Published online by Cambridge University Press: 23 January 2009
The debate about concepts has always been shaped by a contrast between subjectivism, which treats them as phenomena in the mind or head of individuals, and objectivism, which insists that they exist independently of individual minds. The most prominent contemporary version of subjectivism is Fodor's RTM. The Fregean charge against subjectivism is that it cannot do justice to the fact that different individuals can share the same concepts. Proponents of RTM have accepted shareability as a ‘non-negotiable constraint’. At the same time they insist that by distinguishing between sign-types and – tokens the Fregean objection cannot just be circumvented but revealed to be fallacious. My paper rehabilitates the Fregean argument against subjectivism. The RTM response rests either on an equivocation of ‘concept’—between types which satisfy the non-negotiable constraint and tokens which are mental particulars in line with RTM doctrine—or on the untenable idea that one and the same entity can be both a shareable type and hence abstract and a concrete particular in the head. Furthermore, subjectivism cannot be rescued by adopting unorthodox metaphysical theories about the type/token and universal/particular contrasts. The final section argues that concepts are not representations or signs, but something represented by signs. Even if RTM is right to explain conceptual thinking by reference to the occurrence of mental representations, concepts themselves cannot be identical with such representations.
1 A.J.P. Kenny, The Metaphysics of Mind (Oxford: Oxford University Press), 136; G. Rey, ‘Concepts’, in Craig, E. (ed.), The Routledge Encyclopedia of Philosophy (London: Routledge, 1998) Vol. 2, 505Google Scholar; Laurence, S. and Margolis, E., ‘The Ontology of Concepts: Are Concepts Abstract Objects or Mental Representations?’, Noûs 41 (2007), 561Google Scholar.
2 Quoted in Crane, T., ‘Something Else, Surely’, Times Literary Supplement (7.05.2004), 4Google Scholar. See: J. Fodor, Hume Variations (Oxford: Oxford University Press), 9 and ‘Having Concepts: A Brief Refutation of the Twentieth Century’, Mind and Language 19 (2004), 29–47. In this last article, he uses the less misleading label ‘concept pragmatism’ to cover all ‘epistemic’ accounts that treat concept possession as a kind of knowledge, including those which invoke knowing that rather than knowing how. But even in that article he does nothing to dispel the impression that RTM and pragmatism exclude one another. This is misleading, since one can be a representationalist while subscribing to conceptual role semantics.
3 See Volume 19.1 of Mind and Language (2004). Responding to Fodor, ‘Having Concepts …’, op. cit. note 2, Weiskopf and Bechtel reject his allegation that pragmatism cannot avoid circularity in its account of concept possession, Prinz and Clark defend the pragmatist link between conceptual thought and action, Rey deplores his wholesale rejection of the ‘Twentieth Century’ on the grounds that it leads to a first-person meaning mysticism, and Peacocke insists that epistemic accounts of concepts are not committed to unacceptable consequences like making thought interpretation-dependent. Fodor gets the last word in ‘Reply to Commentators’, Mind and Language 19 (2004), 99–112.
4 Fodor, Hume Variations, op. cit. note 2, 8; see 7–10.
5 Crane, op. cit. note 2.
6 Fodor, Hume Variations, op. cit. note 2, 8n, 141, 10.
7 Fodor, Concepts: Where Cognitive Science Went Wrong (Oxford: Oxford University Press), 3, 7–8, 22; Fodor, Hume Variations, op. cit. note 2, 13 + n; Crane, op. cit. note 2.
8 E.g, Rey, op. cit. note 1.
9 On this point I agree with Sutton (‘Are Concepts Mental Representations or Abstracta?’, Philosophy and Phenomenological Research 68 (2004), 89–93) and Margolis/Laurence (op. cit. note 1, 589 n. 10. But whereas they opt for subjectivism (alias RTM or mentalism), I defend Frege's objectivist argument concerning shareability. Unlike Fodor, Laurence and Margolis, however, Sutton does not reject the Fregean argument as fallacious, and his position is compatible with it. See fn. 27.
10 Op. cit. note 7, 22.
11 Frege, G., Grundgesetze der Arithmetik, Vol. 1 (Jena: Phole, 1893), XVIIIGoogle Scholar; see Grundlagen der Arithmetik; English translation The Foundations of Arithmetic (Oxford: Blackwell, 1953), X; ‘Der Gedanke’, in Beaney, M. (ed.), The Frege Reader (Oxford: Blackwell, 1997), original pagination 66–7Google Scholar.
