradox is fairly standard, one might balk at the assumptions that drive it. To assuage the reader’s worry, consider the following: If one accepts the prefaceassertion and the fallibilist-assertion then a contradiction can be adduced, from the fact that, for each proposition that we believe, qua belief, we believe it to be true. ⁴⁰ See also Simon Evnine, ‘Learning from One’s Mistakes: Epistemic Modesty and the Nature of Belief ’, Pacific Philosophical Quarterly, 82/2 (2001), 157–77.
160
Moore’s Paradox: Transparency of Belief
weak conditions. Moreover, the denial of all-out belief fits with one’s treating the fallibility-proposition as an epistemic possibility that all of one’s beliefs are correct.⁴¹ This, in turn, serves to explain why the fallibility-proposition is not all-out believed (though it may be partially believed) and so why it is not judged to be assertible by the believer, even if, in fact, it is true. Can one take oneself to know the fallibility-proposition? We maintain that one cannot, since (i) one will take it to be epistemically possible, even if wildly unlikely, that all of her beliefs are true, and (ii) knowledge implies that it is not possible (from one’s point of view) that one is mistaken. We grant that it is overwhelmingly likely (from your point of view) that at least one of your beliefs is false. But we deny that there is any basis for detaching from this subjective probability to the all-out belief that the fallibility-proposition is true. The situation parallels that of a fair lottery: There is no ground to detach from the overwhelming probability that one will lose (with a single ticket) to take it as simply true (as if the lottery winner had already been announced) that one will lose. That said, there does seem to be something that we believe and, thus, something that we take ourselves entitled to assert. But, as our belief is merely partial, e.g., of the sort that it is likely that at least one of the sentences of my book is false (though I know not which it is), the assertion that expresses such a (partial) belief is guarded (or qualified) as well. The same holds for the fallibilist-assertion: It is guarded or qualified because what it expresses is a partial belief, which lacks the transparency that is a hallmark of full belief. But the (qualified or guarded) assertion that at least one of the sentences of my book is false is, itself, neither paradoxical nor problematic, even given the full belief in each of the sentences of the book. Thus, the paradox is diagnosed— it arises from the false assumption that we fully believe the fallibilist-proposition—and treated— we can fully believe our qualified belief, which belief thus does not (and, indeed, cannot) yield paradox. We have claimed that a key to resolving these paradoxes is to resist the temptation to read such beliefs (like the fallibility proposition) as full beliefs. A further reason for not treating the fallibility-proposition as all-out believed is that, contra the consequences of the Preface Paradox, where full belief in the content of each of the sentences of the author’s book is retained, in the case of the Fallibilist Paradox, the analogous belief casts doubt back on the individual beliefs. So, for example, suppose that, after grading the final exams for your logic class, you discover that no students have scored below 90 per cent, which is far above the class average. You consider each student and provide reasons for thinking that student to be honest and so conclude, of each of the many students, that that ⁴¹ Indeed, if one believes, of each of his beliefs, that it is true, he will not (and cannot, without contradiction) deny the epistemic possibility that all of what he believes—including their conjunction—are true.
Jonathan E. Adler and Bradley Armour-Garb
161
student did not cheat. But the class average is just too high and so you conclude that at least some of the students must have cheated, in spite of the fact that you do not have evidence that would enable you to convict any particular student. It seems then that a dilemma ensues, for it seems that you (fully) believe each of the students to be honest but you also believe that someone has cheated. As in the previous cases, the dilemma is resolved by noticing that you would not (or: should not) all-out believe that each of your students have performed the exam honestly. Accordingly, this case is contrasted with the Preface Paradox, where the warrant for the preface-assertion is simply the high probability that one of p1 , p2 , . . . ,pn is false, which licenses no inference to the preface-assertion itself. In the case of the students, there is a basis for an inductive inference: The best explanation for the coincidence is that some students cheated (not merely that this is probable).⁴² In the case of the exam, once you conclude that something has gone wrong—that it is overwhelmingly likely that someone has cheated—you find that you can no longer fully believe that each of the students are honest test takers, and this is the case consistent with your fully believing that some of your students have cheated.
6 . C O N C LU S I O N Our account implies the controversial claim that instances of the following are variants of MP and, thus, are paradoxical, if asserted or fully believed: p, but I lack sufficient evidence that p. p, but my reasons do not establish that p. Any hearer would refuse to accept the former conjunct, given an assertion of the latter. (Recall previous discussion that assertion and all-out belief both claim to satisfy a condition of total relevant available evidence as representative of the possible non-misleading evidence.) For another example, consider (16) There are no misspellings in my book, and I have not checked the last section. Reasons that establish the truth of p would normally be taken as reasons sufficient to know that p. Reflection on these examples amounts then to an extension, as well as to a corroboration, of our basic analysis. More generally, it seems that any statement of the form p but I M that p, ⁴² Analogously, while you do not all-out believe (FA) (you partially believe it), your partial belief does not undermine the epistemic status afforded to the content of what you fully believe.
162
Moore’s Paradox: Transparency of Belief
if all-out believed, will be a version of Moore’s Paradox, if M serves to cancel the grounds or reasons for fully believing that p. But no paradox ensues, or is heard, from an assertion of a qualified form, such as I am almost certain that p but I do not know that p. This, we maintain, provides further support for our diagnosis of Moore’s Paradox and, with it, the factivity of full belief, from the first-person point of view.
PART IV M O O R E ’ S PA R A D OX AND CONSCIOUSNESS
This page intentionally left blank
8 Consciousness, Reasons, and Moore’s Paradox André Gallois
On a number of occasions Wittgenstein discusses what he calls Moore’s Paradox. Moore himself discussed the phenomenon Wittgenstein so labels in his paper, ‘Russell’s ‘‘Theory of Descriptions’’ ’, and in a reply to Charles Stevenson.¹ Moore’s own treatment of the paradox is perfunctory. He clearly thinks of the problem posed by examples of Moore paradoxicality as having minor significance with a relatively straightforward solution. Evidently Wittgenstein disagreed. Norman Malcolm reports that Wittgenstein thought of Moore’s Paradox as Moore’s great discovery. Wittgenstein was right to attach the importance he did to the puzzling phenomenon Moore identified.² One reason he was right to do so is, far from being an isolated logical curiosity, it plays a role in a large number of central philosophical issues. I propose to investigate the bearing of Moore’s Paradox on some aspects of the relation between consciousness and having reasons. Here is how I will proceed. In the first section, after, all too briefly, discussing the distinguishing feature of Moore-paradoxical statements and thoughts, I identify an instance of Moore-paradoxicality that directly links Moore-paradoxicality to consciousness. In the same section I go on to say why a standard treatment of Moore-paradoxicality fails to yield a persuasive account of that instance. In the second section I distinguish between two ways of attributing a reason for being in a psychological state with propositional content such as believing, imagining, or intending. In the third section I elaborate on the link between Moore-paradoxicality and the thesis, which I defend, that only consciously held beliefs can supply reasons. The fourth section is devoted to exploring the link ¹ G. E. Moore, ‘Russell’s ‘‘Theory of Descriptions’’ ’ in Philosophical Papers (London: Allen and Unwin 1959), 151–95; Moore’s discussion of the paradox is on pp. 172–7. His later discussion of the paradox is on pp. 542–3 of P. A. Schlipp (ed.), The Philosophy of G. E. Moore (La Salle, Ill.: Open Court, 1942). ² To entitle that phenomenon ‘Moore’s Paradox’ is somewhat misleading in something like the way ‘Naturalistic Fallacy’ is misleading. Just as the alleged mistake entitled the Naturalistic Fallacy is no fallacy so ‘Moore’s Paradox’ labels a phenomenon which, however puzzling, is not a paradox. At least it is not a paradox in the sense of presenting something which is evidently true that seems to imply something evidently false.
166
Consciousness, Reasons, Moore’s Paradox
between Moore-paradoxicality and what I call the deletion and augmentation functions of consciously held belief. What emerges is a connection between consciously held belief having those functions, and one being able to form an integrated view of the world. In the fifth section I attempt to explain what is meant by an integrated view of the world by invoking a comparison with integration in fiction. The sixth section provides an account of the distinction between two ways of attributing a reason introduced in the second. Finally, in the last section I utilize the materials developed in the preceding ones to give an account of the target example of Moore paradoxicality relating to consciousness.
1 . M O O R E - PA R A D OX I C A L I T Y Let us use ‘Moore-paradoxicality’ to label a feature that is shared in common by instances of, among others, the following schemas: (1) P, but I do not believe that P, (2) P, but I believe that not-P. ³ What is it about (1) and (2) that would lead us to classify them as Mooreparadoxical?⁴ Here we need to be careful. In one way it is unfortunate that ‘Moore-paradoxicality’ has become such an inevitable label for the feature Moore called our attention to since it strongly suggests that the thoughts and statements possessing it, or sentences expressing those thoughts and used to make those statements, are, in virtue of doing so, automatically defective. We should resist that suggestion. Whatever we think of eliminativism about propositional attitudes there is nothing defective in an avowed eliminativist’s assertion that eliminativism is true, but not believed by her to be so. Likewise there is nothing defective about a dialethist who believes in true contradictions producing an instance of (2). So ‘Moore-paradoxicality’ should not be taken to pick out a feature whose possession automatically indicates its possessor is defective. Since that is so we should not identify Moore paradoxicality with a certain form of irrationality, or bad type of inconsistency. We need to separate what Moore-paradoxicality is from a diagnosis of why it is usually a bad thing to be. What then is Moore-paradoxicality? We may be tempted to say that to be Moore-paradoxical is just to be an instance of (1) or (2), and leave it at that. Here too we should resist temptation. Additional examples of Moore-paradoxicality are provided by: ³ I am making the assumption that there is some feature, other than a disjunctive one whose disjuncts are not significantly connected, that (1) or (2) have in common that deserves to be called Moore-paradoxicality. ⁴ Strictly it is instances of (1) or (2) that are Moore-paradoxical. However, since a statement or thought will be Moore-paradoxical in virtue of instancing a certain schema I will refer to a schema of the appropriate type as Moore-paradoxical.
Andr´e Gallois
167
(3) P, but I am not at all justified in believing that P, (4) P, but I do not know that P, (5) P, but it might not be the case that P, and by; (6) P, but it is not at all likely that P. Despite this proliferation we may well think that we can discern the nature of Moore-paradoxicality even if we restrict our attention to (1) and (2). As a number have pointed out, Moore-paradoxicality seems to arise in the most basic way at the level of thought. When we ask what makes thinking any of (1) or (2) Moore paradoxical many have found the following answer attractive. Consider (1). Belief plausibly distributes over conjunction. That is, believing both P and Q implies believing P and believing Q. Some hold that, at any rate, conscious belief iterates. That is, consciously believing P implies believing that one believes P. In addition, suppose that the following rationality condition holds for belief. Believing that P implies not believing that not-P. If distribution, iteration, and the rationality condition hold for conscious belief then it is impossible for there to be a belief in an instance of (1) that is consciously held.⁵ What, on this view, makes (1) Moore-paradoxical is that a conscious belief in (1) is ruled out. Advocates of this account of the Moore-paradoxicality of (1) allow that the principles dictating the impossibility of consciously believing (1) do not rule out consciously believing (2). What those principles imply is that a conscious believer of (2) has inconsistent beliefs. By distribution, consciously believing P, but I believe Not-P implies consciously believing P and believing that one believes Not-P. In turn, consciously believing P implies believing that one believes that P. So, one who consciously believes P has the belief that she believes P as well as the belief that she does not believe P. Call an account of Moore-paradoxicality along these lines the propositional attitude account.⁶ We may reasonably hope to extend the propositional attitude account to cover (3)–(6). Clearly, (3) and (4) implicate belief.⁷ Moreover, on ⁵ The argument for this being so goes as follows. Suppose you consciously believe some instance of (1). Say you believe both that Moore is a philosopher and that you do not believe that Moore is a philosopher. By distribution you consciously believe that Moore is a philosopher. So by iteration you believe that you believe that Moore is a philosopher. But by distribution you also believe that you do not believe that Moore is a philosopher. So by the rationality condition you do not believe that you believe that Moore is a philosopher. ⁶ The above version of the propositional attitude account is given, without endorsement, by Roy Sorenson in ‘Moore’s Problem with Iterated Belief ’, Philosophical Quarterly, 50/198 ( Jan. 1998), 28–43. Somewhat different versions of the same account are defended by Sidney Shoemaker in a number of papers reprinted in his collection The First Person Perspective and Other Essays (Cambridge: Cambridge University Press, 1996), and Uriah Kriegel in ‘Moore’s Paradox and the Structure of Conscious Belief ’, Erkenntnis, 61/1 (2004), 99–121. ⁷ That said, how to extend the propositional attitude account to cover, for example, (4) is by no means obvious. Consciously believing (4) does not imply believing that one does not believe
168
Consciousness, Reasons, Moore’s Paradox
one natural reading (5) and (6) may be taken to do so as well. After all the modality in (5) is plausibly epistemic, as is arguably the probabilistic estimate implied by (6). Despite that, the propositional attitude account should be rejected as a comprehensive account of Moore-paradoxicality. As Moore in effect pointed out, there are examples of Moore paradoxicality that do not, directly or indirectly, involve the denial or attribution of a propositional attitude. For example there is as much reason to hold that: (7) ‘Snow is white’ does not mean that: snow is white, is Moore-Paradoxical as there is to so regard any of (1)–(6) as being so.⁸ Moreover, if (7) is Moore paradoxical, so is: (8) Napoleon is not called ‘Napoleon’. What then make all of (1)–(8) Moore-paradoxical? Suppose we agree that beliefs, sentences, statements, and singular terms are, in some suitably broad sense, representations, and, in that light ask what, if anything (1) to (8) have in common relevant to characterizing Moore-paradoxicality. Compare the following instances of (1): (1∗ ) Syracuse is in New York, but I do not believe that Syracuse is in New York and: (2∗ ) I believe that Syracuse is not in New York, but Syracuse is in New York, with the Moore-paradoxical: (7) ‘Snow is white’ does not mean that: snow is white, and; (8) Napoleon is not called ‘Napoleon’. that P. Tim Williamson in Knowledge and Its Limits (Oxford: Oxford University Press, 2000) has attempted to account for the Moore paradoxicality of assertions of instances of (4) by appealing to a constitutive rule governing assertion. Since Williamson’s account rests on such a rule it is not obvious how it would apply to the Moore paradoxicality of thinking an instance of (4). ⁸ In ‘Russell’s ‘‘Theory of Descriptions’’ ’, Moore does not initially give instances of any of (1) or (2) to illustrate what has come to be known as Moore paradoxicality. Instead he gives as an opening example ‘The sentence ‘At least one person is the King of France’ means that at least one person is the king of France.’ He observes that it would be a mistake to think that the last mentioned sentence is a tautology. Later, using ‘Z’ to stand in for the sentence ‘At least one person is the King of France’ Moore remarks ‘Hence if we were to assert ‘‘Z does not mean that at least one person is the King of France’’ we should imply that Z can be properly used to mean what, on the second occasion on which we are using it, we are using it to mean’. He continues ‘To make our assertion by the use of this language is consequently absurd for the same reason [my italics] for which it is absurd to say such a thing as ‘‘I believe he has gone out, but he has not’’ ’. Moore, ‘‘Russell’s ‘Theory of Descriptions’’ ’, 173–5.
Andr´e Gallois
169
Consider sentence (7). Sentence (7) contains the sentence ‘Snow is white’ mentioned on the left, and used on the right. Sentence (7) says something true only if, in virtue of having a meaning, it represents something being so. Moreover sentence (7) represents something being so only if any, used rather than mentioned, sentence it contains represents something being so. No sentence that lacks a meaning represents something being so. Hence, (7) represents something to be so which, if it is so, fails to represent anything to be so. Sentence (8) conforms to the same template. Sentence (8) represents something to be so only if the name ‘Napoleon’ represents someone that sentence (8) says it does not represent. Now consider (1∗ ) and (2∗ ). Taking thoughts to represent things being so we can say this about (1∗ ) and (2∗ ). A thought that (1∗ ) represents Syracuse as being in New York, but also represents the thinker of that thought as not representing, in thought, Syracuse as being in New York. Likewise, a thought that (2∗ ) represents Syracuse as being in New York, but also represents its thinker as having a thought that represents the opposite. This suggests an account of Moore-paradoxicality which may be called the representational account. Applied to our initial examples of Moore-paradoxicality, (1) and (2), the representational account tells us that thinking (1) or (2) implies having thoughts that represent in conflicting ways. For example thinking: (1) P, but I do not believe that P, implies having a thought that represents P being so, and also having the thought that one does not so represent P. Asserting (7) by uttering sentence (7) requires using a linguistic representation that, according to (7), fails to represent. Asserting (8) by means of uttering sentence (8) likewise requires using a name to represent an individual who, according to (8) is not so represented. No doubt the representational account could be more precisely elaborated. I shall not attempt to do so since it is, at best, unclear how it would apply to the following example of Moore-paradoxicality which plays a major role in the remainder of the paper: (9) P, but I only non-consciously believe that P. Consider this instance of (9): (9∗ ) Syracuse is in New York, but I only non-consciously believe that Syracuse is in New York. Compare (9∗ ) with the following paradigm of Moore paradoxicality: (1∗ ) Syracuse is in New York, but I do not believe that Syracuse is in New York. Here is an undefended, but, it seems to me, compelling intuition. Whatever renders stating or thinking (9∗ ) unacceptable is the same feature that renders
170
Consciousness, Reasons, Moore’s Paradox
stating or thinking (1∗ ) unacceptable. If that is right, we have reason to classify (9∗ ) with (1∗ ) as Moore-paradoxical. The Moore-paradoxicality of (9∗ ) is not easily accommodated by the representational account. One who thinks (9∗ ) has a thought that represents Syracuse as being in New York. The problem with applying the representational account to (9∗ ) is this. (9∗ ) does not imply that the same thinker either does not entertain a thought representing Syracuse as being in New York, or entertains a thought representing Syracuse as not being in New York. On the contrary, (9∗ ) only implies that a thinker of (9∗ ) does entertain a thought, albeit a non-conscious one, representing Syracuse as being in New York. We might try to extend the representational account to cover (9∗ ) by invoking the following argument. One who thinks that (9∗ ) not only represents Syracuse as being in New York, but also consciously represents Syracuse as being there. Hence, one who thinks that (9∗ ) consciously represents that Syracuse is in New York while denying that she is so representing the location of Syracuse. This brings us back to the propositional attitude account of Mooreparadoxicality. According to the version of that account we looked at earlier, consciously believing (1∗ ) is precluded by distribution: belief distributes over conjunction; iteration: believing that P implies believing that one believes P, and the rationality condition: believing that not-P implies not believing that P. Whatever one thinks of the plausibility of the iteration and rationality conditions, they do not exclude believing (9∗ ). One who believes (9∗ ) does believe that she believes that Syracuse is in New York. So, there is no prospect of deriving from consciously believing (9∗ ) that one who does so fails to believe that she believes Syracuse is in New York. If we dispense with the rationality condition but retain distribution and iteration, we can at least show that consciously believing (1∗ ) implies having inconsistent beliefs⁹. Can we likewise show, by invoking only distribution and iteration, that consciously believing (9∗ ) implies having inconsistent beliefs? It all depends on the version of iteration we employ. There are the following options: I1: Consciously believing that P implies believing that one believes that P¹⁰. I2: Consciously believing that P implies consciously believing that one believes that P. ⁹ The argument goes as follows. By distribution consciously believing (1∗ ) implies believing that Syracuse is in New York, and believing that one does not believe that Syracuse is in New York. By iteration believing that Syracuse is in New York implies believing that one believes that Syracuse is in New York. ¹⁰ This is the strongest version of iteration that so-called higher order thought (HOT) theories of consciousness are typically committed to.
Andr´e Gallois
171
I3: Consciously believing that P implies believing that one consciously believes that P. I4: Consciously believing that P implies consciously believing that one consciously believes that P. Combining distribution with I1 fails to result in consciously believing (9∗ ) implying having inconsistent beliefs. Applying distribution and I1 to (9∗ ) will only yield the following. Consciously believing (9∗ ) implies believing that one believes that Syracuse is in New York, and believes that one only non-consciously believes that Syracuse is in New York. Nothing is inconsistent about that combination of beliefs. The same goes for I2. We can, at most, extract from distribution and I2 that consciously believing (9∗ ) implies consciously believing that one believes that Syracuse is in New York, and consciously believes that one only non-consciously believes that Syracuse is in New York. Again there is no inconsistency of belief. I3 and I4 can be used to show that consciously believing (9∗ ) implies having inconsistent beliefs. The problem with using I3 to explain what is wrong with consciously believing the Moore-paradoxical (9∗ ) is more subtle. I3 would serve that purpose if it could be used to show that consciously believing (9∗ ) implies having an irrational belief. It cannot. The problem is this. What we want is an account of Moore-paradoxicality that shows why it is problematic to consciously endorse (9∗ )¹¹. Let us ask: what follows about consciously believing (9∗ ) from distribution and I3? Just this. Consciously believing (9∗ ) implies having the possibly non-conscious belief that you consciously believe that Syracuse is in New York together with the conscious belief that you only non-consciously believe that Syracuse is in New York. Now let us pose a further question. If distribution and I3 are both true, and you consciously believe that (9∗ ), then, if you are fully rational, what else can you be brought to acknowledge? Since you consciously believe that you nonconsciously believe Syracuse is in New York, you can be brought to acknowledge that you non-consciously believe that Syracuse is in New York. Hence, you can be brought to acknowledge that you do not consciously believe that Syracuse is in New York. You consciously believe that Syracuse is in New York. It follows from I3 that you, it may be non-consciously, believe that you consciously believe that Syracuse ¹¹ Earlier I distinguished an account of what, for example, it is to have a Moore paradoxical thought from what, in those cases where it is wrong, is wrong with having a Moore paradoxical thought. We need to draw such a distinction in the case of even the paradigm of Moore paradoxicality: P, and I do not believe that P. As we saw there are instances of that schema, such as the eliminativist’s: there are no beliefs, and I do not believe that there are no beliefs, that do not result in any defective thought. In contrast, there are, so far as I can see, no non-defective instances of: P, but I only non-consciously believe that P.
172
Consciousness, Reasons, Moore’s Paradox
is in New York. Does it follow that, if you are rational, you can be brought to acknowledge that you have that conscious belief ? That would only follow if you consciously believe that you consciously believe that Syracuse is in New York. Unless, on pain of irrationality, you can be brought to acknowledge that you do consciously believe that Syracuse is in New York, there is no obvious reason why it would be irrational to have the inconsistent beliefs implicit, according to distribution and I3, in consciously believing (9∗ ). What we need for conscious belief in (9∗ ) to incur not only inconsistent beliefs, but irrationality is I4. The trouble is that I4 is implausibly strong. The obvious problem with it is the infinite regress of beliefs it generates. No doubt a resourceful defender of the propositional attitude account would have responses to the above objections.¹² I will not attempt to anticipate those responses. My disquiet with the propositional attitude account has a different source. It is this. Suppose a believer Sam comes to consciously believe: (9∗ ) Syracuse is in New York, but I only non-consciously believe that Syracuse is in New York. We who are not prepared to endorse instances of: (9) P, but I only non-consciously believe that P, are in a position to recognize that Sam does consciously believe that (9∗ ). Hence, we are in a position to recognize that Sam does consciously believe that Syracuse is in New York. So, assuming I3, we are in a position to recognize that Sam has inconsistent beliefs. How, using the propositional attitude account, might we show Sam that he has inconsistent beliefs? Obviously we begin by pointing out to Sam that, whether or not he consciously believes that Syracuse is in New York, he at any rate consciously believes (9∗ ). Having secured that concession, it should, using the propositional attitude account, be an easy matter to persuade Sam that he consciously believes that Syracuse is in New York. The difficulty is that if Sam is willing to endorse an instance of (9) such as (9∗ ), he will have no trouble endorsing the following instance of (9): (9∗ ), but I only non-consciously believe that (9∗ ). Perhaps, on pain of irrationality, Sam can be brought to acknowledge that if he consciously believes (9∗ ), he has inconsistent beliefs. That is no help in showing Sam what is wrong with (9∗ ), unless he can be brought to acknowledge that he does consciously believe that (9∗ ). Again, unless we have recourse to I4, it is hard to see how the propositional attitude account can be used to show Sam the error of his ways. I am suggesting that the propositional attitude account is wanting in that it cannot be used to show someone like Sam who finds nothing amiss in ¹² For example, we might try weakening I4 to: consciously believing P implies being in a position to consciously believe that one consciously believes P.
Andr´e Gallois
173
consciously believing (9∗ ) that there is something amiss.¹³ In doing so I am implicitly imposing what may be called a first-person adequacy condition on an account of the Moore-paradoxicality of: (9) P, but I only non-consciously believe that P. According to that condition an account of the Moore-paradoxicality of (9) should enable anyone who consciously believes an instance of (9) such as (9∗ ) to identify what has gone wrong.¹⁴ I take it to be a virtue of the account of the Mooreparadoxicality of (9) given in the closing section that it satisfies the first-person adequacy condition. Whether or not that is a virtue, the account will, hopefully, have a further one. Developing it will throw light on the relation between two of the most significant features of our mental lives: being conscious and having reasons.
2 . AT T I T U D E A N D C O N T E N T- R E L AT E D R E A S O N S Suppose, you are in the power of a tyrant who controls an efficient team of mind readers called the thought inspectors. You will be horribly tortured unless you come to believe that the tyrant discovered Neptune. If you do not acquire that manifestly false belief, the thought inspectors will, almost certainly, detect your failure to do so, and act to your great detriment. Do you have a reason to believe the tyrant discovered Neptune? Certainly, in one sense you have every reason to self-induce that belief. You have every reason to resort to hypnotism, brainwashing, reading bad works, or whatever else it takes to get yourself to believe the tyrant discovered Neptune. In another sense you have no reason to ¹³ It is important to distinguish between Sam finding something amiss in consciously believing (9∗ ), and finding something amiss with consciously believing (9∗ ). Sam consciously believes (9∗ ). As noted, the propositional attitude account can be used to show Sam what is wrong with consciously believing (9∗ ). So, we may assume he accepts that there is something wrong with consciously believing (9∗ ). Since Sam does consciously believe (9∗ ) he ought, if he is rational, to concede that there is something amiss. Since it is open to Sam to, at least, remain agnostic about whether he consciously believes (9∗ ), the propositional attitude account is impotent to explain why, in consciously believing (9∗ ), Sam ought to find something amiss. ¹⁴ We might ask: gone wrong with what? The difficulty of answering that question is, in my view, symptomatic of something that lies at the heart of Moore-paradoxicality. Someone who consciously believes (9) ought to find something amiss with doing so. Suppose we ask such a believer what is amiss. There seem to be two candidate answers. The first is that (9∗ ) is true. Clearly, that will not do. There is nothing, in the relevant sense, amiss with Syracuse being in New York, and the believer failing to believe that it is. At first sight, more promising is the answer that the believer consciously believes that (9∗ ). We have already seen, in effect, why that answer will not do. Suppose our believer in (9∗ ) ought, on pain of irrationality, to find something amiss with consciously believing (9∗ ), but does not. In that case, there should be something we can call Sam’s attention to which will alert her to what has gone wrong. Without strain Sam consciously believes that Syracuse is in New York without conceding that she does. So she will find nothing difficult about consciously believing (9∗ ) without conceding that she does. Hence, we cannot alert Sam to what has gone wrong by pointing out that she consciously believes (9∗ )
174
Consciousness, Reasons, Moore’s Paradox
believe the tyrant discovered Neptune. It is perfectly acceptable for you to report on your situation by saying something like the following. Even though I have no reason to believe the tyrant discovered Neptune, I have every reason to bring it about that I hold that belief.¹⁵ In this case there seems to be a straightforward way of marking the relevant distinction. In the above scenario you have no reason that bears on its truth to acquire the belief that the tyrant discovered Neptune. The only reason you have for so believing is one that bears on your self-interest. However, that way of marking the distinction will not generalize to the next case. For a variety of reasons it is greatly in your interest to join the society of bottle top collectors. Unfortunately, you regard collecting bottle tops as a deeply irrational activity, and the society, which includes among its members some thought inspectors from the previous example, will admit only those who have the desire to perform that activity. You can, it seems, correctly say the following. I have no reason to want to collect bottle tops even though I have every reason to cultivate the desire to do so. As in the previous case, in one sense, you have every reason to have an attitude towards a proposition P which, in another sense, you have no reason to have towards the same proposition. However, in this case we cannot mark the contrast as the contrast between having a reason to acquire a desire that bears on self-interest rather than one that bears on the truth of that desire. Desires cannot be true or false. Here is a final example, familiar from the literature, illustrating the distinction between attitude and content-related reasons. You have a strong incentive to intend to perform some act even though you have an equally strong incentive not to carry out your intention. In the best known case, you have every reason to intend to ingest a toxin that you have every reason not to ingest.¹⁶ Placed in this predicament, you may correctly say the following. I have no reason to intend to take the toxin even though I have every reason to form the intention to do so. We have before us a distinction between two senses in which someone can have reason to adopt a propositional attitude. So far, I have attempted to illustrate rather than explain the distinction. When it comes to explaining the distinction between attitude and content-related reasons, there are different ways of doing so. One has already been implicitly introduced. A content-related reason for believing, wanting, or intending that P is a reason for believing, wanting, or intending that P is so. An attitude-related reason is a reason for bringing it about that one is in a certain attitudinal state. An attitude-related reason for believing,
¹⁵ Pascal’s Wager arguably illustrates the distinction being drawn here. Pascal’s Wager may give a reason to get oneself to believe in the existence of God. It gives no reason to believe in the existence of God. ¹⁶ This example is due to G.Kavka, ‘The Toxin Puzzle’, Analysis, 43/(1983), 33–6.
Andr´e Gallois
175
wanting, or intending that P is a reason for bringing it about that one believes, wants, or intends that P. Here is a second way of articulating the distinction between attitude and content-related reasons. Consider belief and desire. A content-related reason for believing or desiring that P counts in favor of P being believable or desirable. An attitude-related reason for believing or desiring that P contributes not at all to P being believable or desirable. In the above examples the attitude-related reasons for believing the tyrant discovered Neptune and wanting to collect bottle tops contribute not at all to the believability of the tyrant having discovered Neptune, or the desirability of bottle top collecting. Generalizing, in contrast to an attitude-related reason, a content-related reason for believing, desiring, or intending that P indicates that P merits belief, desire, or intention.¹⁷ I have a preference for a third way of drawing the distinction.¹⁸ So far the distinction between attitude and content-related reasons has been presented as a distinction between two kinds of reason. That is a natural, but not inevitable, way of taking it. Consider the case of believing that the tyrant discovered Neptune. We could say the individual in that case has one kind of reason, an attitude-related one, for believing the tyrant discovered Neptune, but not another kind, a content related one, for holding that belief. Alternatively, rather than acknowledging two kinds of reason, we could say this. She has no reason at all for believing that the tyrant discovered Neptune. Instead she has a reason for wanting to believe that the tyrant discovered Neptune. If we construe the distinction between content and attitude-related reasons in this way, it transmutes into a distinction between reasons for having higher as opposed to lower-order attitudes. What looks like a different kind of reason for having, say, a first-order attitude is really the only kind of reason there is for having a different kind of second-order attitude. When the time comes to make use of the content-attitude related distinction in an account of Moore-paradoxicality I will presuppose this last way of taking that distinction.
3. CONSCIOUSNESS AND REASONS Consider Fred who only non-consciously believes that he is unmusical. Suppose that, despite having that unconscious belief, Fred should pursue a musical ¹⁷ I owe this way of putting the distinction to Eve Garrard. ¹⁸ This way of drawing the distinction is adopted by Derek Parfit in ‘Rationality and Reasons’ in Dan Egonsson, Bjorn Peterson, Jonas Josefsson and Toni Ronnow-Rasmussen (eds.), Exploring Practical Philosophy: From Action to Values (Aldershot: Ashgate, 2001): 17–41, and by Alan Gibbard Wise Choices, Apt Feelings: A Theory of Normative Judgment (Cambridge, Mass.: Harvard University Press, 1990). 37
176
Consciousness, Reasons, Moore’s Paradox
career, and consciously believes that he should. Suspecting that there is an unconscious impediment to his effectively pursuing his chosen career, Fred visits a psychoanalyst. As a result he is persuaded that he does have an unconscious belief in his unmusicality. Fred comes to consciously believe that he has the unconscious belief that he is unmusical. Here is an asymmetry between Fred’s unconscious belief and his conscious belief that he ought to be a musician. Without it becoming conscious, Fred is able to consciously self-attribute his belief that he is unmusical. Since that is so, Fred can take his unconsciously believing himself to be unmusical as a reason for believing, wanting, or acting. For example, Fred may realize that his unconscious belief that he is unmusical inhibits him from properly exploiting his musical talent. In that case, Fred unconsciously believing that he is unmusical may give him a reason to persist with the therapy revealing he has that belief. Now consider Fred’s consciously held belief that he should become a musician. As we have seen, Fred may take the fact that he unconsciously believes himself to be unmusical to be a reason to attempt to divest himself of that belief. Fred may, likewise, take the fact that he believes he should become a musician as a reason for acting. He may, for example, take his believing that he should become a musician as a reason for not discussing his future career with his parents who very much want him to become a lawyer. So far, no asymmetry has emerged between the way in which a consciously, as opposed to a non-consciously, held belief can feature in a reason for acting or adopting a propositional attitude. An asymmetry emerges when we consider the following. Fred is able to take the fact that he believes he should become a musician to be a reason for acting. Since he holds that belief consciously, he is also able to take the fact that he should become a musician as a reason for acting. Fred is also able to take the fact that he believes he is unmusical as a reason for acting. Since he holds that belief only unconsciously, he is not able to take the fact that that he is unmusical as a reason for acting. Consciously holding the belief that P enables one to treat the fact that P, rather than just the fact that one believes P, as a reason for acting, or adopting further propositional attitudes. Only non-consciously holding the belief that P disables one from taking that P as a reason. What explains this asymmetry? One candidate explanation can be immediately discounted. Let us say that if one believes that P, and is able to take the fact that P as a reason, then the belief that P is reason-giving. Suppose we grant that a belief is reason-giving only if one is aware of holding it. It may be suggested that a non-conscious belief cannot be reason-giving because the one who holds it is not aware of doing so. We need to draw a distinction between being conscious of holding a belief, and that belief being consciously held. In the case described above, Fred, as a result of visiting his psychoanalyst, is fully conscious of holding the belief that he is unmusical. Despite that, the belief he is fully conscious of holding is not one
Andr´e Gallois
177
that he consciously holds. So the question remains to be answered: why is it that only a consciously held belief can be reason giving? Here is the principle we are currently examining: C: A believer is in a position to take that P as a reason for acting, or consciously holding a propositional attitude such as believing, wanting, hoping intending etc, only if she consciously believes that P. The question before us is this. What explains the truth of C? One answer to that question has proved unsatisfactory. C is not true because only a belief that an individual is aware of holding can be reason-giving. An initially more promising answer goes like this. Take Fred’s unconscious belief that he is unmusical. He has, we are supposing, every reason to believe that he holds that belief. Nevertheless, he has no reason to believe that he is unmusical. Only a belief held for a reason can, itself, be reason-giving. Moreover unconscious beliefs are, typically, not held for reasons. So, non-conscious beliefs are, typically, not reason-giving. That is why C is true. At most this shows that an unconscious belief, even if consciously selfattributed, is unlikely to be reason-giving. It fails to show that such a belief cannot be reason-giving. We are still left with the question: why cannot a non-consciously held belief be reason-giving? In brief the answer is that only a consciously held belief can engage in the right way with other consciously held propositional attitudes, such as desires, to deliver a reason for adopting those attitudes. To see how only consciously held belief can engage with a consciously held desire to yield a reason consider a type of case familiar from the literature on belief and the will. Sam has a very strong desire to believe that God exists. As he sees it he has compelling truth-related reasons to believe there is no God. As a result he becomes severely depressed. So depressed that he has an overriding non-truth-related reason to reacquire his former consciously held belief in the existence of God. Prior to reacquiring that belief, Sam reasons thus. What I most want is that God exists. There is nothing I can do to satisfy that desire. In particular, my coming, once again, to consciously believe in God’s existence will make no difference to his existing. Still, given that I am unable to modify my desire for God to exist, if I do bring myself to consciously believe that God exists, I will no longer be depressed. Sam has a desire he recognizes he cannot satisfy: the desire that God exists. Having that desire, together with the depression resulting from its non-satisfaction, leads him to form a desire he can satisfy: the desire that he consciously believe that God exists. Suppose that, by adopting the relevant Pascalian procedures, Sam brings himself, for what he takes to be good truth-related reasons, to consciously believe that God exists. At that point Sam is asked whether he feels depressed because he lives in a universe without God. From the point of view of alleviating
178
Consciousness, Reasons, Moore’s Paradox
his depression what matters is that, from his perspective, Sam is in a position to give the following answer. I don’t feel in the least depressed about living in a universe without God because I don’t live in such a universe: God exists. Before he succeeds in repossessing his belief in God’s existence, Sam takes his desire for God to exist to be unsatisfiable, but sees that he is able to satisfy a distinct desire to consciously believe that God exists. After repossessing his belief in God’s existence, Sam not only takes his desire to believe that God exists to be satisfied, but, in addition takes his desire for God to exist to be satisfied. Were he not to take the second desire to be satisfied, he would still feel depressed. What the case of Sam illustrates is the distinctive impact on one’s psychology of having a belief consciously. Consciously having the belief that P enables one to integrate P into one’s view of the world in such a way that it being the case that P can have a distinctive impact on one’s mental life. We may label that impact the integrative function of consciously holding a belief. In summary, in virtue of a belief being consciously held it is possible for what is believed to be integrated into the believer’s world view in a way that what is believed in the case of an non-consciously held belief cannot be. In the remainder of the paper I will attempt to do the following. First to spell out in three interconnected ways the integrative function of holding a belief consciously. Second, to say how conscious belief having that function helps to explain the Moore-paradoxicality of: P, but I only non-consciously believe that P. Third, to relate the foregoing to the distinction drawn earlier between attitude and content-related reasons.
4 . C O N S C I O U S B E L I E F : D E L E T I O N A N D I N T E G R AT I O N Consider the following inferences: (A) Necessarily: there is only one even prime. There is only one even prime. (B) It is true that: snow is white. Snow is white. (C) John intends that: he John will leave soon. John will leave soon. Each of (A)–(C) have the following in common. Each one is an inference consisting of a premiss which supports the corresponding conclusion. In the case of (A) and (B) the premiss entails the conclusion. In the case of (C) the premiss does not entail, but makes probable the conclusion. We may say that in the case of each of (A)–(C) the truth of its premiss supports the truth of its conclusion. Here are some further features shared in common by (A)–(C) that I wish to call to attention. Ceteris paribus, belief in the premiss of each of (A)–(C) makes
Andr´e Gallois
179
rational belief in its conclusion. We may say that, in the case of each of (A)–(C), the transition from believing its premiss to believing its conclusion is a defeasibly rational one. Another feature shared in common by (A)–(C) plays a prominent role in the following discussion. Consider (A). Suppose an individual makes what I called the rational transition from believing (A)’s premiss, necessarily: there is an only even prime, to believing its conclusion: there is an only even prime. Making such a transition amounts to deleting the operator expressed by ‘necessarily’ from that premiss. An alternative way to put the point is to say that making the relevant transition amounts to taking a proposition that falls under the scope of the necessity operator in the premiss outside the scope of that operator. The same point applies to (B) and (C). Making the defeasibly rational transition from a belief in the premiss of each one to a belief in its conclusion results in deleting the operator featured in that premiss. For our purposes calling attention to the features shared in common by (A)–(C) has the following relevance. Compare (A)–(C) with: (D) I consciously believe that: G. E. Moore smoked a pipe. G. E. Moore smoked a pipe. (D) shares the following in common with (A)–(C). It is defeasibly rational for anyone who consciously endorses (D)’s premiss to make the transition to endorsing its conclusion. If you accept, as you might put it, that you believe in full consciousness that G. E. Moore smoked a pipe then, in the absence of countervailing considerations, it is rational for you to believe that G. E. Moore did indeed smoke a pipe. The next feature (D) shares in common with (A)–(C) is the crucial one. Making the transition from believing (D)’s premise to believing its conclusion amounts to deleting the associated operator. In this case it amounts to deleting the operator expressed by ‘I consciously believe that’. Again we may put the point in terms of scope. Moving in thought from (D)’s premise to its conclusion places the proposition that G. E. Moore smoked a pipe outside the scope of the operator expressed by ‘I consciously believe that’. Before examining the significance of this point about deletion or scope, we need to note a difference between (D) and each of (A)–(C). I called the move from a belief in any of (A)–(C)’s premisses to a belief in its conclusion a defeasibly rational transition. Why call that move a defeasibly rational transition rather that something more familiar such as a warranted or justified inference? The reason is this. Belief in the premise of any of (A)–(C) does warrant or justify belief in its conclusion. One indication of this is the following. It is rational for one who believes (A)’s premiss to believe its conclusion. In the case of (A) it is also rational to believe the conditional whose antecedent is (A)’s premiss, and whose consequent is (A)’s conclusion. It is rational to believe:
180
Consciousness, Reasons, Moore’s Paradox
(A∗ ) If it is necessary that there is an only even prime then there is an only even prime. Moreover, unless it is rational to believe (A∗ ) it is not rational to move from a belief in (A)’s premiss to a belief in its conclusion. The same goes for (B) and (C). The rationality of moving from a belief in (B) or (C)’s premiss to its conclusion makes rational, in the case of (B), belief in the conditional: (B∗ ) If it is true that snow is white then snow is white, and, in the case of (C), the conditional: (C∗ ) If John intends that he will leave soon then John will leave soon. As with (A) and (A∗ ) the rationality of belief in (B∗ ) and (C∗ ) is required for the corresponding belief transitions to be rational. In this respect (D) contrasts with (A)–(C). Here once again is: (D) I consciously believe that: G.E. Moore smoked a pipe. G. E. Moore smoked a pipe. It is rational for me to move from a belief in (D)’s premiss to a belief in its conclusion even if it is not rational for me to believe: (D∗ ) If I consciously believe that: G. E. Moore smoked a pipe then G. E. Moore smoked a pipe.¹⁹ We might say that the move from a belief in (D)’s premiss to a belief in its conclusion is inference-like. Despite that, and this cannot be overemphasized, that move is not a rational one to make because it is a justified or warranted inference. Here is another way to make the same point that, as we will see, links the rationality of the belief transition represented by (D) to the Moore paradoxicality of: P, but I only non-consciously believe that P. Each of (A)–(C) is an inference that it is rational to make. As we have observed, each one involves deleting an operator from the proposition constituting its premiss. Clearly, in the case of (A)–(C), in order for that deletion to be rationally made the relevant premiss must, at least, be believed. For example, the truth of (A)’s premiss is by itself insufficient to make the deletion of the operator expressed by ‘it is necessary that’ rational. In contrast, it is enough for (D)’s premiss to be true for the deletion of the operator expressed by ‘I consciously believe that’ to be rationally made. What makes the transition from consciously self-attributing a consciously held belief to the belief consciously self-attributed a rational one? In brief the answer is this. The deletion function of consciously held belief enables a believer to integrate reasons and what they are reasons for into a unified view of the world. ¹⁹ Of course, it may be rational for me to believe D∗ . It would be so if I had reason to believe myself authoritative about G. E. Moore’s smoking habits. The point is it need not be rational for me to believe D∗ in order for it to be rational for me to move from D’s premiss to its conclusion.
Andr´e Gallois
181
My next task is to clarify what kind of integration is at issue. I will attempt to do so in two ways. First, by invoking a comparison I have used elsewhere with the role of integration in fiction.²⁰ Second, and more importantly, by reintroducing the distinction drawn earlier between two ways of attributing a reason.
5 . I N T E G R AT I O N I N F I C T I O N Part of a fictional narrative may consist in some character within the story relating a fictional narrative: a story within a story. Such a story may unfold like this. The original storyteller, it may be an actual author, tells a fictional tale in which a group of individuals gather to have a beer together, and one of them tells a story. Call the story told within the original story the embedded story. Suppose the embedded story begins like this. A traveler journeyed to a distant land, and there met an old man who told her fortune. So long as it remains within the original story, as they say, just a story, the embedded story fails to be integrated with the original. There are a number of ways to signal this failure of integration. The most obvious is this. Suppose we are asked to specify what would have to be the case for the original story to be true. We would need to mention that a group of individuals gathered together to have a beer. We would not need to mention that a traveler journeyed to a distant land. What we would need to mention is that one of those having a beer told a story about such a traveler. Another indicator of the relevant failure of integration is this. Events in the original and embedded stories do not bear on one another in a way that they do in a pair of stories that, to anticipate, present an integrated fictional world. One of the protagonists in Charlotte Brontë’s Jane Eyre is Rochester’s wife. Jean Rhys’s Wide Sargasso Sea is what these days would be called a prequel to Jane Eyre. In the former novel, Rochester’s wife’s early life in the West Indies is narrated in part to explain the way she reacts in Bronte’s narrative. Despite being by different authors, the events in The Wide Sargasso Sea do bear on the events in Jane Eyre in a way that the events in the embedded and original stories do not bear on each other. An example of this is given above. Events narrated in The Wide Sargasso Sea may be invoked to explain events narrated in Jane Eyre, not just by the readers of those works, but also by their participants. For example, though it did not, reflecting on his wife’s earlier life as depicted in Jean Rhys’s novel could have given Rochester explanatory insight into his wife’s behaviour as the events in the embedded story could not provide explanatory insight into the behaviour of the individuals listening to it. Again, the events in Jean Rhys’s novel can provide the protagonists in Jane Eyre with ²⁰ In Gallois, ‘First-Person Accessibility and Consciousness’, Philosophical Topics, 28 (2000)/2 101–24.
182
Consciousness, Reasons, Moore’s Paradox
reasons for, say, acting or believing. Events in the embedded story cannot provide the protagonists in the original story with like reasons to act or believe. The next indication of integration failure relates directly to what was said in the last section about conscious belief. In the narration of the original story use is made of an operator: one, as it may be, expressed by ‘one of the individuals drinking beer began to tell the following story’. Call an operator of this type the story operator. Events in the embedded story fail to be integrated with those in the original because the former events fall within the scope of the story operator. Taking those events outside the scope of that operator has the effect of integrating them into the original story. For example, if one of those in the original story, without entering any qualification, remarks that the traveler in the embedded story has just entered the room, that has the effect of making that traveler part of the same fictional world as the world of the original story. In the case of a fictional narrative we have identified signs or marks of integration. A fictional work F1 fails to be integrated with a fictional work F2 if what would have to be the case for F1 to be literally true need not be mentioned in specifying what would have to be the case for F2 to be true. F1 fails to be integrated with F2 if what is the case according to F1 does not bear, in an explanatory or reason-giving fashion, on what is the case for the participants in F2. F1 fails to be integrated with F2 if what is the case according to F1 cannot be placed outside the scope of the story operator in F2. The point of the preceding remarks about integration is to construct a, hopefully illuminating, parallel with the sense in which consciously held beliefs are integrated. Of the three marks of integration the last two apply to conscious belief. Consider the case mentioned at the beginning of Section 3. Fred consciously believes he should become a musician, but only unconsciously believes he is unmusical. From Fred’s point of view, that he should become a musician bears on what, for example, he has reason to do or think. In contrast, that he is unmusical can have no bearing on what he should do or think, unless his belief that he is unmusical becomes conscious.²¹ The third mark of integration concerns scope. If pair of fictional works are integrated then what is the case according to one will fall outside the scope of the story operator in the other. Fred is, likewise, in a position to place that he should become a musician outside the scope of a belief operator as he is not able to place his being unmusical outside the scope of that operator.
6 . R E A S O N S , C O N S C I O U S N E S S , A N D I N T E G R AT I O N It is now time to return to the distinction between attitude and content related reasons. Consider the case I used to introduce the distinction in Section 2. The ²¹ Again, Fred can take that he believes he is unmusical as a reason for acting, but his belief that he believes he is unmusical is one he consciously holds.
Andr´e Gallois
183
tyrant will torture you unless you believe he discovered Neptune. In that case, you have an attitude- but no content-related reason to believe he discovered Neptune. Now consider a belief which you do have a content-related reason for holding. You have a content-related reason for believing that the tyrant did not discover Neptune. One content-related reason you have for holding that belief is, say, that the tyrant was not born when Neptune was discovered. Here is a striking feature of a content-related reason which differentiates it from its attitude-related counterpart. The belief for which you hold a content related reason is the belief that: (i) The tyrant did not discover Neptune. One content related-reason you have for believing (i) is: (ii) The tyrant was not born when Neptune was discovered. In addition, you have an attitude-related reason for believing: (iii) The tyrant did discover Neptune, which is: (iv) You will be horribly tortured unless you adopt the belief that (iii). (ii) specifies the content of a belief that gives you a reason for believing (i). (iv) likewise specifies the content of a belief that gives you a reason for believing (iii). One of these reasons is content-related, and the other only attitude-related. What makes for this difference? Consider the pair of beliefs at issue in what we may call the tyrant case: (Bi) The belief that the tyrant did not discover Neptune. (Biii) The belief that the tyrant did discover Neptune. Your reason for adopting (Bi) is that: (ii) The tyrant was not born when Neptune was discovered. and for adopting (Biii) that: (iv) You will be horribly tortured, unless you adopt (Biii). Intuitively there is a relation between (ii) and (Bi) which does not hold between (iv) and (Biii). Making use of the terminology used in the last section to characterize one aspect of integration in fiction, (ii) bears on what (Bi) is about in a way that (iv) does not bear on what (Biii) is about. We may say that (ii) bears, in a way (iv) does not, on a question that (i) and (iii) supply competing answers to. (ii), unlike (iv), bears on whether the tyrant did discover Neptune. How should the distinctive connection between a content-related reason and the content of the belief it rationalizes be understood? One answer is clearly inadequate. It goes like this. (ii) is, we are supposing, your reason for having: (Bi) The belief that the tyrant did not discover Neptune,
184
Consciousness, Reasons, Moore’s Paradox
and, we may also suppose, you would have no other reason to hold (Bi) if you had not believed (ii). If so, your having a belief with (Bi)’s content is counterfactually dependent on your believing (ii). We may be inclined to say that the distinctive relation that holds between a reason Q and the belief that P is content-related if one would not have had a belief with that content, unless one had believed Q.²² Of course, the trouble with this suggestion is that you would not have had (Biii) if you had not believed (iv). There is no obvious relation of counterfactual dependence that distinctively holds between a content-related reason and the belief it is a reason for having. What then is the best way to understand the link between a content-related reason and the belief it rationalizes? The answer I favor is best given against two background assumptions whose defence must be left to another time and place. Consider the following pair of claims: (1) That it has rained recently is a reason for it being true that the ground is wet. (2) That Jones’s light is on is a (content-related) reason for Sam believing that Jones is at home. The first assumption is that the expression ‘is a reason for’ in sentences (1) and (2) attributes the same relation that holds between one thing being the case, or being true, and something else being the case or being true. The second assumption is the crucial one. In the case of (1) the reason-giving relation holds between it having rained recently and the ground being wet. In the case of (2) the same reason-giving relation does not hold between Jones’s light being on and Sam believing that Jones is at home. Instead, it holds between Jones’s light being on and Jones being at home. Here is an all too brief sketch of how one might employ the machinery of possible worlds to fill in this basic picture. In the case of (1) the reason-giving relation holds between something true at the actual world, it raining recently, and something else, the ground being wet, also true at the actual world. In the case of (2) the same reason-giving relation holds between something true at the actual world, Jones’s light being on, and something, Jones being at home, holding true at the members of some set of worlds. But the worlds in this case are Sam’s belief-worlds. We may put it like this. If (1) is true, it having recently rained is a reason for it being true in the actual world that the ground is wet. If (2) is true, Jones’s light being on is a reason for it being true in the world as Sam believes it to be that Jones is at home. Content-related reasons for other propositional attitudes behave in the same way. Sam, let us say, has the following content-related reason for going for a ²² An obvious objection to this proposal arises from the possibility of overdetermination of content-related reasons. You have a sufficient content-related reason, (ii), for (Bi). Suppose, you have an additional content-related reason which is also sufficient for (Bi). In that case, it may be false that had you not believed (ii), you would not have had (Bi).
Andr´e Gallois
185
walk: it is a fine day. In that case, it being a fine day is a reason for it being true in the world as Sam wants it to be that he goes for a walk. Seen in this light in the tyrant case we say you have the attitude-related reason: (iv) You will be horribly tortured, unless you adopt (Biii) for adopting: (Biii) The belief that the tyrant did discover Neptune because (iv) being true is a reason for it being true in the world as you want it to be that you believe the tyrant discovered Neptune. (iv) is not a reason for it being true in the world as you either want or believe it to be that the tyrant did discover Neptune. That is why (iv) provides an attitude- but no content-related reason for (Biii).
7. CONSCIOUSNESS, REASONS, AND MOORE’S PA R A D OX We have covered a good deal of territory since taking note of the Mooreparadoxicality of instances of: (9) P, but I only non-consciously believe that P. The representational account of Moore-paradoxicality was found wanting in that it provides no explanation of the Moore-paradoxicality of (9). My aim is to provide, at least, the beginnings of such an account by utilizing as its components what has been covered in the previous five sections. It is now time to review those components, and say how they bear on the Moore-paradoxicality of (9). In Section 2 we looked at two ways, attitude and content-related, in which a reason can be a reason for believing. In the last section we noted that, as so labeling them suggests, content-related reasons have a distinctive connection to the contents of the beliefs they rationalize. In that section an attempt was made to articulate that connection. The relation of being a reason for holds between contents. In the case we are concerned with it holds between belief-contents. If Q is a content-related reason for having the belief that P, that is a matter of Q being true being a reason for P being true. Why not then simply say that Q being true is a reason for P being true without making reference to belief? Because the reference to belief indicated the standpoint from which Q being true is a reason for P being true. Q being true is a reason for P being true from the standpoint of belief rather than desire, hope, fear, intention, or, for that matter, actuality. Section 3 explored the relation between consciousness and reasons. There it was claimed that only a consciously held belief can supply a reason. That observation is linked to the topic of Section 4: the deletion and integration functions of consciously held belief. One may consciously self-attribute both
186
Consciousness, Reasons, Moore’s Paradox
consciously and non-consciously held beliefs. A striking difference between a consciously and non-consciously held belief is this. In the case where the belief that P is consciously held one is typically able to delete the qualification that P is something one believes. Such self-effacement is intimately connected to a function of conscious belief I called its integration function. The contents of consciously held beliefs can be integrated into a single view of the world. If one consciously believes that P and consciously believes Q then one can delete the qualification that P and Q are believed. As a result the contents of those beliefs are integrated into a view of the world in which both P and Q obtain. Now for the story about the Moore-paradoxicality of (9). To see how that story goes, consider the instance of (9) introduced in the first section: (9∗ ) Syracuse is in New York, but I only non-consciously believe that Syracuse is in New York. Suppose Sam comes to consciously hold (9∗ ). The representational account tells us that the following are consequences of Sam consciously believing (9∗ ): (3) Sam consciously believes that: Syracuse is in New York, and: (4) Sam consciously believes that: he only non-consciously believes that Syracuse is in New York. The task is to say why Sam holding the conscious beliefs attributed by (3) and (4) should be problematic for Sam given that (3) does not imply that Sam consciously believes (3), but, at most, that he believes (3). According to (4) Sam attributes to himself the belief that Syracuse is in New York as one that he non-consciously holds. Since that is so Sam will not be prepared to move from: (5) I (Sam) believe that: Syracuse is in New York, to: (6) Syracuse is in New York. But, Sam does consciously believe (6). So, from Sam’s point of view, (6) falls outside the scope of the belief operator in (5). Despite that, since he is only prepared to self-attribute his belief in (6) as a non-conscious one, Sam is not prepared to move from (5) to (6). He is thus not prepared to delete the belief operator in (5). Hence, in virtue of consciously believing (5) and consciously self-attributing a non-conscious belief in (6), Sam has conflicting dispositions. He is disposed to delete the belief operator in (5). He is also disposed not to delete the belief operator in (5). Can we leave it at that? Can we explain the Moore-Paradoxicality of (9) by simply noting that, in virtue of consciously endorsing (9∗ ), Sam will have the conflicting dispositions in question? To do so would be unsatisfactory since the
Andr´e Gallois
187
resulting account of (9)’s Moore-paradoxicality would fail to meet the first-person constraint mentioned at the end of Section 1. Simply observing that consciously believing (9∗ ) induces conflicting dispositions in the believer leaves out what, from the believer’s first-person perspective, generates that conflict. We can do better by taking into account the integration function of consciously held belief, and its connection to content-related reasons. In consciously believing (9∗ ) our believer self-attributes the belief that Syracuse is in New York as one only non-consciously held. As so attributed the believer cannot place Syracuse being in New York outside the scope of the belief operator. Since the believer cannot take Syracuse being in New York outside the scope of that operator, she cannot integrate Syracuse being in New York into her antecedent view of the world. In particular, she cannot take Syracuse being in New York as a content related reason for acting or having further propositional attitudes, or take any content-related reason as a reason for Syracuse being in New York from the standpoint of belief. Our believer does consciously believe (9∗ ). So Syracuse being in New York is integrated into her overall view of the world. She is in a position to take Syracuse being in New York as a reason, and take content-related reasons as reasons for Syracuse being in New York, again from the standpoint of belief. In this way, consciously believing (9∗ ) generates an unstable view of the world believed in. In virtue of consciously believing (9∗ ), our believer consciously believes that Syracuse is in New York. Since that is so Syracuse being in New York is integrated into the world as the believer takes it to be. Our believer also consciously believes that she only non-consciously believes that Syracuse is in New York. Since that is so Syracuse being in New York fails to be integrated into the world as the believer believes it to be. I will conclude by, once again, using the comparison with integration in a story to throw light on what is going on here. In that case the counterpart to consciously believing (9∗ ) would be this. Suppose we begin reading the original story. In doing so we are presented with a fictional world. That fictional world includes a number of individuals having a conversation while drinking beer in a pub. Suppose one of those individuals announces that David Lewis’s realism about possible worlds has been conclusively established. If so, the existence of such worlds spatio-temporally disconnected from the world of the original story has been integrated with that world. So far no failure of integration. Some stories are inconsistent. Suppose the original story turns out to be inconsistent. Suppose the author of the original story implies that one of the beer drinkers is a teetotaler. Still no failure of integration. The fictional world depicted is one in which some beer drinker is a teetotaler. At this point, one of the characters begins relating the embedded story. Now there is a failure of integration. The embedded story introduces into the original story a fictional world that is not integrated into the world of the original story.
188
Consciousness, Reasons, Moore’s Paradox
The world-embedded story may in various ways become integrated into the world of the original story. The counterpart of consciously believing (9∗ ) would arise if a reader of the original story were invited to both integrate and not integrate the world of the embedded story into the world of the original. Such a story would not, as an inconsistent one is, be a story about the impossible, but an impossible story. In like fashion, consciously believing (9∗ ) leads a believer to tell an impossible story about the actual world.
9 Moorean Absurdity and Showing What’s Within Mitchell Green
1 . M O O R E A N A B S U R D I T Y A N D M O O R E ’ S PA R A D OX Moorean absurdity occurs in any utterance or thought in which an agent overtly expresses an intentional state that she also explicitly disavows; it also occurs in any utterance or thought in which an agent overtly expresses an intentional state whose content is incompatible with that of another intentional state that she also explicitly avows. Her expression-and-disavowal, or expression-and-avowal, might occur in her utterance or thinking of a single sentence such as ‘It’s raining but I don’t believe it.’ On the other hand this expression-and-disavowal, or expressionand-avowal, might occur without the medium of a single sentence. It might occur through the utterance or thinking of two consecutive sentences such as It’s raining. I don’t believe that it is. Alternatively, for all we know about what a conscious mind can do, it might occur through the simultaneous thinking or uttering of the above two displayed sentences. The above disjunctive characterization of Moorean absurdity (as either overtly expressing an intentional state that one also explicitly disavows, or overtly expressing an intentional state whose content is incompatible with that of another intentional state that one also explicitly avows) is not a grammatical characterization. Someone who utters a sequence of sentences such as those
Earlier versions of this paper were presented at the University of Virginia and at Texas A&M University. I thank audiences at both institutions for their insightful comments. Special thanks to John Williams and a referee for Oxford University Press for their illuminating comments on an earlier draft. Research for this paper was supported in part by a Summer Grant from the Vice Provost for Research and Public Service at the University of Virginia. That support is here gratefully acknowledged.
190
Showing What’s Within
displayed above may simply be using her second sentence to express surprise or amazement. Doing so is not at all absurd. Nor is it absurd if the above sequence registers a mid-utterance change of mind. Likewise, as I recover from laryngitis I might test my voice by uttering the above sequence. In this case I don’t mean a word of what I say, and nothing in my performance is absurd. An actor who utters ‘It’s raining but I don’t believe it’ on stage in the course of a play about, say, G. E. Moore, is also not behaving absurdly; nor is he harboring absurd beliefs. This is due to the fact that so long as he is not a method actor, he is not expressing any of his beliefs. Instead, he is portraying a character expressing and avowing beliefs, and thereby portraying a character who is behaving absurdly. That is why the non-absurdity of the actor’s performance is compatible with the above characterization of Moorean absurdity.¹ Similarly, in Wittgenstein’s example the train announcer who says over the loudspeaker, ‘Train number 121 arriving on track #3 in two minutes (personally I don’t believe it),’ is not behaving absurdly. The reason is that he is only saying his first conjunct ex cathedra in the hopes of keeping his job, rather than putting it forth as an expression of his state of mind. That is why the non-absurdity of his announcement is compatible with the above characterization of Moorean absurdity. My account of Moorean absurdity invoked the notion of an overt expression. It may be possible to express an intentional state without doing so overtly. Perhaps when I involuntarily grimace as the paring knife slices into my finger I am expressing pain. I am not, however, overtly doing so. Accordingly, if while thus grimacing as the knife slices flesh I also say, ‘. . . but I’m not in pain,’ I am in error but am not absurd. Likewise, my blushing might express my embarrassment even though I cannot help it. If while blushing I also explicitly deny that I am embarrassed, I am simply wrong rather than absurd. Contrary to what these examples suggest, it is not enough for overtness that my expression be intentional. We will consider what more is required in Section 3 below. Whereas Moorean absurdity is a phenomenon for which we theorists need no apology, Moore’s Paradox is a testament to the ignorance of students of language and thought. For Moore’s Paradox is the apparent tension that arises from the facts that (1) such utterances and thoughts as ‘It is raining but I don’t believe it,’ seem absurd, but (2) their absurdity is not due to a semantic contradiction. An assertion of an explicit contradiction P&-P is absurd, but with the help of standard logical techniques we may explain its absurdity without landing in paradox. Likewise, a self-falsifying utterance such as an utterance of ‘I am not now uttering any words’ is absurd, but the absurdity is not paradoxical because there is no mystery how the utterance of that sentence falsifies itself: We explain the phenomenon with standard logic plus a gentle reference to pragmatics. A slightly ¹ It also shows why Szabo-Gendler (2001) is mistaken to contend that the actor’s performance is absurd. Szabo-Gendler is confusing absurdity in the actor with absurdity in the character portrayed. This mistake undermines the gravamen of his criticism of modal fictionalism.
Mitchell Green
191
fuller reference to pragmatic notions such as is available from double-indexing semantics shows what is absurd about an utterance of ‘I am not here now’. With minimal assumptions about the structure of contexts of utterance plus the behavior of indexicals ‘I’, ‘here’, and ‘now’, we may infer that this sentence expresses a contingent proposition even though in every context in which it is (or could be) thought or uttered it is not true. Once again we account for the absurdity without landing in paradox. Matters are not so simple with cases of Moorean absurdity. ‘It is raining but I don’t believe it,’ and ‘It is raining but I believe it is not,’ not only express propositions that could be true. They are also sentences that could be truly uttered or truly thought.² If I utter ‘It is raining but I don’t believe it,’ what I utter could be true so long as I am not sincere. If I say to myself without uttering any words, ‘It is raining but I don’t believe it,’ what I say to myself could be true so long as I am not sincere. (Below I show that it is possible to say something to oneself in the privacy of one’s own thoughts that one does not also believe.)³ We label rather than explain the phenomenon with such expressions as ‘pragmatic paradox’ or ‘pragmatic absurdity’. In what, then, might we locate the source of Moorean absurdity? My aim in what follows will be to resolve Moore’s Paradox by explaining the source of Moorean absurdity. With apologies to existentialists, the first step shall be to elaborate on the notion of absurdity.
2 . A B S U R D I T Y A N D T H E V I O L AT I O N O F N O R M S Absurdity arises from severe violation of a system of norms.⁴ Some well-known systems of norms are theoretical rationality and practical rationality. We find a severe violation of theoretical rationality in any agent whose system of beliefs is guaranteed to put her in error no matter how the world happens to be, and in a way that she could in principle discern with no empirical investigation. Jane’s belief that Hesperus is shining puts her in error if in fact Venus is not shining, but it is not absurd simply to be in error. So too, Jane’s belief that Hesperus is shining ² Cargile (1967) is to my knowledge the first to observe this. ³ Kriegel (2004) asserts that although Moorean sentences can be true, they cannot be truly asserted and they cannot be truly believed. As we shall see below, only the latter claim is correct. Kriegel’s view that a Moorean sentence cannot be truly asserted seems to rest on his view that an utterance is an assertion only if it is an expression of a belief, that is, only if it is sincere. This view simply stipulates away the possibility of a lie, and will not be assumed here. ⁴ ‘In ordinary life, a situation is absurd when it includes a conspicuous discrepancy between pretension and aspiration or reality: someone gives a complicated speech in support of a motion that has already been passed; a notorious criminal is made president of a major philanthropic foundation; you declare your love over the telephone to a recorded announcement; as you are being knighted, your pants fall down.’ This characterization from Nagel (1979), is more inclusive than the one given in the text, but I will only need the less inclusive characterization in what follows.
192
Showing What’s Within
but Phosphorus is not shining puts her in error no matter how the world turns out to be compatible with metaphysical possibility. However, Jane’s ignorance of the identity of Hesperus and Phosphorus may prevent her from being able to discern her error except by empirical investigation, and if that is so, then her belief that Hesperus is shining but Phosphorus is not, need not be absurd. On the other hand, if Jane believes that Taylor saw Hunter but that Hunter was not seen by Taylor, then not only is she in error no matter what else is true of the world, she can in principle discern her error with no empirical investigation. That is why it is absurd for her to believe that Taylor saw Hunter but that Hunter was not seen by Taylor. (Harboring an absurd belief need not make her irrational; perhaps she is irrational only if she violates theoretical rationality in a way that would be open to minimal, as opposed to extensive, reflection.) A belief or system of beliefs that severely violate the norms of theoretical rationality is absurd even if prudential, moral, or other norms enjoin one to hold it. In that case it would be absurd from the point of view of theoretical rationality, while permissible, and perhaps even mandatory, from the point of view of some other system of norms. Thus suppose that the only way to save the life of a loved one is to take a pill and undergo hypnosis with the result that I come to believe a contradiction. Suppose further that I agree to induce that belief in this way. Then I undertake to do something that is absurd from the point of view of theoretical rationality, but morally mandatory—or at least permissible. Instead of this gaseous ‘point of view’ talk, we could just describe the situation as one in which my coming to believe a contradiction is both absurd (full stop) and morally permissible. Theoretical rationality is not the only system of norms admitting of severe violation. We find a severe violation of practical rationality in one whose system of plans, together with her utilities and subjective probabilities, guarantee that she does not maximize subjective expected utility, and in a way that she could in principle discern with no empirical investigation. Suppose I accept a sure thing of $1 over a wager in which a fair coin is tossed, such that if it comes up heads I win $1,000 and if it comes up tails I gain nothing. This choice guarantees that I do not maximize subjective expected utility, and I could in principle discern this with no empirical investigation. A rival to this classical style of decision theory, prospect theory,⁵ offers a distinct set of norms whose violation takes a different form from that adduced by subjective expected utility theory. We need not settle this rivalry here. Suffice it to say that each theory offers an account of practical rationality, and with it an account of what constitutes a severe violation of that form of practical rationality. No matter the theory, a severe violation of what is by its lights practically rational will be deemed absurd. As with the case of theoretical rationality, we may expect the norms of practical rationality to collide ⁵ See for instance Tversky and Kahneman (1992), and Wakker and Tversky (1993). Green 1999a discusses the philosophical significance of these approaches to decision theory.
Mitchell Green
193
with other norms. When that happens, what is absurd from the point of view of theoretical rationality might be acceptable, and perhaps even mandatory, from the perspective of that other system of norms. My system of beliefs might be absurd without its following that I am being absurd in reporting on or otherwise acknowledging them. My therapist shows me films of my behaving in phobic ways around spiders, even the harmless ones. I conclude from these observations of myself that I believe that all spiders are dangerous. However, after learning how few spiders are truly dangerous, I come to see that on the whole they are not, and that those that are tend to be reclusive. Yet old phobias are hard to shake, and I might acknowledge both this fact and my new empirical investigations into arachnids with the words, ‘Even though I believe that all spiders are dangerous, really they aren’t.’ This is an eminently reasonable thing for me to report. After all, each conjunct of that report is both true and justified. Yet if this report is sincere and correct, it follows (as we shall see below) that I am in error. Indeed I could know that at least one of my beliefs is in error with no further empirical investigation. According to our account of absurdity, it follows that my system of beliefs is absurd. That is compatible with the fact that when I assert, ‘Even though I believe that all spiders are dangerous, really they aren’t,’ this assertion is true, justified, and not the least absurd. The assertion might be true, justified, and not absurd. It might also be sincere. If it is, then it reports-and-expresses a pair of beliefs that cannot fail to put me in error. It reports my belief that all spiders are dangerous. It expresses my belief that not all spiders are dangerous. This pair of beliefs is absurd. But there the pair sits, and I have good evidence for the presence of that pair of beliefs. So it can be theoretically rational for me to report or express a system of beliefs that are sure to violate theoretical rationality, even when those beliefs are my own. If by dint of phobia or obtuseness you’ve violated norms of theoretical rationality to the point of absurdity, you’re only being reasonable in acknowledging the fact. It can be theoretically rational for me to acknowledge someone’s severe violation of theoretical rationality, even my own. It can also be theoretically rational to commit such a violation. Anyone who writes a sizeable book has a good chance of making an error somewhere. If she predicts in her preface that she has made an error somewhere in the ensuing five hundred pages, she is only being reasonable. On the other hand her preface together with the ensuing text cannot all be true. In addition, if we assume that she believes everything she says in her book (including the preface), her belief system is absurd. How then can her prefatory prediction of error be reasonable? The phobic stands back from his belief that all spiders are dangerous and acknowledges its falsity while also acknowledging that, being bound up with habits of mind, autonomic responses, and the like, it is hard to shake. The modest author stands back from the many beliefs she expresses in her book to acknowledge that the limitations of her intellect (and those of her research assistants and other authors on whom she has drawn) make it likely that she has erred somewhere in those five hundred pages.
194
Showing What’s Within
Those limitations are epistemic limitations—limitations on one’s ability to live up to norms of theoretical rationality. It is theoretically rational to acknowledge those limitations; after all, the author knows she is no less fallible than others. This is to say that it can be theoretically rational to harbor a system of beliefs that severely violate norms of theoretical rationality.⁶ Hence, it can be theoretically rational to harbor a system of beliefs that is absurd from the point of view of theoretical rationality. The so-called Preface Paradox is thus a misnomer. Instead we should acknowledge the existence of prefatory absurdity while denying that such utterances or beliefs are paradoxical at all. The phobic puts his finger on his absurdity; he knows which of his beliefs is wrong without being able to shake it. The author draws a circle around her absurdity; she acknowledges that at least one of her many beliefs is likely to be wrong without being able to pinpoint which ones. You might even know that I harbor an absurdity without knowing whether the source of that absurdity is theoretical rationality, practical rationality, or some other system of norms. Suppose that a severe violation of one system of norms N1 would result in absurdity, and a severe violation of another system of norms N2 would do so as well. Suppose now that you know that my behavior severely violates either N1 or N2 but you do not know which. In such a case, surely, you know my behavior is absurd in spite of your not knowing the precise source of that absurdity. Similarly, suppose N1 and N2 are systems of norms, and I perform an action of which I can know, with no further empirical investigation, that it will violate either N1 or N2. Here, while we cannot infer, of either N1 or N2, that I have severely violated it, we can infer that I have severely violated their conjunction, namely N1 & N2. In that case my behavior is again absurd in spite of our not being in a position to locate the source of that absurdity in a violation either of N1 or of N2.
3 . S PE A K E R M E A N I N G In what follows I shall use ‘meaning what s/he says’ and its cognates to refer to cases in which an agent says something in such a way as to take responsibility for it. On this usage, one means what one says even if one is a liar, and even if one is mistaken. On this usage, one who assertively utters an indicative sentence P means what she says only if she stands to be right or wrong on the issue ⁶ Descartes’s theism and views about evil lead him to conclude that so long as we are using our minds in the most rigorous and careful way possible, God would not permit us to be in error. For this reason, Descartes would hold that so long as the author in question has done her utmost to ensure the accuracy of her claims, that is, so long as she clearly and distinctly perceives the truth of each of these claims, she could not be justified in making her prefatory prediction of error. In lieu of Descartes’s extraordinary but implausible position, we do well to acknowledge the likelihood of our error even when, as authors, we have done our best.
Mitchell Green
195
of P depending on whether P is true; she also puts forth P as something she believes, whether or not she really does believe it. Ronald Reagan did not, we hope, mean what he said in warming up a microphone before a press conference with the words, ‘The bombing begins in five minutes.’ So too one does not generally mean what one says in rehearsing lines from a play. By contrast, you mean what you say in promising to educate a dying friend’s children only if you commit yourself to educating those children, and only if your sincerity in making that promise depends upon your intending to educate them. The notion of meaning under scrutiny here is often referred to by the misleading phrase, ‘speaker meaning’. What is speaker meaning? I shall pursue only as much of an answer to this question as is needed for the account below. Paul Grice is widely believed to have shown that to achieve a case of speaker meaning, one must make an utterance with the intention of producing an effect on an audience, with the further intention that this effect be achieved at least in part by the audience’s recognition of your intention. (Grice 1957, 1969, 1982) This is a so-called reflexive communicative intention. Unfortunately, Grice’s conditions for speakermeaning are too restrictive (Davis 2003; Green 2003; forthcoming). Speaker meaning is possible in the absence of a reflexive communicative intention. For instance, when Herod presents Salome with St John’s severed head on a charger, he both shows her that St John is no longer and means that St John is no longer.⁷ He intends Salome to come to believe that St John is dead, but presumably intends her to conclude this from the presence of the severed head rather than from any recognition of his intention that she believe anything. After all, the severed head is there for her to see. Speaker meaning does not require a reflexive communicative intention. In fact, it does not even require a communicative intention (an intention to produce a cognitive effect, such as a belief, on an audience). A framed suspect might mean that she is innocent in saying, ‘I am innocent!’ Yet she is fully aware that no one will believe her and, being realistic, she does not intend to convince anyone. She might not even intend her interrogators to believe that she believes she is innocent, since she might know that they’re certain she’s lying. Or, gazing into my newborn daughter’s eyes I might say, ‘All things valuable are difficult as they are rare,’ meaning what I say, without having the slightest intention to produce beliefs or other attitudes in her or in anyone else. Again, in the film Sleeper, Woody Allen’s character Miles Monroe discovers in his solitary exploration of a futuristic world a genetically modified chicken the size of small house. Miles remarks, ‘That’s a big chicken.’ In saying this he does not seem to be intending to produce an effect on anyone, himself included. ⁷ The fact that he is not telling her that St John is no longer is neither here nor there: We know independently that speaker meaning does not require telling. One can, for instance, speaker-mean, ‘How many apples are in the bowl?’, without making an assertion.
196
Showing What’s Within
An intention to communicate, to say nothing of a reflexive intention to communicate, is not required for speaker meaning. However, one way of achieving speaker meaning is to harbor another kind of reflexive intention, specifically, that built into the notion of overtness. Doing something overtly involves as a minimum making some aspect of oneself manifest. I might make manifest the bruise on my arm by pulling back the sleeve that had been covering it. In so doing I need not cause anyone actually to perceive that bruise. What matters instead is that the bruise be open to view by appropriately situated observers. I can likewise manifest my commitment to some proposition P by making that commitment open to inspection to appropriately situated interlocutors. I might do that by asserting P in front of an audience, be they distracted or not; I might do it by asserting P in the privacy of my study. In my locked, unbugged study, the only audience to whom I am manifesting my commitment is myself. I nevertheless mean what I say when I say to myself, ‘All things valuable are difficult as they are rare.’ I can, however, make some aspect of myself manifest without being overt. The bulging vein on my forehead makes my anger manifest without itself being, or being a part of, any overt behavior on my part. I can even intentionally make some aspect of myself manifest without behaving overtly. As we approach each other in a dark alley I cough to keep from startling you when we get closer; but I need not be overtly coughing. Contrast this with a case in which I am a schoolteacher who has come upon a young couple displaying affection, and I stentoriously clear my throat before they see me. Here my throat-clearing will be overt: Not only do I intentionally manifest my presence, I also manifest my intention to manifest my presence. Overtly to do something requires doing it with the intention that the act be manifest, and further with the intention that that very intention itself be manifest. But it is not true that I can achieve overtness merely by having two intentions, namely (a) an intention that some commitment of mine be manifest, and (b) a further intention that this very intention be manifest. Consider the following Strawson-inspired example (Strawson 1964). You are exploring a house for possible purchase. I want to manifest to you my belief that it is rat-infested, and so enact the following plan. I will enable you to see me leaving a rat in the house, while acting as if I think I am unobserved. I also know that you think me a good and veracious friend, and know that, although you won’t conclude from the presence of the rat that it is genuine evidence that the house is rat-infested, you will nevertheless conclude from my odd behavior that I must believe that the house is rat-infested. In acting as I do in the above example, I intend that my belief that the house is rat-infested be manifest. I thus satisfy condition (a) above. I will also intend in so acting that this very intention (that my belief that the house is rat-infested be manifest) itself be manifest. Hence, I satisfy condition (b) as well. However, in this case it does not seem that I overtly display my belief that the house is
Mitchell Green
197
rat-infested. The problem appears to be that, as in the case of my secreting your monogrammed handkerchief into the crime scene in order to implicate you, something about my performance is under the table—or at least the floorboards. Contrast this with a case in which you and I are discussing the aforementioned house over lunch, and you ask whether I think it infested with anything. I happen to have a rat in my briefcase, and wordlessly take it out, dangling it by the tail in front of you. Here, surely, I overtly display my belief that the house has a vermin problem. From the foregoing it seems that at least one means to achieve overtness in a way that will make speaker meaning possible, requires all the relevant intentions be out in the open.⁸ One approach to filling this need is to demand that to be overt, the intention that my commitment be manifest self-referentially requires this very intention to be manifest as well.⁹ That suggests the following sufficient condition for speaker meaning: Sufficient Condition for Propositional Speaker Meaning: S means that P if 1. S performs an action A intending that: 2. In performing A, S’s commitment to P be manifest, and that it be manifest that S intends that (2). P might be the proposition that it is blustery outside, with A being the uttering of certain words or a non-conventional action such as the throwing open of curtains to reveal a looming storm. In the Herod case the P in question is the proposition that St John is dead, and the action is Herod’s presenting of St John’s head on the charger. Once again, one’s commitment to P might be manifest without anyone being aware of this fact. As a result one can overtly intend that one’s commitment to P be manifest without intending to produce effects on others. In light of our Sufficient Condition for Propositional Speaker Meaning, then, one might mean something without intending to produce effects on any audience, and one can mean something in the course of overtly making one’s commitment manifest.¹⁰ ⁸ That is compatible with our having ulterior motives in cases of speaker meaning. In remarking on the weather I might be trying to be sociable, or for that matter be intending to distract you while purloining your maraschino. In either case I nevertheless speaker-mean that it’s a nice day when I say, ‘Nice day.’ For further discussion of the distinction between ulterior and ostensible motives in speaker meaning see Green (1999b). ⁹ Green (forthcoming: ch. 2), develops this point in further detail. In addition to elucidating the sufficient condition given below for propositional speaker meaning, speaker-meaning that P, that work also develops the notion of objectual speaker meaning, in which an agent means α, where α is an object. ¹⁰ Some authors have been skeptical of the very possibility self-referential intentions. (Such scepticism is not to be confused with scepticism of the doctrine that all intentions are self-referential. One can accept that self-referential intentions are possible while remaining neutral on the question whether all intentions are self-referential.) Thus for instance Seibel (2003) writes, ‘the content of [the self-referential intention] contains an element which refers to the intention itself. But what does that element look like? . . . How does it single out the intention and nothing but it? By identifying features, i.e., properties which are exclusively possessed by the intention? But what could be these
198
Showing What’s Within
4 . S PE A K E R M E A N I N G I N YO U R H E A RT ‘Speaker meaning’ is a misleading label because it applies to cases involving no speech. Suppose that we share no language and I want to alert you to the danger of the quicksand nearby. I mime out before you my being pulled under the surface by quicksand. In so doing I might speaker-mean that there is quicksand nearby without making a sound—indeed without moving my lips. The label ‘speaker meaning’ is also misleading because one can speaker-mean things without doing anything publicly observable. To see this, consider first of all that the distinction between speaker-meaning and merely mouthing words has an analogue in the realm of thought. I can think through the lines of a poem in my head without assenting to what those lines say. On the other hand, I might not only say something to myself but also mean it in such a way as to be committed to it. As with speech, I might thereby mean some proposition P in such a way as to stand to be right or wrong on the issue of P depending on whether P is true. If in the course of a vigorous morning run I resolve to make it up to the top of the hill before me by saying in my heart the words, ‘I shall conquer that hill!’, I might mean what I say. If I do, then I undertake a commitment that will be satisfied if and only if I conquer that hill.¹¹ I take the phenomenon of speaker-meaning something to oneself to be familiar and uncontroversial. In spite of this, talk of so-called ‘inner speech’ tends to make philosophers nervous because of its association with discredited attempts to base an account of linguistic meaning upon an incorrigibly private language.¹² However, I here make no claim that linguistic meaning can be explained in features?’ Intentions admit of the same act/object dichotomy as do many other intentional states, and like other mental events, are spatiotemporally located. On the modest assumption that no two intentions have identical spatiotemporal coordinates, we may then use such coordinates to individuate intentions. One might still wonder what the content is of an intention whose content refers to that very intention, which itself comprises both an intending (a state or act) and a content. One answer may be given in terms of an analogue of truth conditions applicable to intentions, namely satisfaction conditions. Just as the thought, had as I regain consciousness after a near-fatal accident, ‘This thinking is miraculous’, will be true just in case that thinking is, indeed, miraculous, so too, the intention, ‘This intention shall be manifest’ will be satisfied just in case that intention is, indeed, manifest. ¹¹ Anselm distinguishes between two things the fool might be thought to be doing when he says in his heart, ‘There is no God’ (1995: 101). On the one hand the fool might be silently saying these words to himself. Anselm thinks this case is possible. What he does not think possible is another case in which the fool not only says to himself, ‘There is no God,’ but also understands what he is saying. Anselm’s point is that if the fool were to grasp the concept of God, he would immediately see that God could not fail to exist. In what follows I will not lay down any limitations on what absurd thoughts a person can entertain even as he understands the words—if such there be—in which they are couched. ¹² For a creditable such attempt, however, see Davis (2003).
Mitchell Green
199
terms of mental acts of ascribing meaning to words. For all I say in this essay, acts of inner speech are only possible in the context of a linguistic community. However, one might still challenge the idea that it is possible overtly to do something in the privacy of my own thoughts. Does overtness, invoked in my sufficient condition for speaker meaning above, make any sense in this context? I contend that it does. My believing that it is raining outside does not guarantee that I will tell myself that it is. For instance if I am conscious of the rain but not attending to it, I will not tell myself that it is raining. Hence I can believe that it is raining outside without speaker-meaning anything. I might also make my belief that it is raining manifest to myself by, for instance, going to the closet to take out an umbrella. Here too, however, I haven’t speaker-meant anything, but have merely acted on and thereby displayed my belief that it is raining outside. What if I intentionally manifest to myself my belief that it’s raining? This too is not sufficient for speaker-meaning something to myself. I film myself going outside with an overcoat, hat, and umbrella. Unfortunately I also have anteriograde amnesia, which prevents me from retaining new information for more than a few seconds. (Assume that I know that I suffer from this malady.) I leave the film of myself for a later stage of myself to watch. I know that that later person-stage will watch the film and see an earlier person-stage going out dressed for bad weather. I also know that the later person-stage will not know who or what made the film, and with what intent. Because of this, the earlier person-stage might intend, in placing the film where he does, to manifest to himself his belief that it is raining. However, he will in all likelihood not believe that in so doing he is manifesting to himself this very intention. After all, the later person-stage will have no idea where the film came from; and the earlier person-stage knows this. Because of this, the earlier person-stage will in all likelihood be unable overtly to manifest his belief that it is raining. This squares with our intuition that in leaving the film there for the later stage of himself to see, the earlier stage doesn’t speaker mean that it is raining. In order to speaker-mean something to myself, what is needed instead is a case in which I not only intentionally manifest to myself my belief that it is raining, but also intentionally make my intention to manifest that belief to myself, itself manifest. If I have anteriograde amnesia, and know it, I might create a movie to be a screen-saver on my computer showing me ostentatiously going to the closet to get a hat and umbrella. In that movie I stare significantly at the camera while donning gear for the weather. In placing the movie where I do, I may reasonably intend not only to make my belief that it’s raining manifest to my later self, but also intend to make manifest to my later self this very intention. (Assume that I have retained enough knowledge from before the accident producing my amnesia that I know that the only way this movie could have ended up as a screen-saver was if I had put it there myself.) As with the
200
Showing What’s Within
case of the quicksand example above, here I speaker-mean something without producing any words.¹³ The situation is similar when I say something to myself while uttering no words aloud. If I say in my heart, ‘It’s raining,’ I may reasonably intend to make my belief that it’s raining manifest to myself. Further, in a standard such case, not only do I intentionally manifest to myself my belief that it’s raining, I also intentionally make that very intention manifest to myself. For I will normally be aware that the only reason I would ‘hear’ myself saying these words was that I had intended to manifest my commitment to myself. In an unusual case, such as schizophrenia, I might be suffering from delusions of the sort that cause me to hear voices. If I am schizophrenic and know it, then I probably also know that I am prone to hear such voices. In that case it is less clear that I can intend, by saying something to myself, to manifest my intention to manifest a belief to myself. One symptom of schizophrenia, then, may be an inability to speaker-mean things to oneself by saying things in one’s heart. Speaker-meaning something in the privacy of one’s own thoughts seems possible, and indeed not at all unusual. This becomes clear in light of our sufficient condition for speaker meaning, which requires no communicative intentions, much less reflexive communicative intentions. However, it might seem that a disanalogy between speaker meaning in one’s heart and speaker meaning done in public is that only the latter can fail to be sincere. No mystery surrounds the idea of lying to others. What about lying to oneself? Sometimes we use this expression to refer to people believing things on insufficient evidence. For instance, Hunter’s lying to himself in thinking he can make it in time for his Chicago flight might be due merely to the facts that (a) his chances are so slim, and (b) he should have known better. This need not involve his telling himself anything that he thinks untrue. On the other hand, everyday experience also suggests that I can say something to myself, meaning it, without believing what I say. I tell myself that this is my last piece of pie when I know perfectly well that by the time I leave the room, the pie plate will be clean. The phenomenon is also reflected in literature. For instance, in Ann Packer’s, The Dive from Clausen’s Pier (Knopf, 2002), the narrator Carrie is visiting her fianc´e in the hospital after his spinal injury from a dive off a pier. She is gradually losing interest in caring for him, and wants to move away: Again he closed his eyes, and now tears seeped out, a single trail moving down each cheek. I set his hand down and began stroking his forearm again. I wish I could say I felt selfless then, unaware of myself. That I was thinking only of him, or that I wasn’t even thinking. But I was: This is me doing the right thing. This is me being brave and strong for Mike. (p. 102) ¹³ Those sympathetic to a ‘memory links’ account of diachronic personal identity need not be distracted with the question whether a person could survive an amnesia of this sort. Grice’s (1941) version of such an account can accommodate this kind of case.
Mitchell Green
201
When Carrie says, ‘This is me being brave and strong for Mike,’ she is telling herself that she is being brave and strong for Mike. She speaker-means it. Yet she doesn’t really think she is being brave and strong for him. She knows full well that she is moving away from him emotionally, that her visits to the hospital are becoming less frequent and more halfhearted, and that she may not even love him any more. We have intuitive support for the idea that one can speaker-mean something to oneself. We now have, in addition, intuitive support for the idea that one can speaker-mean something to oneself without believing it. Intuitions are fallible, however, and this latter intuition might seem mistaken. A speech act that is not sincere is typically made with the intent to deceive. Is it plausible that one can intend to deceive oneself by saying something to oneself? Likewise, speaker-meaning is widely construed as an attempt to produce an effect on an audience by means, at least in part, of their recognition of the speaker’s intention to produce that effect. I have rejected that account and suggested that it is at least sufficient for speaker meaning that one intentionally and overtly manifest one’s commitment to a proposition. Is it possible to do this in the privacy of one’s own thoughts without being sincere? I manifest my commitment to P by making that commitment available. I might do that by asserting P in front of a distracted audience; I make my commitment available to them even if they don’t acknowledge that commitment. If I manifest my commitment to P by asserting it in the privacy of my own study, I make that commitment available to myself and no one else. (Assume that there are no recording devices or eavesdroppers.) But much the same holds for things said in my heart. There are many things I believe that are not manifest to me. Introspection, psychotherapy, and elenchus are all ways of dredging up beliefs into consciousness. When I follow one or more of these paths, I might articulate what I find by consciously thinking to myself some such thing as: ‘I do seek my colleagues’ approval!’; or ‘I don’t think that consequences are all that matter for morality!’ As in the case of what I utter in my locked study, when I say one of these things ‘in my heart’ I make my belief manifest to myself and to no one else. That is still enough for me to speaker-mean that I reject consequentialism, and it is still enough for me to mean that not enough of my self-worth comes from within. I cannot make manifest, to myself or anyone else, a belief that I do not have. How then can I fail to be sincere in the confines of my own thoughts? In the cases in which I make a sincere assertion I not only make my belief manifest, I also (intentionally and overtly) manifest my commitment to the content of the proposition that I assert. Hence, when I sincerely assert that it is raining, I not only manifest my belief that it is raining, but also manifest my commitment to the proposition that it is raining. By contrast, I might manifest commitment to a proposition that I do not believe. Similarly, when Carrie says to herself, ‘This is me being brave and strong for Mike,’ she intentionally and overtly manifests
202
Showing What’s Within
commitment to the proposition that she is being brave and strong for Mike, without actually believing that proposition. She would like to be being brave and strong for Mike, but she doesn’t think she is (and in fact she isn’t). 5 . S PE E C H AC TS A N D T H E I R N O R M S We now see that speaker meaning is possible in the context of one’s own thoughts, and that it is even possible to speaker-mean something in such a milieu without being sincere. Further, like beliefs (which are governed by the norms of theoretical rationality), and actions (which are governed by the norms of practical rationality), speech acts are governed by their own system of norms. That system overlaps with but is not coextensive with the norms of theoretical rationality. According to speech act norms, assertions are to be relevant, justified by the speaker’s evidence, and sincere. Promises are to be sincere and not obviously impossible to fulfil. Questions are not to presuppose anything controversial. And so forth. What would be a severe violation of speech act norms? It is absurd to attempt to promise you something that is obviously beyond my control to bring about, e.g., to make it the case that the Napoleonic wars did not occur if it is common knowledge between us that those events took place in the past. It is absurd to attempt to bequeath something to you that, as you and I both know, is not my own to give, such as the Horsehead Nebula. These cases square with the account of absurdity given in Section 2 above, for in all of them one can discern with no further empirical investigation that one will violate a system of norms. Unlike some other speech acts, assertion is beholden to a norm of theoretical rationality, namely that one is to assert only what is true. It also includes a norm whose source is less clear: Assert only what one believes. We don’t need to decide here whether this sincerity norm governing assertion flows from theoretical rationality or from some other system of norms. Perhaps it does, or perhaps on the other hand it is a sui generis norm of assertion. For in either case, an assertion about which it can be known, just by inspecting its content, that it is false, is absurd. So are assertions whose content is falsified by particular aspects of their use. For instance, an utterance of ‘I am not now uttering any words,’ is false even though its content could be true. It is thus absurd knowingly to utter such a sentence even if it expresses a proposition that could be true. Similarly for a conscious thinking of, ‘I am not now thinking’: It is absurd consciously to think this sentence even if it expresses a proposition that could be true. (I assume that if one consciously thinks this sentence, one knows that one is doing so.) Suppose that we know of some promise P that anyone who makes it or purports to make it, meaning what he says, has either promised to do something that she will not, in fact, do, or has made a lying promise. Suppose, in addition, that we may infer this with no further empirical investigation. Were we able to
Mitchell Green
203
infer with no further empirical investigation that the promise will not be kept, we would be able to conclude that the promise is absurd. Were we able to infer with no further empirical investigation that the promise is not sincere, we would be able to conclude that it is absurd. Knowing that no matter how the world is, the promise is either one or the other of these is also enough to justify our concluding that it is absurd. It does not follow from this, though it may also be true, that such a promise is impossible. Fortunately we need not settle that issue. Similarly, suppose that we know of some sentence or thought S that anyone who asserts (thinks) it, meaning what they say (think) is either mistaken or is not sincere. We know this without requiring any empirical investigation, and no matter what else is in the speaker’s (thinker’s) mind. Did we know with no further empirical investigation that the speaker’s (thinker’s) utterance is mistaken, we could conclude that the utterance is absurd. Did we know with no further empirical investigation that the speaker’s (thinker’s) utterance is not sincere, we could conclude the same thing. As with the case of the promise just contemplated, our ability in principle to infer with no further empirical investigation that the assertion is either mistaken or not sincere is also enough to justify our conclusion that it is absurd. Here too it may be going too far to say that such an assertion (thought) is impossible, and here too we are fortunate that we need not settle that issue. I will argue that all instances of Moorean absurdity are absurd on account of severely violating norms for speech acts in this way: We may know with no further empirical investigation that either the speaker is in error, or is not sincere. We might not know which one it is, but as we saw in Section 2 above, this does not matter. What matters is that we may know with no further recourse to facts about the world that the speaker is violating one of the norms internal to assertion (or whatever speech act is the vehicle of her Moorean utterance). That is perfectly compatible with the content of what is asserted being true. It is also perfectly compatible with the speaker believing what is asserted. 6 . M O O R E A N S PE E C H A N D M O O R E A N T H O U G H T My approach assumes that all instances of Moorean absurdity are speech acts, even while some are not acts of speech. This assumption seems to be in conflict with the widely shared view that Moorean absurdity can be realized merely by believing some such thing as ‘P, but I don’t believe it’, rather than saying it. Surely belief is not a form of assertion, not even a form of mental assertion? It is considerations like these that make those interested in Moorean absurdity doubt that a speech-act approach could possibly be broad enough to cover the relevant explananda. It is not, however, true that merely believing, ‘P, but I don’t believe it,’ is absurd, or causes one to be absurd. If either that entire belief or one component
204
Showing What’s Within
of it is not accessible to conscious introspection, then the agent may be unable to discern with no further empirical investigation that she is in error. Suppose that either the entire belief, ‘Spiders are harmless, but I believe they aren’t,’ or one of its two conjuncts, is locked in my unconscious in such a way that it would require at least a year of intense psychotherapy including films of my own behavior, virtual reality exercises, fMRI information, and so forth, to come to be aware of it. In believing both conjuncts of ‘Spiders are harmless, but I believe they aren’t,’ I cannot fail to be in error.¹⁴ However, the same goes for my belief that Hesperus is shining but Phosphorus is not. In both cases it would take empirical investigation to determine that I must be in error. In the latter case I must learn the identity of Hesperus and Phosphorus; in the former case that I believe both conjuncts of, ‘Spiders are harmless, but I believe they aren’t.’ Just as it is difficult to see what would be absurd about believing that Hesperus is shining but Phosphorus is not, so too it is difficult to see what would be absurd in this case of believing both conjuncts of, ‘Spiders are harmless, but I believe they aren’t.’ If some ‘part’ of a Moorean belief is unconscious, then one may be unable to bring it into consciousness by introspection. Rather, one may need to do some empirical investigation to detect its presence. Because of that, an unconscious belief of this sort is no more absurd than believing both that Hesperus is shining and that Phosphorus is not. It may not be intuitively clear what an unconscious belief having the form either of a commissive or omissive Moorean sort would be like; yet we need not dwell on this issue. The reason is that approaches taking Moorean belief as the explanans and other cases of Moorean absurdity as the explananda,¹⁵ predict that believing the following two propositions puts one in a situation of Moorean absurdity: (a) P, (b) I believe that not-P. One need not believe their conjunction in order to exemplify Moorean absurdity, according to this standard account. After all, the argument used in the footnote to the last paragraph applies equally well to anyone who merely believes both (a) and (b). Similarly for the commissive case. Accordingly, all we need to imagine is an ¹⁴ We discuss the omissive and commissive cases in order. Assume that belief distributes over conjunction. Then one who believes ‘P, but I don’t believe it’ believes P, and believes that he does not believe that P. That latter belief must be in error. For the commissive case suppose that I believe that (p & I believe that not-p). Then assuming that belief distributes over conjunction, I believe that p. But what I believe, that p & I believe that not-p, is true only if I also believe that not-p. Thus what I believe is true only if I have contradictory beliefs about p, one of which must thus be mistaken. A Moorean belief, be it omissive or commissive, conscious or unconscious, cannot fail to put the person harboring it in error. I should mention as well that we are here assuming that the unconscious belief about spiders can only be made known to me by empirical investigation: it is ‘subconscious’ rather than ‘preconscious’. According to theories of the ‘adaptive unconscious’ now gaining currency (canvassed in Wilson 2002), some unconscious material can only be known in a ‘third personal’ way rather than in a ‘first personal’ way; they can thus only be known by empirical investigation. I take the spider belief to be a case of this kind. ¹⁵ Positions of this kind are discussed in Chapter 1, Section E.2 in this volume.
Mitchell Green
205
agent whose belief (a) is unconscious while his (b) belief is conscious. In that case, while his entire set of beliefs put him in such a position that he cannot fail to be in error, he is not absurd for holding this set of beliefs. He cannot fail to be in error, yet he will be unable to discern this fact without further empirical investigation. For this reason, his system of beliefs is no more absurd than his believing that Hesperus is shining and Phosphorus is not shining. I claim that a necessary condition of an agent’s being absurd is that her severe violation of norms is one that she can in principle come to be aware of with no further empirical investigation. Accordingly, if some of my Moorean belief is submerged in my unconscious, then I am not, at least on this basis, absurd. Rather, one needs consciously to think some such sentence as ‘P, but I don’t believe it,’ in order to generate absurdity recognizably Moorean. But this by itself is not enough, for there are many propositions we think through without committing ourselves to them. Instead, one needs not just consciously to think it, but more specifically to speaker-mean it in the way described in Section 3 above. As we saw in that section, this does not require that one believe what one says to oneself. It does require that one manifest commitment to what one says to oneself, and in particular a commitment to the truth of what one says. (Strictly speaking, I have given only a sufficient condition for speaker meaning. However, I take it as not in need of argument that merely thinking a thought, and merely manifesting commitment to oneself, are not sufficient for speaker meaning.) 7 . M E N TA L A S S E N T The approach adumbrated thus far might also seem to be superseded by the view that takes ‘mental assent’ as the core notion, and explains other cases in its terms. However, as this term is normally used, mental assent must be sincere (Shoemaker 1995). The reason is that mental assent is construed by authors such as Shoemaker as an episodic instantiation of belief. On this usage, one cannot mentally assent to a proposition that one does not believe. Because of this, Shoemaker’s approach does not have sufficiently broad scope. It does not account for cases in which a person says to himself a Moorean sentence that he does not in fact believe. As noted above we have everyday familiarity with the experience of saying things to oneself that one means but does not believe. Doing so with a Moore sentence can still be absurd. It follows that the mental assent approach lacks adequate scope to be the source of a general explanation of Moorean absurdity. 8 . S H OW I N G W H AT ’ S W I T H I N Showing comes in at least three forms. First of all, I might show my courage by acting bravely. My brave behavior is good evidence for my courage. A
206
Showing What’s Within
grammatical tag for this category is showing-that, since the brave behavior also shows that I am courageous. Second, recall from our discussion of manifestation above that I might show something in such a way as to make it perceptible. I show my bruise, and thereby enable others to see that bruise. Let us put this perception-enabling form of showing under the rubric of showing-α, where ‘α’ is a singular term. Finally, I might also show how something looks, feels, sounds, etc. I present your nose with a durian and enable you to know how it smells. Similarly, the trepidation in my voice might enable you to know how my anxiety feels if you are sufficiently empathetic. If you are sufficiently empathetic, then hearing my voice may enable you to imagine feeling as I do. If you can do that, then you know how I feel. These three forms of showing we may label showing-that, showing-α, and showing-how. The discussion below depends primarily on the notion of showing-that; showing-α has a cameo in Section 10. We are now in a position to tie together various strands to support an account of what is absurd in the utterances that Moore was the first to discover. Assertions purport to show beliefs. If they are sincere, they do show those beliefs. This is not because they make beliefs perceptible (that doesn’t seem to make sense), nor because they show how a belief feels (beliefs don’t seem to feel like much of anything). Rather, an assertion is (inter alia) evidence that the speaker believes what is asserted. That is why the sincerity of an assertion shows that you believe what is asserted. Now suppose you show a belief or other attitude that you also deny having. Then since ‘show’ is a success verb, that denial must be in error. The showing might be public (in an utterance) or private (in a saying to oneself). If you show a belief (or other attitude) and then go on to describe yourself as believing its contradictory, then whether or not this latter statement is true, you are in error. On the other hand if you purport to show a belief (or other attitude) that you do not in fact have, then you are not sincere. Suppose then that you (perhaps silently) assert 1. P but I don’t believe that P. Then by the assumption that assertion distributes over conjunction, you have asserted P and have asserted that you don’t believe that P. The former assertion is either sincere or not. Suppose it is sincere. In that case, it shows your belief that P, but then your other assertion, that you don’t believe that P, is in error. In that case you’re in violation of the norm that assertions are to track the truth. On the other hand suppose that the assertion of P is not sincere. Once again you are in violation of a norm of assertion, namely to assert only those things you believe. So either the assertion is sincere or it is not; but in either case we may infer with no further empirical investigation that you are in violation of a norm of assertion. Hence with no further empirical investigation we may conclude of someone who asserts, ‘P but I don’t believe it’, that she is in violation of a norm of assertion. It is a severe violation of a norm or system of norms to perform an act of which
Mitchell Green
207
it can in principle be inferred with no further empirical investigation, that it will violate that norm. From our original contention that absurdity consists in a severe violation of a system of norms, we may infer that assertion of ‘P but I don’t believe it,’ is absurd. So too, suppose you assert 2. P but I believe that not-P, With assertion-distribution we may infer that you have asserted P and have asserted that you believe that not-P. Your first assertion is either sincere or it is not. Suppose it is sincere. Then your second assertion, that you believe that not-P, is either correct or incorrect. In the former case, you are in error: for your first assertion, being sincere, shows your belief that P, and so you believe that P; while the correctness of your second assertion implies that you believe that not-P. Anyone who believes both P and not-P is in error. On the other hand, if the second assertion, that you believe that not-P, is incorrect, then you are in violation of a norm of assertion. Likewise and as before, if the first assertion, namely that P, is not sincere, you still violate a norm of assertion. It follows that if you assert ‘P but I believe that not-P’, then we may conclude with no further empirical investigation that you are in violation of some norm of assertion. It is a severe violation of a system of norms to perform an act of which it can in principle be known with no further empirical investigation that it will violate those norms. From our original contention that absurdity consists in a severe violation of a system of norms, we may infer that assertion of ‘P but I believe that not-P,’ is absurd. I cannot be sure that the norm that assertions are to track the truth is a norm of assertion. It might instead be a norm of theoretical rationality, applying to assertion, as with any other activity aiming at the truth, simply by universal instantiation. However, if this is so it will not undermine our explanation of the absurdity in cases such as (1) and (2). For if this is a norm of theoretical rationality only, then one who asserts (1) performs an act that, no matter how the world turns out to be, either violates a norm of assertion or violates a norm of theoretical rationality. Likewise, if the norm that assertions are to track the truth is a norm of theoretical rationality only, then one who asserts (2) performs an act that, no matter how the world turns out to be, either violates a norm of assertion or violates a norm of theoretical rationality. As we saw in Section 2 above, that still suffices for absurdity. 9 . M I X E D I L LO C U T I O N C A S E S The approach offered here generalizes with little difficulty to cases other than those involving only assertion. I shall consider two such cases, one involving supposition, and the other involving interrogatives.
208
Showing What’s Within
Even off stage, one can utter an indicative sentence with other than assertoric force by, for instance, forwarding a proposition merely as something to be entertained, as a supposition for the sake of argument, or as a conjecture. Taking, for brevity, only the case of supposition for the sake of argument (which hereafter I shall just refer to as supposition), offhand it seems we can intelligibly inquire whether a speech act of supposing is sincere no less than we can inquire into the sincerity of a compliment, an assertion, an apology, or a promise.¹⁶ A supposition’s sincerity does not require the speaker to believe what she has proffered as a supposition. Rather, it is at least sufficient for her sincerity that the speaker perform such a mental act of supposing as one might do in one’s unspoken hypothetical deliberations. (I doubt that such a mental act is required for the speaker to be sincere, but that question need not be settled here.) The result of such a mental act is that the speaker is in the intentional state of supposing, occupancy of which I shall take to be at least a necessary condition for her speech act of supposing to be sincere. That a speech act of supposing can be assessed for sincerity might be obscured by the fact that it is not pragmatically deviant to assert or believe 3. P, though my state of mind is not one of supposing that P. One can also suppose this sentence (or its content) by imagining a case in which: P holds but one refrains from supposing P. Further, one can assert or believe without oddity 4. P, though my state of mind is one of supposing that not-P. One can also suppose it, thereby supposing both P and the proposition that her state of mind is one of supposing not-P. No paradox need result.¹⁷ On the other hand, as exemplified by the relevance of sentences involving interrogatives to the question what it is for an interrogative to be sincere, we need not restrict our inquiry to examples in which the two conjuncts of either of the sentences just displayed are put forth with the same illocutionary force, or are held under the same propositional attitude. Accordingly, consider a situation in which a speaker inscribes P under the scope of a supposition sign of the sort used in natural deduction systems. That sign indicates without asserting that P is put forth with the force of supposition, and will indicate that all reasoning carried out to its right and below P are within P’s scope. Assume further that the speaker ¹⁶ In its use as part of a verb phrase of the form ‘A supposes’ taking complements of the form ‘that P’, ‘suppose’ is often used to impute beliefs, sometimes with the suggestion that the believer is in error. I shall nevertheless consider only its use to refer to the acceptance of a premise for the sake of argument. This usage of ‘suppose’ is thus also to be distinguished from uses of ‘assume’ to refer to a person’s commitment, often unacknowledged, to the truth of a proposition. (‘Assume’ is however used at other times to refer to the use of a premise for the sake of argument, as are ‘say’, ‘pretend’, and ‘imagine’, and what is said below will apply to all these uses.) ¹⁷ Sorensen (1988) argues along similar lines that there are no imagination blindspots, and I take it that he would say the same for supposition. We are about to see that these points may be granted without its following that there is no analogue for the case of supposition of the Moore paradox.
Mitchell Green
209
is using the sign of supposition to guide her informal reasoning rather than to aid her in the use of a formal system, and that she appends to her inscription of P the parenthetical remark that her state of mind is not one of supposing P. Such a parenthetical remark would normally be read not as within the scope of the supposition sign but rather as being put forth assertorically, and thus would behave analogously to the parenthetical that occurs in 5. If (as is indeed the case) snow is white, then grass is green. Here the speaker asserts the conditional, ‘If snow is white, then grass is green’ while also putting forth ‘snow is white’ assertorically even though grammatically speaking the parenthetical clause occurs in the antecedent.¹⁸ Likewise, in 6. P (though my state of mind is not one of supposing P for the sake of argument)
the content of the parenthetical clause will normally be read as being put forth assertorically rather than as part of what is being supposed. In addition, the content P and the content expressed in the parentheses can be conjoined to form a logically consistent proposition. Nevertheless, if someone were to write the above display on a chalkboard addressing an audience familiar with the conventions of natural deduction, their audience may have no choice but to find the performance absurd. A first, charitable response to this performance might be to construe the parenthetical remark as retracting the supposition of P. However, just as we may be unable to construe one who says, ‘P but I don’t believe it’ as expressing a mid-utterance change of mind, so too this interpretation may be unavailable if, for instance, the speaker goes on to infer things from P. A second charitable response is that the speaker is dissociating herself from her supposition of P, perhaps because her commitment to not-P is so deeply entrenched that she cannot bring herself to reason as if P is true. This construal will also be ruled out by the speaker’s going on to reason under P’s scope with adequate facility. A third charitable response might be to construe the speaker as signaling that her state of mind is not merely one of supposing P, but is instead one of accepting P in a way that may seem stronger than supposition (i.e., belief or conjecture). This interpretation, too, may be ruled out by contextual factors, and would have been explicitly ruled out had the speaker instead inscribed ‘P (though my state of mind is not one of accepting P in any way at all)’, within the scope of the supposition line. Attempts at charitable interpretation might, in the end, ¹⁸ This perspective on parentheticals is defended in Green (2000b).
210
Showing What’s Within
meet with failure, with the result that the audience would have no choice but to find the speaker’s performance absurd. Such a case would be an analogue for supposition of the absurdity to be found in certain utterances of ‘P but I don’t believe it’.¹⁹ How shall we explain what is absurd in such a case? Suppose that the speaker’s putting forth P as a supposition for the sake of argument is sincere. Then her doing so shows her state of mind of supposing that P. Her parenthetical utterance, having the force of an assertion, must therefore be in error, whence she is in violation either of theoretical rationality or norms of assertion (or both). On the other hand, if the speaker’s putting forth P as a supposition for the sake or argument is not sincere, then she is in violation of a norm of the speech act of supposition. (No absurdity arises from a person’s putting forth P as a supposition for the sake of argument while going on to avow that she supposes not-P.) Accordingly, anyone who utters (6) above in the conditions described (including the condition of being within the scope of the supposition sign) is in violation of some system of norms, and we may determine this with no further empirical investigation. That is what makes her utterance absurd. Some authors have suggested that the following utterance exhibits Moorean absurdity: 7. It’s raining but I don’t know that it is. The point, however, needs to be handled with care. It is clear that some utterances of this sentence in a speech act generate no absurdity. We know from the discussion of supposition just offered that a speaker can utter (or think) an indicative sentence in a speech act without making an assertion. Suppose, then, the first conjunct of (7) is uttered as, say, a conjecture. It is perfectly appropriate to put forth a conjecture while making clear that you don’t know it to be true. This is attested by the fact that while it is appropriate to respond to my assertion of P with the challenge, ‘How do you know?’, it is not appropriate to challenge my conjecture with that question. It seems, more generally, that (7) is a case of Moorean absurdity only if the first conjunct is put forth assertorically. ¹⁹ Some authors, for instance Searle and Vanderveken (1985), and Rosenthal (1998), have suggested analogues of the Moore paradox involving neither belief nor assertion without developing the possibility of such a paradox for the case of supposition. However, not all of these authors keep sight of the fact that a speech act can generate Moorean absurdity only if it characteristically expresses an intentional state. As Heal (1977) argues, although a speech act such as an imperative might seem to generate Moorean absurdity, as in ‘Shut the door, but I don’t want you to shut the door’, this appearance is probably deceptive. The reason is that imperatives are not speech acts one of whose roles is the expression of an intentional state, and it is for this reason not the case that imperatives are speech acts that characteristically express an intentional state. One uttering an imperative might provide her addressee with evidence of her intentional state (perhaps a desire), but it does not follow from this that the imperator expresses any such state. Further, one performing an imperative might adventitiously express such an intentional state as a desire, but this fact is of little interest to the study of Moorean absurdity.
Mitchell Green
211
Let us suppose then that the first conjunct of (7) is an assertion. Then it does seem plausible that we have a case of Moorean absurdity. If so, then we may account for this fact as follows. An assertion, if sincere, justified, and correct, shows not only one’s belief, but also one’s knowledge. Going on to deny that one knows what one asserts must put one in error. Similarly, in 8. It’s raining but I know that it isn’t if one is sincere, justified, and correct, the first conjunct shows one’s knowledge, whence the second conjunct must be in error. (Unlike believing p and believing not-p, one can’t know this pair of propositions.)²⁰ I close this section with a remark about non-indicative versions of Moorean absurdity. It is not clear to me whether non-indicative cases of Moorean absurdity exist. For instance, I do not know whether, ‘Shut the door, but I don’t want you to shut the door,’ exemplifies Moorean absurdity. Likewise, I do not know whether ‘What time is it, even though I have not the slightest interest in knowing the time?’, exemplifies Moorean absurdity. Both cases are pragmatically odd, but it would be rash to infer that they exhibit the same sort of oddity that we find in Moore’s cases. However, if either one of these cases does exemplify Moorean absurdity, the approach offered in this paper explains why. The explanation would proceed by observing that if the first conjunct is sincere, then it shows a state of mind that the second conjunct mistakenly disavows; thus whether or not the first conjunct is sincere, the speaker is in violation of some system of norms, and she is in violation of such norms in a way that is open to inspection with no further empirical information.
1 0 . N O N - I L LO C U T I O N A RY C A S E S We glossed Moorean absurdity as any utterance or thought in which an agent overtly expresses an intentional state that she also explicitly disavows; or any utterance or thought in which an agent overtly expresses an intentional state whose content is incompatible with that of another intentional state that she also explicitly avows. This account does not strictly require that the agent in question use words, even in the privacy of her own thoughts. That raises the question whether we find Moorean absurdity in cases in which an agent scowlingly denies that she is angry, or exuberantly avows her lack of exuberance. Just as we have found cases of speaker meaning that do not require uttering any words, a speech act does not require any act of speech. For instance, an extended finger at an auction is a promise to purchase the item at the amount bid ²⁰ In (8) of course, we also have a semantic contradiction, but we would need some reason to think that absurdity is ‘additive’ before predicting that (8) ought to sound ‘more’ absurd than other cases that we have considered.
212
Showing What’s Within
on the condition that no one else outbids that offer. Pulling down one’s lower eyelid in some Mediterranean cultures is a warning. So too, nodding one’s head is an assertion under the right conditions. That is why it can be the vehicle of Moorean absurdity. As Williams (forthcoming) observes, if you ask me whether the pubs are open and I nod my head in emphatic agreement while saying, ‘I don’t believe so,’ this seems no less absurd than one of the standard cases 1 or 2 above. This and similar cases involving nonverbal speech acts can be explained in terms of the line of thought we have developed thus far: A nodded head, for instance, shows one’s acceptance of some salient proposition if in fact that head’s owner is sincere; together with her subsequent utterance we may infer with no further empirical investigation that one is either in error or not sincere. Other cases involving nonverbal behavior are quite different. When an agent scowlingly denies that she is angry, or exuberantly avows her lack of exuberance, she might be in error only. She might simply be unaware of the fact that she is scowling or that she is behaving exuberantly. The scowl might not be one that she has noticed, and she might not be conscious of the exuberant behavior. In those cases, her disavowal or avowal is simply a mistake of fact and so is not absurd: It is not the case that with no further empirical investigation she could conclude that she is in violation of a system of norms. If we find such cases amusing it is because it is easy to be amused at people who are blind to their own emotional displays. Likewise, one’s companion on a midnight walk through a cemetery who tremblingly says ‘Not scary at all,’ might be unaware of the tremors in her voice, and might be unable to detect her own fear without empirical investigation. Given these possibilities, we cannot conclude from these performances that the agent is behaving absurdly, or is in some other way absurd. The emotional expressions in these last cases are not intentional, to say nothing of overt. Might an agent overtly express an intentional state without conventional devices such as words or gestures? If so, that would suggest that a case of Moorean absurdity might be found in which an agent behaves both expressively and overtly (rather than performing a speech act) while disavowing what she expresses. To that end, imagine that I not only scowl, but overtly do so: according to the gloss given above, it would be sufficient to achieve this result if I not only intentionally display my anger, but also intend to display this very intention. In such a case it would be natural to describe me as scowling significantly. In fact, in such a case it is also natural to describe me as speaker-meaning that I am angry. That does not fall under our sufficient condition for propositional speaker meaning as given in Section 3 above. The reason is that condition applies when an agent manifests her commitment (to a proposition, state of affairs, etc.). However, it is not clear that I am committing myself to anything in scowling, even overtly. Instead, what I am doing in overtly scowling is displaying my anger—making that anger perceptible rather than merely giving evidence for the presence of that anger. (This is suggested by the fact that if I am dissimulating, not showing my anger but merely seeming to do so, I may be misleading or mendacious but no
Mitchell Green
213
liar.) In this case, I am showing my anger in the showing-α way discussed in Section 8, rather than merely showing that I am angry. Speech acts are not the only vehicles by which we express ourselves. They are also not the only way in which speaker meaning is achieved. In a case in which we intentionally and overtly show our intentional state, we speaker-mean that state without performing a speech act, and we express that intentional state as well. Suppose that I scowl in such a way as not only to express my anger, but to do so overtly. If I could at the same time deny that I am angry, then we would have the makings of Moorean absurdity. However, it is not clear that I could do both these things at once. For in light of what I literally say when I deny that I am angry, it is hard to see how an interpreter could sensibly construe me as intending overtly to display my anger. My literal utterance will put pressure on the interpreter either to construe my facial behavior as inadvertent or covert; or at least as facetious. That is why, when it might seem as if a case of Moorean absurdity involving nonverbal, non-conventional expressive behavior is in the offing, the best we may be able to do is either to describe the agent as protesting too much or as hamming it up. REFERENCES Anselm (1078 [1995]), Monologion and Proslogion, ed. T. Williams (Indianapolis: Hackett). Collins, A. (1996), ‘Moore’s Paradox and Epistemic Risk’, Philosophical Quarterly, 46: 308–19. Cargile, J. (1967), ‘On Believing You Believe’, Analysis, 27: 177–83. Davis, W. (2003), Meaning, Expression and Thought (Cambridge: Cambridge University Press). Green, M., (1999a), ‘Attitude Ascription’s Affinity to Measurement’, International Journal of Philosophical Studies, 7: 323–48. (1999b), ‘Illocutions, Implicata, and What a Conversation Requires’, Pragmatics & Cognition, 7: 65–92. (1999c), ‘Moore’s Many Paradoxes’, Philosophical Papers, 28: 97–109. (2000a), ‘The Status of Supposition’, Noˆus, 34: 376–99. (2000b), ‘Illocutionary Force and Semantic Content’, Linguistics & Philosophy, 23: 435–73. (2003), ‘Grice’s Frown: On Meaning and Expression’, in G. Meggle and C. Plunze (eds.), Saying, Meaning, Implicating (Leipzig: University of Leipzig Press), 200–19. (forthcoming), Self-Expression (Oxford: Oxford University Press). Grice, P. (1941), ‘Personal Identity’, Mind, 50: 330–50. (1957), ‘Meaning’, Philosophical Review, 66: 377–88; repr. in Grice (1989). (1969), ‘Utterer’s Meaning and Intentions’, Philosophical Review, 78: 147–77; repr. in Grice (1989). (1982), ‘Meaning Revisited’, in Mutual Knowledge, ed. by N. Smith (New York: Academic Press), 223–43; repr. in Grice (1989). (1989), Studies in the Way of Words (Cambridge, Mass.: Harvard University Press).
214
Showing What’s Within
Heal, J. (1977), ‘Insincerity and Commands’, Proceedings of the Aristotelian Society, 77: 183–202. (1994), ‘Moore’s Paradox: A Wittgensteinian Approach’, Mind, 103: 5–24. (2003), Mind, Reason, and Imagination (Cambridge: Cambridge University Press). Kriegel, U. (2004), ‘Moore’s Paradox and the Structure of Conscious Belief ’, Erkenntnis, 61: 99–121. Nagel, T. (1979), ‘The Absurd’, in Mortal Questions (Cambridge: Cambridge University Press). Rosenthal, D. (1998), ‘Thinking That One Thinks’, in A. Burri (ed.), Language and Thought (Berlin: De Gruyter), 259–87. Searle, J., and Vanderveken, D. (1985), Foundations of Illocutionary Logic (Cambridge: Cambridge University Press). Shoemaker, S. (1988), ‘On Knowing One’s Own Mind’, Philosophical Perspectives, 2: 183–209. (1995), ‘Moore’s Paradox and Self-Knowledge’, Philosophical Studies, (77): 211–28. (1996), The First Person Perspective and Other Essays (Cambridge: Cambridge University Press). Siebel, M. (2003), ‘Illocutionary Force and Attitude Expression’, Linguishes and Philosophy, 26: 351: 66. Sorensen, R. (1988), Blindspots (Oxford: Clarendon Press). Strawson, P. (1964), ‘Intention and Convention in Speech Acts’, Philosophical Review, 73: 439–60. Szabo-Gendler, Z. (2001), ‘Fictionalism and Moore’s Paradox’, Canadian Journal of Philosophy, 31: 293–308. Tversky, A. (1975), ‘A Critique of Expected Utility Theory: Descriptive and Normative Considerations’, Erkenntnis, 9: 163–73. and Kahneman, D. (1992), ‘Advances in Prospect Theory: Cumulative Representations of Uncertainty’, Journal of Risk and Uncertainty, 5: 297–323. Wakker, P. and Tversky, A. (1993), ‘An Axiomatization of Cumulative Prospect Theory’, Journal of Risk and Uncertainty, 7: 147–76. Williams, J. (forthcoming), ‘Wittgenstein, Moorean Absurdity and its Disappearance from Speech’, Synthese. Wilson, T. (2002), Strangers to Ourselves: Understanding the Adaptive Unconscious (Cambridge, Mass.: Harvard University Press).
PART V A RG U M E N TS F RO M M O O R E ’ S PA R A D OX
This page intentionally left blank
10 My Philosophical Position Says p and I Don’t Believe p Alan Hájek
I N T RO D U C T I O N There is typically something strange about asserting or believing Mooreparadoxical sentences. In this paper I want to harness this strangeness to do a particular kind of philosophical work. I will argue that various prominent philosophers are committed to asserting and believing various Moore-paradoxical sentences in virtue of the very philosophical positions that they hold. Some of the philosophers in question may be surprised to learn of their commitment and find it unwelcome; others may not be troubled by their commitment and claim that not all Moore sentences are paradoxical after all; still others may positively celebrate their commitment and the paradoxicality, perhaps in the name of notoriety or boldness of thought. In any case, observing the commitment may help lay bare what we find peculiar in these philosophical positions, where previously we may only have had vague feelings of unease; or we may regard the Moore-paradoxical commitments as further reductios of the positions; or we may conclude that not all Moore sentences are paradoxical after all. Whichever way things go, I hope that some philosophical progress will be made. Along the way, I will consider some philosophical positions that may not actually have been held by any philosopher, prominent or otherwise, but that still have some interesting Moorish consequences.
I thank Jon Kvanvig for helpful discussion at an early stage, and Graham Priest, Roy Sorensen, and especially Andy Egan, Jordi Fernandez, Mitch Green, and John Williams for very astute comments on earlier drafts.
218
My Philosophical Position
MOORE SENTENCES, AND MORE SENTENCES (OF THE SAME KIND) We begin with the canonical versions¹ of Moore sentences—sentences of the form: (1) p and I believe that not-p. (2) p and I don’t believe that p. In very much the same spirit, we have: (3) p and I believe that ‘p’ is not true. (4) p and I don’t believe that ‘p’ is true. In fact there is a spectrum of sentences displaying the same peculiarity to varying degrees. For example: (5) p and I assign ‘p’ low probability. (6) p and I assign ‘p’ middling probability. (7) p and I don’t assign ‘p’ high probability. Along the way we will encounter further sentences that are similarly anomalous. The simplest way to reveal the varying degrees of oddity of such sentences is to begin with the idea that in asserting p, one represents oneself as knowing that p. (Cf. Unger 1975; Slote 1979; De Rose 1991; Williamson 1996.) But the smaller the probability that one assigns to p, the further one is from believing and a fortiori from knowing that p, and thus the greater is the discrepancy between how one represents oneself by an assertion of p, and how one represents oneself by an assertion of a probability assignment to p. Even if we do not accept this rather stringent account of assertion, we can surely agree that in asserting p, one conveys at least reasonable confidence in p—one conveys at least a moderately high subjective probability assignment to p. (Cf. Lewis 1976.) But one sends mixed messages by conveying also that this subjective probability is low, or middling, or not moderately high. After all, probability functions, even subjective probability functions, are functions: they cannot assign two different values to the same proposition. So there is apparently ¹ There are non-canonical instances of the same puzzling phenomenon—e.g. ‘God knows that I am an atheist’ (Sorensen 1988). Conversely, there are non-puzzling instances of Moorean-looking sentences: for example, ‘it is raining, and I don’t believe that there is precipitation’, uttered by someone who does not know what the word ‘precipitation’ means. More generally, someone may not know that the sentence denoted by ‘p’ expresses or is implied by the proposition p. Thus, a sentence’s having the canonical Moorean syntax is neither necessary nor sufficient for its being Moore-paradoxical. I will not attempt to give an analysis of just what Moore-paradoxicality consists in. Instead, I will play the Justice Stewart defense, resting content with knowing the phenomenon when I see it; moreover, I believe that all the examples I adduce clearly display it. I thank John Williams for discussion on this point.
Alan H´ajek
219
no way of interpreting someone who utters (5)–(7) as having a single probability assignment for p. And the lower the probability stated in the second conjunct, the harder this task of interpretation becomes. (Note that it is not strange in the same way to say p, and ‘p’ is probable, since both conjuncts send the same message of confidence in p. Strangely wordy, perhaps—but not strange in the same way.) Thus, if you assert any of (1)–(7), you apparently represent yourself to others as having a certain attitude to the world, which you then undermine or contradict with the representation of another attitude. But Moore’s paradox is as much a puzzle for belief as it is for assertion. And it would be equally puzzling to represent yourself to yourself in any of these ways—by believing any of these sentences. Moreover, it should seem puzzling for you to believe something that entails any such sentence—thus implicitly committing yourself to the truth of the sentence—if the entailment is easily recognized. The lore has it that asserting or believing Moore-paradoxical sentences is problematic. And yet a number of philosophers are implicitly committed to doing both in virtue of the philosophical positions that they espouse: things that they explicitly say entail instances of (1)–(7), where the entailments are easily recognized. If we alert them to their commitment to Moore-paradoxical sentences, they should either assert and believe these sentences—which many, following Moore, find ‘absurd’—or rethink their philosophical positions.
S O M E PH I LO S O PH I C A L P O S I T I O N S W I T H M O O R E - PA R A D OX IC A L C O N S E QU E N C E S
‘There are no beliefs’ Churchland (1981) and Stich (1983) are skeptical about the very notion of belief. ‘Belief ’ is part of a suspect folk psychology, likely to go the way of phlogiston and vital spirits—that is, ultimately to be discarded by science. On this view, neither I, nor you, nor anybody else ever has, ever had, or ever will have beliefs. Rather, we have whatever mental items will be postulated by a fully developed psychological theory. Thus, Churchland and Stich are committed to uttering sentences such as: ‘It is raining and I don’t believe that it is raining (and neither do you, nor anybody else, for ‘‘belief ’’ is part of a suspect theory of the mental)’. We have an easy way of generating sentences of type (2). Now, Churchland and Stich would surely not be fazed by this gambit—nor by the putative reductio that ‘they offer a philosophical position that, by their own lights, they don’t really believe!’ The right thing for them to say, of course, is
220
My Philosophical Position
that their mental state regarding the rain, or their own position, is whatever the fully developed psychological theory postulates it to be. Still, their commitment to Moore sentences is genuine, as is the puzzlement that it may induce in many of the rest of us. And we see our first trick for generating Moore sentences, one that we will see again: subject belief to different standards from those for (sincere, warranted) assertion. Indeed, according to Churchland’s and Stich’s standards for belief, nothing meets them: nobody ever had, ever has, or ever will have a belief. Yet presumably their standards for assertion are the usual ones. Thus, on their view, one may succeed in asserting something (sincerely and with warrant) while lacking a belief in that thing.² Now suppose that you are not totally convinced by Churchland and Stich, but you think that it is at least an open question—something else close to Moore’s heart—whether or not they are right. It seems that you are committed to sentences such as: ‘It is raining, and it is an open question whether I believe it is raining (and whether you do, and whether anybody else does, for it is an open question whether Churchland and Stich are right)’ and perhaps: ‘It is raining, and I am agnostic about whether I believe it is raining . . .’ These, too, sound pretty Moorish. If you want to avoid such a predicament, you should not be even agnostic about whether Churchland and Stich are right. Not that they would welcome this talk of agnosticism either, since presumably by their lights it is another folk psychological notion to be jettisoned. So if you really want to keep an open mind regarding their position, perhaps even the latter sentence is too committal. Better to play it safe: ‘It is raining, and it is an open question whether I am agnostic about whether I believe it is raining . . .’ And so on. Churchland’s and Stich’s position is programmatic; they await the details of a fully developed psychological theory. Let me go out on a limb and suggest one way that things could conceivably go. Psychology could adopt wholesale the terms of Bayesianism, eschewing talk of beliefs in favor of subjective probabilities. There is already this tendency in the work of Jeffrey (e.g. 1968), who seeks to replace the concept of knowledge with that of subjective probability, thus downplaying ² I say ‘presumably’ because I assume that even qua eliminativists, Churchland and Stich feel free to assert things very much as the rest of us do—after all, even their philosophical works on eliminativism are full of assertions, any one of which I could use to make my point. To be sure, assertion is usually characterized in terms of its role as the (purported) expression of belief, or in similar mentalistic terms. An eliminativist account of assertion would presumably look rather different. The mentalistic adjective ‘sincerely’ may similarly require an eliminativist gloss. ( Thanks here to an anonymous referee for this volume.)
Alan H´ajek
221
skeptical concerns about knowledge. Pushing this further, a Jeffrey-inspired psychologist might insist that ‘belief ’ cannot simply be reduced to subjective probability (citing, perhaps, the lottery and preface paradoxes, of which more shortly), and that subjective probability is the proper doxastic notion: folk psychology should be eliminated in favor of Bayesianism. Then speaking in his capacity as eliminativist, the psychologist may well say: ‘It is raining, and I don’t believe that it is raining (instead, I assign high subjective probability to it raining).’
‘Beliefs are propositions assigned subjective probability 1’ or ‘. . . very high subjective probability’ On the Churchland/Stich/Jeffrey-inspired psychologist view, beliefs are much harder to come by than you might think (since earning their keep in a mature psychological theory is a tough standard to meet)—so much so that there simply aren’t any. But we need not adopt positions as radical as theirs in order to generate Moore sentences. It suffices to adopt unusually high standards for belief while keeping normal standards for sincere, warranted assertion. Thus, we might be less demanding than Churchland, Stich, and the Jeffrey-inspired psychologist about the notion of belief, but demanding enough. For example, Hawthorne and Weatherson (2004) argue that ‘S believes that p’ should be analyzed as ‘S assigns subjective probability 1 to p’. Beliefs do exist on this view, but still they are harder to come by than you might think. If sincere, warranted assertions remain as easy to come by as you think, we can find propositions that make the cut for assertion, but that do not make the cut for belief, so understood. For such a proposition p, presumably S may properly assert p while disavowing belief in it. We might be moved by the lottery paradox,³ for example, to say that no threshold of subjective probability below 1 is sufficiently high to count as belief: you don’t really believe that your ticket will lose, you merely assign it probability 0.999999 of doing so. As long as a lower threshold suffices for (sincere, warranted) assertion, the conditions for Moore sentences are in place. Thus, it apparently becomes reasonable to say: ‘My ticket will lose (I assert this because my probability is above the threshold for assertion), but I don’t believe that my ticket will lose (since my probability falls below the threshold of 1 for belief)’. We may lower the standard for belief while arguably maintaining some daylight between the new standard and that for warranted assertion—and this still suffices to generate Moore-paradoxical sentences. Suppose that we set the bar for belief not at probability 1, as Hawthorne and Weatherson did, but at ³ Kyburg (1961).
222
My Philosophical Position
0.999. Plausibly, various sentences will still clear the bar for assertability, but not the bar for belief. As it might be: ‘the pubs are open (I am entitled to assert this since my credence in it is sufficiently high), but I don’t believe that the pubs are open (since that credence does not quite reach 0.999)’. The preface paradox⁴ furnishes an example of a related phenomenon, this time producing commissive Moore sentences of type (1). You preface a long book that you have written with the modest words: ‘Despite all my efforts, I am sure that there is at least one mistake somewhere in the book’. But the book itself can be regarded as a very long assertion—the conjunction of many individual assertions. Indeed, suppose that we replace all periods but the final one throughout the book with ‘and’s.⁵ You should be as committed to this unwieldy sentence as you were to the original book (even if you have acquired some stylistic qualms about it). Conjoining the unwieldy sentence to your preface, we have an assertion of the form: [MY BOOK] (the unwieldy sentence), and I believe that MY BOOK is false (since I believe that at least some conjunct in it is false). Schematically, we have the dreaded p, and I believe that p is false. The trouble is that assertion is an on/off, all-or-nothing act, whereas degrees of belief come in degrees. We do not have devices for giving assertion all of the nuance that we might want—say, boldness of typeface that varies with the strength of our convictions. Imagine that with each ‘and’ that we insert between successive sentences, their print fades accordingly; by the time we have conjoined them all, they become invisible! In a way, all of us—and not just certain idiosyncratic philosophers—find ourselves in the same uncomfortable shoes as the modest preface-writer, for on pain of gross immodesty, we all admit that some of our beliefs are false. Each of us would thus assert something of the form: [LONG CONJUNCTION OF MY BELIEFS] and I believe that LONG CONJUNCTION OF MY BELIEFS is false. Note that this does not require you to provide a complete enumeration of all your beliefs—an impossible task, surely. It suffices that you can find some long conjunction of beliefs of yours that you believe is false—an easy task, surely. For example, a few dozen beliefs of yours about the capital cities of countries in the world, or about phone numbers, or about the names of the children of your friends and colleagues, may well do the job. The conjunction might not even have to be at all long. Consider a recalcitrant paradox that has a grip on you that can be presented as a short list of premises, ⁴ Makinson (1965). ⁵ I assume that the book consists solely of sentences that have truth-value—I ignore, for example, imperatives or questions.
Alan H´ajek
223
each of which you believe, but which you recognize to be jointly contradictory. You assert: [SHORT CONJUNCTION OF THE PREMISES] and I believe that SHORT CONJUNCTION OF THE PREMISES is false (since I recognize them to be jointly contradictory). Perhaps one of Kant’s antinomies will fit the bill. In fact, soon we will see a single premise giving rise to this phenomenon—although only for certain idiosyncratic philosophers.
‘There are no higher-order beliefs’ Suppose that you are not skeptical about beliefs, the way that Churchland and Stich are, but that you are skeptical about higher-order beliefs. So you are happy to speak of beliefs (‘I believe that it is raining’, and so on) but you have no truck with beliefs about one’s own beliefs (‘I believe that I believe that it is raining’, and so on). Now take a belief of yours—say, that it is raining. Then you should be prepared to assert: ‘I believe that it is raining, and I don’t believe that I believe that it is raining (for that would be a higher-order belief, with which I have no truck)’ —something of the form ‘p, and I don’t believe that p’. Now suppose that you are not totally skeptical about whether there are higherorder beliefs, but you think that it is at least an open question. Then it seems that you are committed to sentences such as: ‘I believe that it is raining, and it is an open question whether I believe that I believe that it is raining’ —something of the form ‘p, and it is an open question whether I believe that p’.
‘There are only so many higher-orders that beliefs can reach’ Less radically, you might allow second-order, third-order, and perhaps still higher-order beliefs, but insist that the hierarchy must stop somewhere. You might say, for example, that the finiteness of our heads imposes limits on just how many times the belief operator can be iterated. (This is a commonplace in the literature on common knowledge in which it is acknowledged that the putative infinite iterations of ‘I know that you know that I know that . . . p’ are an idealization.⁶ See also Sorensen 2000.) Suppose, then, that you think that you have a highest order of belief: an nth order belief, for some n > 2. Find, if you can, an nth order belief of yours—say, that it is raining. Then you should be prepared to assert: ⁶ I thank Mitch Green for this point.
224
My Philosophical Position
‘I nth -order-believe that it is raining, but I don’t believe that I nth -orderbelieve that it is raining (for that would be an (n+1)th order belief, which exceeds my limit)’. And as before we can open the door to an ‘open question’ version of this.
‘There are no higher-order probabilities’; or ‘There are only so many higher-orders that probabilities can reach’ De Finetti (1972) and Savage (1954) are champions of subjective probabilities, but they are skeptical about higher-order subjective probabilities, or probabilities concerning one’s own probabilities. For example, by their lights it is nonsense to say ‘my probability that the coin lands heads is 1/2, and my probability that this really is my probability is 0.99’. De Finetti writes: ‘Any assertion concerning probabilities of events is merely the expression of somebody’s opinion and not itself an event. There is no meaning, therefore, in asking whether such an assertion is true or false or more or less probable.’ (1972:189). One of their arguments, roughly, is that any putative second-order probabilities would collapse to ordinary first-order probabilities—a reductio of the idea that there really were second-order probabilities in the first place. Both of these authors also seem to regard the threat of infinite regress as fatal to higher-order probabilities. Savage: ‘once second order probabilities are introduced, the introduction of an endless hierarchy seems inescapable. Such a hierarchy seems very difficult to interpret, and it seems at best to make the theory less realistic, not more’ (1954:58). De Finetti: ‘we have events and probabilities of events only; otherwise we would have the beginning of an infinite regression (probability of a probability, and so on)’ (1972:193). Consider, then, some statement of probability that de Finetti or Savage is prepared to make—say, ‘The probability that the coin lands heads is 1/2’. Now conjoin to it their skepticism about the notion of this statement in turn having a probability, and thus a fortiori, skepticism about this statement having a high probability: ‘The probability that the coin lands heads is 1/2, and I don’t assign this claim high probability (for this claim is not the sort of thing that has a probability at all)’. That is, we have a sentence of the form (7). Less radically, you might allow second-order, third-order, and perhaps still higher-order probabilities, but insist that the hierarchy must stop somewhere (again, perhaps because our heads are finite). Suppose, then, that you think that you have a highest-order probability assignment: an nth -order assignment, for some n > 2. Find, if you can, an nth -order probability assignment of yours—say, that it is raining. Then you should be prepared to assert: ‘My nth -order probability that it is raining is x, but I don’t assign high probability to the claim that my nth -order probability that it is raining is x
Alan H´ajek
225
(for that would be an (n+1)th order probability assignment, which exceeds my limit).’ Again, we have a sentence of the form (7).
‘There is no such thing as truth’ Paralleling our discussion of Churchland and Stich, now consider a philosopher who is skeptical of the very notion of truth. Let us suppose that he is prepared to assert sentences as usual, but he balks at assertions of the truth of sentences. Thus, a true⁷ Nietzschean ought to be prepared to assert sentences such as: ‘It is raining, but I believe that ‘it is raining’ is not true (for I reject the very notion of truth)’, a sentence of form (3). Similarly: ‘It is raining, but I don’t believe that ‘‘it is raining’’ is true (for I reject the very notion of truth)’, a sentence of form (4). Certain French philosophers and certain literary critics have a similar disdain, disrespect, or disregard for truth. They will pay for it similarly in Moorean ways (which is not to say that this is the worst of their problems). If someone were happy with the notion of (first-order) truth, but skeptical of the notion of higher-order truth, we could set them some Moorean bait: ‘ ‘‘It is raining’’ is true, but I don’t believe that ‘‘ ‘It is raining’ is true’’ is true (for that would involve higher-order truth, of which I am skeptical).’ And, much as before, we could also shanghai someone who thinks that there can be only n levels of higher-order truth, for some n > 2, with an nth -order truth-assertion of theirs that they do not believe is true. A skeptic about subjective probability—perhaps along the lines of Harman (1986)—could presumably be prepared to assert: ‘It is raining, and I don’t assign ‘‘it is raining’’ high probability (or indeed any probability, for I am skeptical about subjective probability)’, a sentence of form (7).
‘Truth is just what is useful for our purposes’ or ‘what is the consensus of our community’, or somehow ‘is relative’ We have just seen how to drive a wedge between an assertion and a belief in the truth of the assertion, by being unusually demanding about the notion of ⁷ I can’t resist this jab. Was Nietzsche a true Nietzschean in this respect? He seems to be when he writes: ‘Truths are illusions about which one has forgotten that this is what they are’ (Nietzsche 1994: 47). I thank Harold Langsam for this reference.
226
My Philosophical Position
truth. Going in the other direction, we could drive the wedge by being unusually undemanding about it. A certain kind of na¨ive pragmatist, for example, might translate talk of truth into talk of what is useful for our purposes. Thus, perhaps: ‘God does not exist, but I believe that ‘‘God exists’’ is true (it being useful for our purposes generally to act as if God exists)’. I dare say that this position is too na¨ive to take seriously. A bit more promising, perhaps, is the position of a rabid communitarian about truth: he thinks that the truth of a proposition p consists in a consensus of his community that p. He may find himself with a case of the Moores when he disagrees with such a consensus. For example: ‘God does not exist, but I believe that ‘‘God exists’’ is true (this being the consensus of my community).’ A certain kind of Peircean might append the words ‘. . . at the end of inquiry’, with similar results. Anyone who holds a relativist account of truth should strictly speaking make no sense of locutions of the form ‘p is true’. For by a relativist’s lights, there is no one-place predicate ‘ is true’ at all, but rather a two-place relation of the form ‘ is true relative to ’. The second argument-place will be filled in different ways by different relativists. A Protagorean may relativize an attribution of truth to a person; a Foucaultian may relativize it to a discursive formation; a Kuhnian may relativize it to a paradigm; a Quinean may relativize it to a theory or a language; a MacFarlanean may relativize it to a context of utterance or of assessment⁸ . . . The upshot is that a relativist should hold that an attribution of truth simpliciter is ill-formed, and thus not something that can properly be believed. Much as strictly speaking it is nonsensical to believe or to assert ‘Ren´ee is younger’ or ‘Los Angeles is west’, so it is nonsensical to believe or assert ‘p is true’, according to the relativist. And yet relativists seem to have no trouble believing or asserting things in the normal way (some of them, anyway). Thus, Moorean sentences should glide off their tongues: ‘It is raining, and I don’t believe that ‘‘it is raining’’ is true (being nonsensical, as it lacks a needed second relatum).’ To be sure, context often makes missing relata clear. In a conversation in which I am comparing my age to various people, you will have no trouble understanding me when I say ‘Ren´ee is younger’, because it is obviously elliptical for ‘Ren´ee is younger than me’. Likewise, if we are both standing in New York, I may permissibly say ‘Los Angeles is west’, leaving tacit the relativization to here, understood by both of us to be New York. That does not show that ‘younger’ or ‘west’ may suddenly become one-place predicates. Rather, context simply ⁸ See MacFarlane (2005).
Alan H´ajek
227
makes obvious what the second argument-place is. So to be sure, context may make obvious which person, or discursive formation, or paradigm, or theory or language, or context of utterance or of assessment . . . is salient. Then a relativist may permissibly say that something is true, it being understood to what this attribution of truth is being relativized. But this only shows that sometimes it is permissible not to speak strictly. And context does not always make clear the missing relatum. When it does not, and the empty argument-place remains unfilled, the result is nonsense. Or so should say the relativist—about ‘younger’, or ‘west’ (as we all are), and about truth (as are the Protagorean, the Foucaultian, the Kuhnian, the Quinean, the MacFarlanean . . .).
‘There are truth gluts’ We already saw with the preface paradox that someone can be prepared to assert something while believing that it is false. There, at least, the Moore-paradoxicality was in a sense diffused over a long conjunction of beliefs, with no individual conjunct being simultaneously asserted and believed to be false. But there are cases where the Moore-paradoxicality is far more localized—indeed, to a single sentence. Faced with stubborn paradoxes like those generated by ‘liar’ sentences (‘this sentence is false’, and its brethren), some philosophers embrace the existence of truth gluts —sentences that are both true and false. Priest (1987) is perhaps most famous for this view. Presumably, then, Priest will not bat an eye at saying: ‘The liar sentence is true, and I believe that the liar sentence is false (and true)’. Now, I suspect that he would hardly be troubled biting the bullet of asserting a Moore sentence, having already bitten the nuclear bomb of contradiction. And those of us who reject dialethism probably recoil at the very notion of there being true contradictions. Thus, you might call my drawing attention to these Moore-paradoxical commitments overkill. I prefer to call them icing on the cake. An approach to vagueness, common to both supervaluating and subvaluating, considers all the various permissible ways of precisifying a vague predicate. (See Hyde 1997.) Subvaluating treats a statement as true simpliciter if it is true according to any such precisification. (We will consider supervaluating shortly.) Suppose that Bruce is a borderline case of ‘tall’. Then there are some precisifications of ‘tall’ according to which he is tall, and others according to which he is not tall. The subvaluator then adjudicates him to be both tall and not tall. Moore awaits (as well as the nuclear bomb): ‘Bruce is tall (and not tall), and I believe that Bruce is not tall (and tall)’, or deleting the parenthetical conjuncts, ‘Bruce is tall, and I believe that Bruce is not tall.’
228
My Philosophical Position
‘There are truth gaps’ Another non-classical reaction to the ‘liar’ sentences has it that they are truth gaps —sentences that lack truth-value altogether. Various philosophers consider various other sentences to be truth gaps. Expressivists about moral discourse, such as Ayer (1946), regard moral claims such as ‘murder is wrong’ as expressions of disapproval, but not truth-apt. Such an expressivist, then, should happily say things of the form: ‘Murder is wrong (as I am wont to express), and I don’t believe that ‘‘murder is wrong’’ is true (since the sentence is not truth-apt)’. To be sure, ‘murder is wrong’ may not be an assertion according to expressivists (any more than ‘Boo to murder!’ is). Still, there is no denying that our example fits the Moorean template, and that expressivists should be prepared to utter it. Wittgenstein apparently has a similarly expressivist view about certain mental state reports, such as being in pain: ‘To say ‘‘I have a pain’’ is no more a statement about a particular person than moaning is’ (1958:67). Then he should not flinch at saying: ‘I have a pain, and I don’t believe that ‘‘I have a pain’’ is true’. Adams (1975), Edgington (1995), and Bennett (2003) argue that conditionals lack truth-values although they may well be assertible. They should thus be prepared to say things like: ‘If it’s raining then the ground is wet, and I don’t believe that ‘‘if it’s raining then the ground is wet’’ is true (since the sentence lacks a truth-value)’. Some authors believe that vague predicates generate truth gaps. On this view, if Bruce is a borderline case of ‘tall’, then the sentence ‘Bruce is tall’ lacks a truth-value. In particular, this sentence is not true (as well as being not false). Now suppose that it is permissible, although of course not required, to believe a sentence that one takes to be a truth gap. Then this has Moorish consequences. As it might be: ‘I believe (permissibly) that Bruce is tall, and ‘‘Bruce is tall’’ is not true (being a truth gap).’ Consider, for instance, the supervaluational approach to vagueness, which identifies truth simpliciter with truth according to all permissible precisifications, so-called ‘super-truth’. Suppose that ‘Bruce is tall’ is true according to some permissible precisifications (as must be the case, on this approach, if Bruce is a borderline case). Then arguably my belief can permissibly follow these precisifications. Thus: ‘I believe that Bruce is tall (permissibly, since he is tall on some admissible precisfications), and ‘‘Bruce is tall’’ is not true (since it is not super-true).’
Alan H´ajek
229
There are probabilistic analogues of this phenomenon. Suppose that your state of opinion cannot be represented by a single probability function—e.g., your probability for rain is vague over the interval [0.5, 0.75]. Following Levi (1974), Jeffrey (1983), and van Fraassen (1990), we may represent you with a set of probability functions, each of which precisifies your probability for rain with some sharp number in the interval. What is determinately true of your opinion is agreed upon by all the functions in this set. It is determinately true in the example that your probability for rain is at least 1/2. On the other hand, any statement about your opinion that is true according to some functions in the set and false according to others is indeterminate, and thus not determinately true—e.g. that your probability for rain is at least 0.6. Now suppose that there is a proposition p for which your probability assignment is vague over a wide interval—wide enough that according to the left-hand endpoint, p is not probable, while on the basis of the right-hand endpoint, it is assertable. For example, suppose that your probability of there being life on Mars is vague over the interval [0.4, 1]. According to some permissible precisifications of your opinion (e.g. 0.4), you do not assign high probability to ‘there is life on Mars’; thus you do not determinately assign high probability to ‘there is life on Mars’. According to others (e.g. 1), ‘there is life on Mars’ is assertible; thus, it is permissible for you to assert ‘there is life on Mars’. Conjoining these facts, we now get you to assert: ‘There is life on Mars (I permissibly assert), and I don’t (determinately) assign ‘‘there is life on Mars’’ high probability.’ We arrive at a sentence of form (7) once we drop the parenthetical reminders of how we got there. I said at the outset that there is apparently no way of interpreting someone who utters (7) as having a single probability assignment for p. I am offering a way of interpreting such a person if we may ascribe to them vague opinion, represented as multiple probability assignments for p. So far I have tried to impale various specific philosophical positions on Moorish sentences, suggesting that even if the philosophers who propound these positions don’t feel any discomfort there, the rest of us may well do so. But let us not get too smug. Perhaps I should not single out these philosophers. Perhaps they are in good company.
A PH I LO S O PH I C A L PE S S I M I S T I C M E TA - I N D U C T I O N Philosophy is a strange business. We have strong incentives to assert things, preferably in print (and the higher the profile of that print, the better), where our commitment to them is made even more manifest. This much is common to any intellectual discipline. But unlike many other disciplines, there is good
230
My Philosophical Position
reason to think that much that is asserted in philosophy is simply false. Moreover, we philosophers know this. We need merely remind ourselves how much philosophers disagree; at most one party to a disagreement can be right (assuming that the disagreement is genuine and the parties are not merely talking past each other, and ignoring dialethism). Even prior to such disagreement, we must acknowledge that philosophy is a subtle business, and that saying something philosophically interesting and true is no mean feat; yet what is philosophically uninteresting is less likely to make it to print (still less high-profile print), so we have incentives to stick our necks out. And whatever the explanation, there is surely the brute historical fact that many, and perhaps even most, substantive philosophical positions that have been offered are false. (Indeed, some of the positions, believe it or not, even have Moore-paradoxical commitments!) For every important philosophical position that you claim is true, I will respond with ten such positions that we agree are false.⁹ But I’m sure you don’t need me—you can do it yourself. Laudan (1981) offers a ‘pessimistic meta-induction’ concerning the truth of scientific theories, on the basis of the historical track record of science. A philosopher like you may well likewise run a pessimistic meta-induction concerning the truth of philosophical theories, on the basis of the historical track record of philosophy. And now here you are, advancing your own ambitious, bold philosophical position: p. You assert it vigorously, you defend it in (high-profile) print, and so on. But do you really believe it? Suppose that you must bet at high stakes on p’s truth, and that God will settle the matter. Still feeling confident in it? You may not think, after all, that you are that much more reliable than various philosophers who have come before you whose equally vigorous assertions have not withstood the test of time. And so your standards for assertion and for belief are sundered. You publicly assert p, maybe even in (high-profile) print, but if you are honest with yourself, you admit that you believe that p is probably false, or that you are at best agnostic about p; in any case you don’t believe that p. At such a reflective and reflexive moment, then, you may catch yourself asserting sotto voce: ‘My philosophical position says ‘‘p’’ and I don’t believe ‘‘p’’ ’. Not that you will ever say that in print. REFERENCES Adams, Ernest (1975), The Logic of Conditionals (Dordrecht: Reidel). Ayer, A. J. (1946), Language, Truth, and Logic (Middlesex: Penguin Books). Bennett, Jonathan (2003), Conditionals (Oxford: Oxford University Press). Churchland, P. M. (1981), ‘Eliminative Materialism and the Propositional Attitudes’, Journal of Philosophy, 78/2: 67–90. ⁹ Lewis’s list of philosophical positions with dubious credentials (1991:59) is a good start; see also Stove (1991), throughout.
Alan H´ajek
231
Edgington, Dorothy (1995), ‘On Conditionals’, Mind, 104/414:235–329. De Finetti, Bruno (1972), Probability, Induction, and Statistics (New York: John Wiley). De Rose, Keith (1991), ‘Epistemic Possibilities’, Philosophical Review 100/4:581–605. Harman, Gilbert (1986), Change in View (Cambridge, Mass.: MIT Press). Hawthorne, John, and Weatherson, Brian (2004), ‘Beliefs Old and New’, http://brian. weatherson.net/belief.pdf Hyde, Dominic (1997), ‘From Heaps and Gaps to Heaps of Gluts’, Mind 106/424: 641–60. Jeffrey, R. C. (1968), ‘Probable Knowledge’, in I. Lakatos (ed.), The Problem of Inductive Logic (Amsterdam: North-Holland Publishing Company), 166–80; repr. in Probability and the Art of Judgment (Cambridge: Cambridge Studies in Probability, Induction, and Decision Theory, 1992). (1983), ‘Bayesian with a Human Face’, in J. Earman (ed.), Testing Scientific Theories (Minnesota Studies in the Philosophy of Science, X; Minneapolis: University of Minnesota Press), 133–56; repr. in Probability and the Art of Judgment (Cambridge: Cambridge Studies in Probability, Induction, and Decision Theory, 1992). Kyburg, Henry (1961), Probability and the Logic of Rational Belief (Middletown: Wesleyan University Press). Laudan, Larry (1981), ‘A Confutation of Convergent Realism’, Philosophy of Science, 48/1: 19–49. Levi, Isaac (1974), ‘On Indeterminate Probabilities’, Journal of Philosophy 71/15: 391–418. Lewis, David (1976), ‘Probabilities of Conditionals and Conditional Probabilities’, Philosophical Review, 85/3:297–315. (1991), Parts of Classes (Oxford: Blackwell). MacFarlane, John (2005), ‘Making Sense of Relative Truth’, Proceedings of the Aristotelian Society, 105/3:321–39. Makinson, D. C. (1965), ‘The Paradox of the Preface’, Analysis, 25/6:205–7. Nietzsche, Friedrich (1994), The Portable Nietzsche, trans. Walter Kaufmann (Harmondsworth: Penguin Classics). Priest, Graham (1987), In Contradiction (Dordrecht: Martinus Nijhoff). Savage, Leonard (1954), The Foundations of Statistics (New York: John Wiley). Slote, Michael (1979), ‘Assertion and Belief ’, in J. Dancy (ed.), Papers on Language and Logic (Keele: Keele University Library). Sorensen, Roy (1988), Blindspots (Oxford: Clarendon Press). (2000), ‘Moore’s Problem With Iterated Belief ’, Philosophical Quarterly, 50/198: 28–43. Stich, Stephen (1983), From Folk Psychology to Cognitive Science (Cambridge, Mass.: MIT Press). Stove, David (1991), The Plato Cult and Other Philosophical Follies (Oxford: Blackwell). Unger, Peter (1975), Ignorance: A Case for Scepticism (Oxford: Oxford University Press). Van Fraassen, Bas (1990), ‘Figures in a Probability Landscape’, in J. M. Dunn and A. Gupta (eds.), Truth or Consequences (Dordrecht: Kluwer). Williamson, Timothy (1996), ‘Knowing and Asserting’, Philosophical Review, 105/4: 489–523. Wittgenstein, Ludwig (1958), The Blue and Brown Books (New York: Harper and Row).
11 Moorean Pretense Robert M. Gordon
Wittgenstein noted that the absurdity of assertions of the form, p but I don’t believe that p, does not carry over to suppositions. He was right. The absurdity also does not carry over to what I take to be the more inclusive category, pretense.¹ For example, you can pretend—or suppose or imagine—that a certain species of mushroom is poisonous but that you are ignorant of this fact. You might even picture yourself innocently saut´eing your mushrooms, then eating them and getting very sick. Or you can imagine walking down the crowded aisle of a busy department store, not looking where you are going, and walking into a supporting column in the middle of the aisle. You imagine: I am walking right into the column, but I don’t believe I am. By a Moorean Pretense, I mean a pretense that has at least one premise of either the ‘omissive’ form already mentioned, p but I don’t believe that p, or the ‘commissive’ form, p but I believe that not-p.² I speak of the premises of a pretense because pretense is plausibly construed as having an inferential structure. For example, when young children pretend that certain globs of mud are cherry pies, they are typically able to use that identification, together with additional information drawn from observation, The author is much indebted to Mitchell Green and John Williams for comments on an earlier version of this paper. ¹ The intuition behind calling it more inclusive is roughly this: When S pretends that p, S intentionally acts in some way as if p. The action may be limited to arguing (e.g., ‘supposing for the sake of argument’). There is of course also a cognitive or epistemic condition: perhaps, ‘S does not believe that p’ or perhaps only, ‘S is not certain that p.’ For example, during an athletic event, athletes may find it useful to pretend they are executing a perfect performance. They may well believe they are doing so—they are just not sure. ² See J. N. Williams, ‘Moore’s Paradox—One or Two ?’, Analysis, 39/3 (1979), 141–2 and Chapter 1 in this volume.
Robert M. Gordon
233
to answer such questions as, ‘How many cherry pies are there?’ and, ‘Which cherry pie is biggest?’³ To answer the ‘How many’ question, for example, the relevant inference might be, ‘Each mudpie is a cherry pie [pretended premise], there are three mudpies [empirical finding], therefore there are three cherry pies [pretense-bound conclusion].’
1 . T H E I N D E X I C A L C H A R AC T E R O F M O O R E A N ABSURDITY Before discussing pretense, it is important to be clear about the indexical character of Moorean absurdity. The belief clauses, I don’t believe that p I believe that not-p contain the first-person indexical ‘I’ and implicitly the temporal indexical ‘now’, signified by the present tense. For an assertion to be Moorean-absurd, the belief clause must contain these or relevantly similar indexicals. Consider a contrasting case: You are walking down a crowded aisle, peering intently at a video monitor as you go. On the monitor you see people walking down a crowded aisle, and you presume correctly that you are watching a live real-time video of the aisle you are in. Suddenly you point to the monitor and remark, That person doesn’t know it, but (s)he is about to walk into a column! Or, to force the point, suppose you say, That person is about to walk into a column, but (s)he doesn’t believe (s)he is! If I happen to know that in fact you are the person in the monitor, I may find your remark comically absurd in context, but I do not find it Moorean-absurd. There is nothing inconsistent or self-stultifying about asserting, p but that person doesn’t believe that p.⁴ Although the person you are referring to as not believing what you have just stated is yourself, you are not referring to that person indexically as yourself. The corresponding point holds for the temporal indexical. Suppose you recognize yourself in the video but do not appreciate that it is a live video: I didn’t know it at the time, but I was about to walk into a column. I was about to walk into a column, but I didn’t believe I was. ³ See K. L. Walton, ‘Pictures and Make-Believe’, Philosophical Review, 82/3 (1973), 283–319. ⁴ Even referring to yourself by name would be insufficient for Moorean absurdity, for it would leave open the possibility that you do not know that the name refers to you.
234
Moorean Pretense
These are not examples of Moorean absurdity either.⁵ Similar indexical requirements hold for Moorean Pretense. A premise of the form, p but that person doesn’t believe that p, does not make a pretense Moorean. Nor can one convert it into a Moorean Pretense merely by identifying the person who doesn’t believe that p with oneself, the person who is doing the pretending. For example, I mentally picture someone who looks like me walking unknowingly into a column, and I think, ‘That’s me I’m picturing walking unknowingly into a column.’ This is not a case of Moorean Pretense, because the indexical ‘I’ is introduced only in my comment on the pretense, not within the pretense itself. A pretense is Moorean only if there is an ‘I’ within the scope of the pretense. It should be noted that an external identification with the ignorant nonbeliever is not only not sufficient for Moorean Pretense; it is also not necessary. That is, it is not necessary that the ‘I’ be understood to refer to myself, the very person who is doing the pretending. I can pretend to be someone else, a real or fictitious individual other than the person who is doing the pretending. I can also pretend the time to be other than the time of the pretending. The ‘I’ and ‘now’ within the pretense need not be co-referring with the ‘I’ and ‘now’ outside the pretense. That is, they need not refer, respectively, to the pretender and the time of the pretending. What matters is not what is referred to but the mode of reference, the I-now character of the reference.
2. THE DIVIDED PRETENDER Psychologically there is a major difference between pretending, That person doesn’t believe that p, and pretending, ⁵ A curious problem arises as it dawns on you that the unfortunate person on the monitor is actually you in realtime. Suppose you make this inference: That person is about to walk into a column, but (s)he doesn’t believe she is! Wait, that’s me! I’m about to walk into a column, but I don’t believe I am. (Or: I’m the one who is about to walk into a column, but doesn’t believe it.) Would your conclusion be Moorean-absurd? I have some inclination to say no, as long as the belief denial is strictly evidence-based: (Evidently) I don’t believe I am. (I would be similarly inclined where one’s assertion of the belief conjunct in a ‘Moorean’ sentence is based solely on neuroscientific evidence, such as results from brain imaging.) This is an unstable situation, however; once the ‘I and now’ identification is made, people would and should abandon the belief denial; otherwise, it would lead to absurdity.
Robert M. Gordon
235
I don’t believe that p. What is the difference? The latter pretense requires that the ignorance or nonbelief be represented in a ‘first person’ way, by pretending not to believe that p. Likewise, pretending, I believe that not-p, requires pretending to believe that not-p. When I pretend to believe or not to believe something, I thereby modify the set of ‘facts’ that my decision-making within the pretense will have access to. For example, if I pretend to believe that the mushrooms on the plate are poisonous, then within the pretense I might refrain from eating them on the grounds that they are poisonous. If I pretend to believe that there is an obstacle in my path, then within the pretense I can plausibly change direction because there is an obstacle in my path. Not only actions, but also emotions, desires, and other beliefs will have access to this ‘fact’. If on the other hand I pretend not to believe that p, then I will not do anything, for example, change directions, on the basis of the ‘fact’ that p; nor will I form emotions, desires, or further beliefs on that basis. In short, one pretends that I believe (do not believe) that p only if one pretends to believe (not believe) that p; and pretending to believe (not believe) something constrains the facts on the basis of which one can act, emote, or make inferences within the pretense. Such constraints pose a psychological problem for Moorean Pretense. If one pretends p but it I don’t believe that p, then one pretends that p in a special way: a way that makes the pretend ‘fact’ that p unavailable to the ‘ignorant’ agent one plays. The pretend fact that p must somehow be cordoned off so that it cannot move one to action or emotion or lead one by rational inference to form new beliefs. Thus Moorean Pretense requires the pretender to represent the world twice over: once as the ‘objective’ or ‘outer’ world, which includes the pretend-fact that p; and once as the ‘subjective’ or ‘inner’ world, which excludes the pretend-fact that p. In a commissive Moorean Pretense one would feed contrary or contradictory premises into the two pretenses. In the mushroom example, one feeds into the outer pretense the premise, F. These mushrooms are false morels. However, one feeds into the inner pretense, as a possible basis for action, emotion, and other beliefs, the contrary premise, T. These mushrooms are (true, genuine) morels,
236
Moorean Pretense
Only the outer pretense has access to F, and only the inner has access to T. There is no leakage of information from one to the other; each is informationally insulated from the other. Compartmentalization has its limits, however. Consider the clown who plays the part of an innocent person walking, eyes averted, into a column—feigning ignorance, yet cognizant that unless he swerves from his path he will indeed walk into a column. It is sometimes difficult to feign such ignorance. A novice might telegraph his prescience by visibly decelerating as he nears the column, and by showing anticipation on his face. A seasoned performer, on the other hand, somehow manages to hide his knowledge from himself, so that it does not show in any readily observed emotions or actions. His finesse helps his audience engage in the two tasks as well: to project themselves into the clown’s innocence and yet at the same time to be fully clued in. For certain types of action, it appears to be not just difficult but humanly impossible to prevent our knowledge from influencing what we do. Discovering what we already know seems to be such a type. Consider a two-player game in which Player A hides something and Player B tries to discover or at least to guess correctly what it is. The hidden object may simply be what A is ‘thinking of ’, such as a particular number. Or it may be the particular coordinates of the squares that A has filled in on a hidden grid. In the Battleship game, the squares represent ‘ships’ of various types, each consisting of a certain number of contiguous filled-in squares in a straight line. Player B’s goal is to hit (bomb, torpedo) all of the squares comprising each of the opponent’s ships. B calls out the coordinates of the targeted square, and A replies by indicating whether the shot is a hit or a miss, and if a ship was hit, what type of ship was hit. Obviously it is important that A’s placement of the ships be hidden from B. (A similar game is the Windows game Minesweeper, which pits the human player against the computer.) Suppose one were to try to play solitaire Battleship or solitaire Minesweeper. One and the same human being would be both A the scene-setter, who sets up the ships or the mines, and B the scene-player, who tries to locate A’s ships or mines. The aim of the solitaire game would be to discover (as Player B) something you already consciously know (as Player A). Barring certain pathologies or a time interval long enough to allow one to forget what one did in setting the scene, such a task would seem at least psychologically (if not logically) impossible. For nothing set up by Player A will be hidden from Player B: Knowledge will ‘leak’ from scene-setter to player. Even more obviously, there appears to be no possible solitaire version of the game of ‘Guess what number I am thinking of.’ Moorean Pretense need not be as difficult as playing a game that requires that we discover something we already know. I can play in my mind the innocent morel-eater, knowing at the same time that I am the unknowing dupe of nature’s deceit. Nonetheless, pretending that p and simultaneously pretending not to believe that p (or to believe that not-p) is pretense of a sophisticated sort.
Robert M. Gordon
237
3 . FA L S E B E L I E F I defined a Moorean Pretense as having at least one premise of either the ‘omissive’ form, p but I don’t believe that p or the ‘commissive’ form, p but I believe that not-p. Thus far, I have spoken as if the pretender were pretending each of the conjuncts to be true—that is, merely pretending. However, it is often the case that one of the conjuncts is imported into the pretend world from the actual world—that is, the world as the pretender actually believes it to be. Here are three distinct types of Moorean Pretense: 1. (ascription-preserving) I pretend that p, carrying over into my pretense my knowledge or belief that I don’t really believe that p (or do believe that not-p). Here my inner pretense, but not my outer, is simply carried over from the actual world. 2. (world-preserving) I pretend that I don’t believe that p (or do believe that not-p), carrying over into my pretense my knowledge or belief that p. Here my outer pretense, but not my inner, is simply carried over from the actual world. 3. (non-preserving) I pretend both that p and that I don’t believe that p (or do believe that not-p), carrying neither conjunct over from what I actually know or believe. Here neither pretense, inner or outer, is carried over from the actual world. Type 1 (ascription-preserving) Moorean Pretense is not uncommon. Philosophers and scientists pride themselves on asking themselves, ‘What if I am (we are) wrong?’ Sometimes this is a matter simply of pretending something contrary to what they actually believe, where it is not essential that at the same time they continue to believe as they do. But in other cases it is essential that they imagine continuing to believe as they do—and being wrong. For example, they may want to assess the cost of acting on a false positive relative to that of acting on a false negative. Accordingly, they carry out in their mind an action plan based on their actual belief and imagine the consequences likely to ensue if the belief is false. Descartes’s pretenses in the Meditations are also ascription-preserving Moorean Pretenses. He tries to persuade himself that nothing he currently believes is so, that the physical world he seems to perceive does not in fact exist, and that nothing he seems to remember is true. However, along with this pretense, he imagines himself continuing to believe as he actually does:
238
Moorean Pretense
I will consider myself as without hands, eyes, flesh, blood, or any of the senses, and as falsely believing that I am possessed of these. (Meditation I, paragraph 12)
Descartes’s beliefs within the pretense do not match the facts that shape the pretend world within which he holds these beliefs. It was this Moorean Pretense that made it possible to introduce the cogito and, in its wake, three of the major issues in modern philosophical thinking about the mind: the mind–body problem, the problem of phenomenal consciousness, and the problem of other minds.⁶ Type 2 (world-preserving) Moorean Pretense may be even more common. It may underlie our ability to anticipate and explain the behavior of people whom we know to be (or perhaps intend to be) mistaken about their environment. Developmental psychologists have devised a number of tests of the ability of children of various ages to recognize false beliefs in others. In the most famous of these, that of Wimmer and Perner, the child subject observes two puppet characters, trustful Sally who puts her marble away in her toy box and covetous Anne who transfers it to her own basket while Sally is out of the room.⁷ When Sally wants to play with her marble again, where will she go to get it? An adult or an older child is likely to say, Sally will go to her toy box, where she left the marble. Asked why she doesn’t go to Anne’s basket, where the marble actually is, they might answer, Sally was unaware that the marble had been transferred to Anne’s basket. She was unaware of this because she was out of the room at the time and therefore didn’t see it being moved. Young children, on the other hand, don’t seem to get it. Asked where Sally will go to get her marble, they will point to Anne’s basket, evidently using their own awareness of where the marble actually is. Wimmer & Perner and numerous other experimental studies have shown that nearly all children make incorrect predictions—until about age 4, when something seems to click, and nearly all get it right. What is it that clicks? According to the ‘simulation’ account I have given elsewhere⁸, it is the capacity for a kind of pretense—what I would now call Type ⁶ David Rosenthal has drawn a connection between Descartes’s cogito and Moore’s paradox in a number of his writings. The distinction between types of Moorean pretense is my own. ⁷ H. Wimmer and J. Perner (1983), ‘Beliefs about Beliefs: Representation and Constraining Function of Wrong beliefs in Young Children’s Understanding of Deception’, Cognition, 13/1: 103–28. ⁸ I have argued for such an account, although I did not label the kind of pretense involved as ‘Moorean Pretense’, in several publications, including ‘Folk Psychology as Simulation’, Mind and Language, 1/2 (Summer 1986), 158–71; repr. in W. Lycan (ed.), Mind and Cognition: An Anthology 2nd edn. (Oxford: Blackwell Publishing, 1998). A similar view has since been adopted by a number of developmental psychologists and has led to numerous experimental studies. Most of the early contributions by philosophers and developmental psychologists to the ‘simulation vs. theory’ debate first appeared in Mind and Language, 7/1–2 (Spring–Summer 1992), Special Issue on ‘Mental Simulation: Philosophical and Psychological Essays’); repr. in M. Davies and T. Stone (eds.), Folk Psychology: The Theory of Mind Debate (Oxford: Blackwell, 1995). For some of the later interdisciplinary work on the topic, see P. Carruthers and P. Smith (eds.), Theories of Theories of Mind (Cambridge: Cambridge University Press, 1996).
Robert M. Gordon
239
2 Moorean Pretense.⁹ Suppose the child subject decides where Sally will go to get her marble by mentally playing the role of Sally. As Sally, the child wants to get her marble, and the place to get it, as the child knows, is Anne’s basket! To move beyond this automatic application of what the child knows, what must be pretended is the following: M. Sally’s marble has been moved to Anne’s basket, but I, Sally, don’t believe it has. The logical form of M is, of course, p, but I don’t believe that p. Again supposing that the child subject answers this question, ‘Where will Sally go to get the marble?’ by role-playing, it is essential that she be capable of compartmentalized reasoning. Otherwise, she would fail to cordon off the information she gained as spectator while Sally’s head was turned. In that case, she would make the incorrect prediction that Sally will look for her marble where it actually is, namely, in Anne’s basket. Does the empirical evidence support the hypothesis that in human beings the capacity to ascribe false beliefs depends on the capacity for Type 2 Moorean Pretense? Although far from conclusive, the evidence does offer some support. One relevant empirical issue concerns neuropathologies that deprive people of the capacity for Type 2 Moorean Pretense. It is well established that autism is associated with a notable lack of spontaneous pretense, particularly sophisticated forms such as Type 2 Moorean Pretense. A number of experimental studies have shown that people with autism generally exhibit a profound deficit in the capacity for false belief ascription.¹⁰ Nonetheless, some people with autism do manage to ‘pass’ various false belief tests. Can the same individuals also carry out sophisticated pretense? Do they, perhaps, find ways to ‘cheat’ on these tests, providing the right answers without invoking the concept of belief? Further, even if it is likely that a causal connection explains the coupling of these twin deficits in most people with autism, it is not yet clear which way the connection flows. Perhaps the lack of Type 2 Moorean Pretense explains the failure to ascribe false beliefs; perhaps the failure to ascribe false beliefs explains the lack of Type 2
⁹ I also believe that the capacity for Type 1 Moorean Pretense is needed, too, for ascribing belief to oneself —not just saying ‘I believe’ as a prefix to an assertion, but understanding one’s utterance as a genuine ascription of belief. I develop this point, among others related to belief ascription and Moorean Pretense, in a paper to appear in a special issue of Synthese on the topic, ‘Self-Ascriptions of Attitudes’. For an earlier application of Moore’s Paradox to self-ascriptions of belief, see my paper, ‘Sellars’s Ryleans Revisited’, Protosociology: An International Journal of Interdisciplinary Research, 14 (2000), 102–14. ¹⁰ For a fuller application of the simulation theory to autism, with some references to the empirical literature, see R. M. Gordon and J. Barker, ‘Autism and the ‘‘Theory of Mind’’ Debate,’ in G. Graham and L. Stephens (eds.), Philosophical Psychopathology: A Book of Readings (Cambridge: Mass.: MIT Press, 1994), 163–81.
240
Moorean Pretense
Moorean Pretense; and perhaps both deficits are connected only by having a common cause. Another relevant empirical issue concerns the age of onset of the capacity for Type 2 Moorean Pretense in children who are not autistic. Obviously, if the capacity to ascribe false beliefs depends on the capacity for Type 2 Moorean Pretense, it must not emerge earlier. However, methodological issues cloud the issue. Not only is it difficult to establish when such pretense begins; there has also been disagreement about the age at which genuine false belief ascription emerges.¹¹ Pretense of a less sophisticated sort usually becomes evident before age 2, long before the child can master false belief tests such as the Sally–Anne task.¹² However, the crucial and more difficult question concerns the capacity for Type 2 Moorean Pretense. Although I am not aware of any study that specifically asks when children begin to engage in such pretense, my bet is that it is not later than the age of false belief ascription. Finally, in Type 3 (non-preserving) Moorean Pretense, I pretend both that p and that I don’t believe that p (or do believe that not-p), carrying neither conjunct over from what I actually know or believe. The fantasies described in the first paragraph may be construed as examples of this type. In the inner pretense I am saut´eing ordinary non-poisonous morel mushrooms, and in the outer pretense I am saut´eing poisonous false morel mushrooms. Actually, I am doing neither. Or, in the inner pretense I am walking down the aisle with no obstacle in the way, and in the outer pretense I am walking into a supporting column in the middle of the aisle. Actually, I am merely sitting at home, fantasizing. Neither the premise of the outer pretense nor that of the inner pretense is carried over from the actual world of the imaginer.¹³ When a Moorean Pretense is acted out, whether in stage or film acting or in children’s pretend play, it is typically of Type 3. The actual world does not furnish the premise of either the inner or the outer pretense. For example, suppose, as before, children pretend certain mudpies to be ordinary cherry pies. However, that is just the inner pretense. The children also share an outer pretense: that one of these pretend-pies is a pie with secret magical powers. Eating such a pie ¹¹ However, the 4-year-old chronology does seem to be holding its own. See H. Wellman, D. Cross, and J. Watson, ‘Meta-Analysis of Theory of Mind Development: The Truth about False Belief ’, Child Development, 72/3 (2001), 655–84. ¹² See P. L. Harris and R. D. Kavanaugh, ‘Young Children’s Understanding of Pretense’, Monographs of the Society for Research in Child Development, 58/1 (1993). ¹³ Whether these examples should be classified as Type 3 Moorean Pretenses depends, strictly speaking, on how the premise of the inner pretense is stated. If it is, ‘I am walking down the aisle of a department store, and not walking into a column’, then it is of course not true of someone who is merely sitting at home, fantasizing. However, if the premise is stated merely as, ‘I am not walking into a column’, then it is, strictly, true of the imaginer. Likewise, ‘I am eating ordinary (non-poisonous) morels’ is not true of the imaginer, but ‘I am not eating poisonous mushrooms’, is. Although this relativity is worth noting, I think it can be fixed with a little more work. (I noticed it just before going to press.) I am sure it raises no important problem for my distinction between three types of Moorean Pretense.
Robert M. Gordon
241
immediately transforms one into a donkey. The children innocently pretend-eat the other pies, to no bad effect. Then they partake of the magic pie; and in a moment, acting out the sequelae of unseen forces, they begin to walk on all fours and bray. The mudpies are of course neither cherry pies nor magic pies. Neither of these premises is a holdover from the actual world. Moorean Pretense, particularly Type 3 Moorean Pretense, appears to be a defining property of dramatic irony. For example, in Oedipus Rex, Sophocles secures the audience’s complicity in a double pretense. We pretend that the actor who plays the protagonist is a man who has slain his father and married his mother, two events that mark him for destruction. We also pretend to see the world through the protagonist’s innocent eyes: The king he slew was not his father and the queen he married was not his mother. The audience is at once clued in and empathetic. It shares in the outer pretense that sets the scene for the action, and it shares in the inner pretense that allows the actor to get behind his scripted behavior.
4 . S U M M A RY Moorean Pretense is not absurd. It underlies Cartesian doubt and is rampant in modern science; it may underlie the ascription of false belief, and it is a defining feature of dramatic irony. Although not absurd, it is psychologically complex. Its complexity hinges on a feature that is also crucial to the Moorean absurdity of assertions: namely, the I-now character of the belief clause, I don’t believe that p, or, I believe that not-p. In an omissive Moorean Pretense, one is pretending ignorance of a premise of one’s own pretense, namely that p. In a commissive Moorean Pretense, one is pretending that a premise of one’s own pretense is false. What keeps the commissive pretense from being not only absurd but self-contradictory is the capacity to compartmentalize. Although one is at the same time pretending that p and pretending that not-p, one assigns these contradictories to distinct logical spaces, one space containing the world and the other, the world relativized to a particular point of view.
This page intentionally left blank
Index of Names Abelard 42 Adams, E. 228 Adler, J. 3, 30–1 Albritton, R. 144 Allen, W. 195 Alston, W. 71 Atlas, J. 29–30, 126 Armour-Garb, B. 30–1 Audi, R. 22, 106 Augustine 29, 41–3 Ayer, A.J. 228 Bach, K. 126 Baldwin, T. 18, 25, 30 , 76–8 Benaccerraf, P. 143 Bennett, J. 228 Bentham, J. 42 Berry, G.G. 37 Black, M. 14, 23 Bo¨er, S. 120 Bonjour, L. 71 Brandom, R. 27 Brentano, F. 18, 20 Bront¨e, C. 181 Buridan, J. 38 Burge, T. 85 Bu˜nuel, L. 31 Burnyeat, M. 42 Cantor, G. 37 Carroll, L. 44 Casta˜neda, H. 118, 136 Church, A. 38 Churchland, P. 219–21, 223, 225 Collins, A. 3, 13 Crimmins, M. 32 Da Vinci 44 Davis, W. 195 Davison, A. 135 De Finetti, B. 224 De Almeida, C. 16, 30 De Morgan, A. 61 Dennett, D. 84 DeRose, K. 4, 14, 218 Descartes 29, 33, 38, 43–4, 47, 85, 142–4, 237–8, 241 Doran, K. 27 Dretske, F. 70
Edgington, D. 228 Emerson 42 Evans, G. 16–17, 31, 92, 94–7, 100–2, 104, 107–9 Fitch, F. 37 Foucault 226–7 Foley, R. 65, 71 Galileo 81 Gallois, A. 3, 32 Geach, P. 46 Gettier, E. 65 Goethe 45 Goldstein, L. 13 Gombay, A. 3 Gordon, R. 4, 29, 33 Green, M. 3, 30, 31, 195 Grice, P. 77, 195 H´ajek, A. 31–2 Halpern, B.S. 47 Harman, G. 225 Hawthorne, J. 221 Heal, J. 3, 13, 17, 31, 80, 84–85 Hegel 38 Hintikka, J. 28 , 143 Hume 84 Hunter, J.F.M. 135 Hyde, D. 227 Jacobson, R. 12 Jeffrey, R.C. 220–1, 229 Jones, O.R. 26 Jourdain, P. 37 Kant 39, 45, 223 Kepler 44 Klein, P. 16, 65 Koestler, A. 81 Kriegel, U. 18–20 Kripke, S. 133 Kuhn 226, 227 Lakoff, G. 141–3 Langford, C.H. 7, 140–2 Laudan, L. 229 Lehrer, K. 65
244
Index of Names
Levi, I. 229 Levinson, S.C. 13, 126 Lewis, C.I. 40 Lewis, D.K. 118, 187, 218 Linville, K. 12 Locke, 85 Lycan, W. 120, 135 MacDonald, M. 37 MacFarlan, J. 226–7 Maclver, A.M. 140 McTaggart, J.M.E. 38 Makinson, D. 159 Malcolm, N. 6, 12, 165 Mariotte, E. 44–45 Martinich, A. 13 Merton, R. 37 Milgram, E. 3 Moore, G.E. 3–6, 9, 29, 37–8, 46–8, 53, 76–7, 85, 88, 90–1, 128–9, 155, 165 Moran, R. 4, 29, 128, 133–6 Morris, C. 141 Murdoch, I. 135 Nietzsche 225 Nozick, R. 70 Ockham 42 O’Connor, D.J. 39 Orwell, G. 81 Packer, A. 200 Parmenides 29, 39, 43 Pecquet, J. 45 Peirce, C. 226–7 Perner, J. 238 Perry, J. 118 Plato 29, 40, 42 Priest, G. 227 Prior, A.N. 118 Protagoras 226–227 Putnam, H. 135 Quine, W.V.O. 117–8, 120, 124, 226–7 Ramsey, F. 44 Reagan, R. 195 Reid, T. 85 Rhys, J. 181
Ring, M. 12 Rosenthal, D. 4, 15, 23, 31–2, 137 Russell, B. 37–38, 46–47, 57 Salerno, J. 38 Sanford, D. 46 Sartre, J.P. 134 Savage, L. 224 Schopenhauer, A. 44–46 Searle, J. 21–22, 24, 31, 106–7, 109, 135 Sellars, W. 139–140 Sextus, E. 41 Shoemaker, S. 3, 12, 23–4, 28, 30, 137–140, 142–4, 205 Slote, M. 218 Smith, F.E. 41, 157 Sophocles 241 Sorensen, R. 3, 6, 21–2, 29, 31, 91–2, 104–7, 109, 111, 223 Sosa, E. 65, 71 Spinoza 29, 43 Stevenson, C.L. 48, 165 Stich, S. 219–21, 223, 225 Stigler, S. 37–8 Stoljar, D. 31 Strachey, R. 47 Strawson, P. 196 Stroud, B. 84 Szabo-Gendler, Z. 4 Tversky, A. 14 Unger, P. 14, 146, 154–5, 218 Vahid, H. 17 Van Fraassen, B. 229 Vanderveken, D. 24, 31 Weatherson, B. 221 Welbourne, M. 26 Wiener, N. 38 Williams, B. 44 Williams, J.N. 3, 15–17, 27–8, 30–2, 53–8, 76–7, 92, 212 Williamson, T. 14, 32, 155, 218 Wimmer, H. 238 Wittgenstein, L. 3, 4, 6–8, 12, 44–5, 47, 122, 134, 146, 154, 165, 190, 228, 232 Wolgast, E. 24
Subject Index absurdity 5–8, 10–13, 15, 21, 25–7, 29, 62–71, 73, 76, 78, 90, 92, 100, 106, 109, 137, 190–3, 192, 194, 202–10, 212, 219, 232–3, 241. commissive 5, 8, 17, 19, 20, 21, 31, 91–2, 107–9,112, decrease of 20, 31–2, 92, 104, 109, 112–13 examples of 3, 5, 8, 21, 30–1, 56, 72, 76, 90–1, 104, 147–8, 152, 153–5, 161, 166–7, 190, 192–3, 210, 211–12, 233. grammatical characterisation of 29, 56, 189 Moorean 4, 9, 10–12, 14, 19, 22, 29–32, 44, 48, 53–8, 77–80, 82, 84–8, 90–2, 100, 107, 110–112, 189–91, 203–6, 210–13, 233–4, 241 ommissive 6, 8, 17, 20–1, 31, 91–2, 101, 104–5, 107–9, 112–13 paradigmatic 91, 104 acceptance 153–4, 212 agnosticism 13, 81, 173, 220 alienation 134 all-seeing eye 29, 37, 42 amnesia 130–2, 199 antinomies 223 apartheid epistemologies 71 assent 198, 205 conscious 139 manifest 139–40 assertability 154–160, 222 assertion 10–14, 26–8, 30, 40–1, 55, 82, 96, 111, 118, 125, 128, 132–3, 136–7, 146–150, 153–4, 158, 166, 169, 190, 193, 196, 201, 207–8, 210–12, 218–22, 224–5, 228, 230, 232 absurdity of 3, 5, 7–8, 18, 32, 86, 88, 90–1, 110, 207, 190 and belief 1, 3, 7, 12, 30, 41, 43, 77–8, 82–3, 85–6, 88, 121, 137, 140, 143, 146–7, 148–9, 154, 156, 167, 161, 202–3, 206–7 and expression 12, 15, 24, 128–40, 146, 148–9, 153, 160 commissive 5, 27, 31–2, 86, 91–2, 110, 112–13 context of 26, 27, 158, 131, 226–7, 191 its distribution over conjunction 12, 10, 14, 23, 26, 67, 110, 158, 193, 204, 206–07 and implication 4–6, 8, 10, 13–14, 76–8, 86–7, 90, 146
logic of 7, 8, 143, 118, 153 Moorean 4, 5, 11–13, 23–7, 53, 78–9, 82, 86–7, 92, 110–112, 137, 142, 149, 154, 155, 210–211, 233, 241 omissive 5, 13, 26–7, 31–2, 43, 86, 91–2, 112–113 purported 143, 156, 206 rational 142–3, 202, 207, 210 sincere 11, 93–4, 112, 139, 141–3, 148, 201–3, 206–8, 211–12 uttering assertively 5, 129, 131, 137, 140, 142–3, 194 attitude-related reasons 148, 173–5, 178, 183, 185 authority 92–3, 96–7, 156–7 autism 239 Bayesianism 220–1 belief 174, 178, 183–4, 190–3, 195, 197, 208, 219–25, 227, 230, 233, 241 and closure 11, 21, 61, 65, 69, 105 concept of 99, 106 conscious 78–9, 165–7, 171–2, 175–80, 182, 185–8, 201, 204–5 commissive 91, 92, 101, 102, 109, 204–5 de dicto 59, 120, 126–7, 143 degree of 152, 159 de re 119–20, 126–30, 133, 136, 143 de se 119–20, 126–30, 133, 136, 143 dispositional 22, 106 its distribution over conjunction 54–6, 58, 61, 140, 142, 149, 167, 170–2, 204, 206, 207 and the distribution of justified belief over conjunction 99–102, 107, 109 expression of 55, 81, 128, 135, 139–40, 190, 193, 222, 224, 228 elimination 21, 105 higher-order 79–80, 144 and iterated operators 167, 170, 180, 182, 186, 187 manifest 77, 139, 196, 199–201 meta-linguistic 131 Moorean 53–57, 64–73, 77–9, 84, 86, 91, 93, 96, 97, 101–04, 204–5 notional sense of 117 and occurrent thought 21, 22 omissive 10–11, 16–21, 25, 28, 91–2, 104–5, 107, 109, 112, 204–5
246
Subject Index
belief (cont.) and principle of elimination 21, 100, 102, 105, 108 rational 55, 72, 142–3, 171, 179, 191, 194 relational sense of 117–18, 124 reporting 7, 12, 135, 137, 193 and Searle’s principle 106–7, 109 second-order 92, 98–9, 101, 107 self-ascriptions of 13 blindspot 38, 44–5 Cartesian doctrines 33, 38, 43–4, 47, 85, 142–4, 237–8, 241 coherence 12, 15, 45, 53, 58–61, 63–7, 70, 73, 78–9, 94, 99, 101, 124, 134, 159 commitment 61, 135–6, 144, 196–202, 205, 209, 212, 217, 219–20, 227, 229–30 communitarian 226 community 199, 225–6 conditionals 143, 209, 228 consciousness 150, 165–6, 175, 179, 182, 185, 189, 199, 201–2, 212, 236, 238 consensus 225–6 content-related reasons 173–5, 178, 183–5, 187 contradiction 90, 99–103, 105, 108–9, 190, 192, 227 counterevidence 16, 58–9, 64, 67, 70 deletion 166, 178–80, 185 deliberative perspectives 133 delusion example 200 dialectical games 40 discourse 118, 226–7 dissociation 6, 133, 134 eliminativism 32, 166, 221 EP test 57–8 epistemic goal 55 epistemic probability 59 Evans’s principle 15–17, 92, 94, 96–7, 99, 101, 102, 104, 107–8 expressivism 228 fact/value distinction 135 fallibility 55, 57, 85, 91, 98, 101–2, 157–60, 194, 201, first-person point of view 122, 149, 154, 162 folk psychology 219, 221 foundationalism 66, 70 functionalism 3, 84 God 13, 21, 24, 31, 42–4, 46, 91–2, 104, 105, 109, 112, 177–8, 226, 230 grammaticality 14, 29, 119, 122–6, 130, 135–6, 189, 206, 209
higher-order thought 137 idealism 37–8, 42, 46–7 illocutionary force 135, 146, 208 inconsistency 25, 28, 30, 40–1, 43, 47, 54–5, 59–60, 62, 63, 77, 79, 86–7, 103, 131, 137–8, 141–3, 147, 150–1, 157–9, 167–8, 170–2, 187–8, 233 Strong versus weak 63, 67 incorrigibility 141–2 indeterminacy 229 indexicals 191, 233–4 infinite regress 80, 144–5, 155, 172, 224 Inner and Outer 7, 134–6, 198–9, 235–7, 240–1. Integration 32, 134, 166, 178, 180–3, 185–8 intention 25–8, 42, 77–8, 80, 82, 84, 92, 110–11, 132, 174–5, 185, 190, 195–7, 199–201, 212 introspection 42–3, 47, 55, 63, 93, 96, 98, 134, 201, 204 irrationality 53, 55–6, 59–61, 101–3, 107, 166, 172 judgment (judgement) 3, 6, 10, 80, 85–8, 107, 113, 152, 154, 160 justification 3, 16–17, 27, 30, 32, 65, 67, 70–3, 85, 92–104, 107–9, 143, 159, 167, 179, 180, 198, 202–3, 211 knowledge 3–4 ,11, 14, 22–31, 37–8, 41–3, 45, 47–8, 63, 70–1, 94, 104, 106, 110, 149, 154–6, 160, 199, 202, 211, 220–1, 223, 236–7 language 23, 47, 126, 135–6, 143, 190, 198, 226–7 malapropism 122 meaning 3–5, 7, 10, 14, 39–40, 48, 57, 77–8, 120, 123, 140–2, 169, 199 speaker 141, 194–5, 197–9, 200–1, 205, 211–13 literal 141–2 pragmatic 140–2 meta-induction 32, 229–30 moral discourse 228 norms 56, 82–3, 84, 85, 88, 146–7, 154–6, 191–4, 202–3, 205, 207, 210–12, 221 226 omnipotence 44 omniscience 92, 102, 138, 157 open question 142, 220, 223–4
Subject Index pain 190, 222, 228 paradigm 84, 91, 104, 146, 169, 171, 226–7 paradox 222, 227 history of 37 Jourdain’s Visting Card 37 liar 3, 38, 57, 227–8 lottery 160, 221 Moore’s 3, 37–8, 43–4, 90–1, 100, 189–91, 203, 219 preface 54, 157–61, 193–4, 221–2, 227 Russell’s 37 perception 45, 83–6, 93, 95–6, 98, 100–1, 206 performatives 30, 80–2, 85, 135 phlogiston 219 plastic surgery example 133 pragmatics 10–11, 27, 32, 39, 90, 95, 141 precisification 227–9 predicate 117–18, 125–6 pretense 232–241 commissive 232, 235, 237, 241 omissive 232, 237, 241 privileged access 117, 137, 139, 143–4 probability 152, 159, 161, 218–19, 221, 224–5, 229 higher-order 224 subjective 160, 192, 218, 220–1, 225 pronouns anaphoric 118, 120–1 indirect reflexive 118, 136–7, 144 of laziness 121 propositional attitudes 79, 117, 120, 146, 149, 153, 166–8, 170, 172, 174, 176–7, 184, 187, 208 quantification 118 quasi-performative 85, 135 rationality 9, 15, 23, 28, 32, 33, 53, 55, 56, 59–61, 65–6, 72, 95, 97, 101–4, 107, 112, 117, 133, 141, 143, 166–7, 170, 172, 180 degrees of 22, 65, 107 doxastic 11 53–6, 66–7, 70–1 ideal 21–2, 105–7 practical 10, 12, 23, 24, 30,191–2, 194, 202 theoretical 9–10, 12, 13, 24, 29, 30, 191–4, 202, 207, 210 relativism 32, 37, 225–7, 237, 241 scope 57, 141, 146, 148, 179, 182, 186–7, 205, 208–10, 234 self-awareness 79
247
self-knowledge 93, 97, 102, 134, 137, 138 self-refutation 41, 53–5 sentence identity criteria of 123 Moorean 11, 13, 15, 20, 23–4, 27–9, 47, 105, 134, 136–8, 140–3, 205, 217–22, 226–7 string 119, 122–4 system 123, 138 token 125 type 123, 125–6, 219, 222 simulation theory 238 sincerity 80–1, 112, 141, 143, 195, 202, 206, 208 singular term referential 117, 127 transparent 124–5 skepticism 224 social pragmatism 27 speaker-meaning 194–202, 205, 211–13 speech acts 24, 77, 82, 88, 135–7, 146, 156, 201–3, 208, 210–13 spoonerism 122–3 Stigler’s Law 37–8 sub-valuationism 227 super-valuationism 227–8 supposition 7, 13–14, 31, 48, 83, 100, 109, 207–10, 232 theoretical perspectives 133 thick concepts 136 transparency 30–1, 143, 146–9, 151, 153, 158, 160 truth 9, 10, 13, 22, 23, 27–8, 37, 39, 40, 42, 44, 47, 55, 56, 60–3, 70–2, 79, 81–2, 85–6, 88, 92, 101, 106, 110–112, 121, 135–6, 139, 143, 149, 153–5, 158, 161, 174, 177–8, 180, 205–7, 219, 226, 230 gaps 32, 228 gluts 32, 227 higher-order 225 possible 32, 90, 102–4, 107 truth-tracking epistemologies 9, 70–2 Unger-Williamson thesis 146, 154 utterance Moorean 77–8, 203 token 122, 125–6, 128 type 122–24, 126–8, 130, 133 vagueness 227, 228 validity 9, 17, 61–2 vital spirits 219 visual field 44–6