12 Frege, Grundgesetze…, op. cit. note 11, XVIII–XIX.
13 ‘Letter to Husserl 24.5.1891’, in Beaney, op. cit. note 11, 149.
14 G. Frege, ‘Über Sinn und Bedeutung’, in Beaney, op. cit. note 11, original pagination 29–30.
15 Cp. Laurence and Margolis, op. cit. note 1, 567.
16 Posthumous Writings (Oxford: Blackwell), 253; ‘Die Verneinung’, in Beaney, op. cit. note 11, original pagination 151.
17 This term is preferable to Fodor's ‘publicity’: my nose is public in the sense of being accessible to more than one observer; but I am relieved to state that I do not share it with anyone.
18 Concepts, op. cit. note 7, 28.
19 C.S. Peirce, The Collected Papers of Charles Sanders Peirce, Vol. II (Cambridge, Mass.: MIT Press).
20 In both cases we have a repeatable entity and a non-repeatable entity. But the relation between believing that p and Anne's believing that p at time t is that between a (universal) property and a particularised instance of that property (a trope), whereas the relation between a type-word and a particular occurrence of it appears to be closer to that between a universal property and the concrete object that instantiates it, which in our example would be Anne herself.
21 The Language of Thought (New York: Crowell).
22 Hume Variations, op. cit. note 2, 13n; see Concepts… op. cit. note 7, 3n.
23 Concepts, op. cit. note 7, 28.
24 In a passage quoted below Fodor seems to adopt the orthodox distinction between particulars and universals. But in the last but one quotation and in his explanations of what it is for concepts to be mental particulars (quoted in my introduction) he contrasts particulars with abstracta, insisting that they are in the head and stand in causal relations. This inaccuracy does not matter for our purposes. For all universals are abstract (barring unorthodox views like that of Armstrong discussed in section 3). In any event, the contrast at the centre of the controversy between RTM and Frege is that between a particular confined to individual subjects and something shareable.
25 Pace Sutton, op. cit. note 9; Laurence and Margolis, op. cit. note 1, 567–9.
26 Something akin to this fall-back position is adopted by Davis, (Nondescriptive Meaning and Reference (Oxford: Oxford University Press, 2005))CrossRefGoogle Scholar and Sutton (op. cit. note 9, 92–4, 103). The latter explicitly concedes the Fregean claim that subjects are related to abstract types, while insisting that they can do so only through the occurrence of concrete tokens. Contrast what Margolis and Laurence call ‘the mixed view’, a position they eventually reject as insufficiently subjectivist (op. cit. note 1, 569, 579–81). It is based on rejecting rather than accommodating the Fregean argument, and hence identifies concepts with mental particulars, while conceding that the latter are typed in terms of Fregean senses.
27 Laurence and Margolis, op. cit. note 1, 567; my emphasis.
28 They extend this charge to a Fregean argument by Peacocke (‘Rationale and Maxims in the Study of Concepts’, Noûs 39 (2005), 167–78) according to which the subjectivist equation of concepts with mental representations fails since there are concepts but not mental representations that will never be acquired by a subject. The response I give on Frege's behalf can be adapted to defend Peacocke: if concepts are to exist independently of subjects they must be types rather than tokens, and hence cannot be mental particulars. However, while all parties to the debate rightly assume that concepts can be shared, it is more controversial to maintain that concepts can exist even if they are never employed by a subject of conceptual thought. For what it is worth, I reject this claim while accepting an analogous claim about properties.
29 S. Laurence and E. Margolis, ‘Concepts and Cognitive Science’, in Concepts: Core Readings (Cambridge, Mass.: Bradford Books/MIT Press), 7.
30 Fodor, Concepts…op. cit. note 7, 28; my emphasis.
31 Concepts…op. cit. note 7, 20–1. This passage disregards the fact that for Frege a concept is not the sense or mode of presentation expressed by a predicate. As mentioned above, this does not matter in the present context. The underlying question is whether whatever distinguishes co-extensional predicates and determines the contents of propositional attitudes is ‘ipso facto in the head’ as Fodor contends (1998: 15). Prima facie more threatening to my reading of Fodor's passage is the fact that it identifies a concept not with an MOP simpliciter, but with ‘an MOP together with a content’. Nevertheless the passage directly concerns the shareability of concepts. Contents are ipso facto shareable by Fodor's lights; his task is therefore to show against Frege that MOPs are as well, in spite of being mental rather than abstract.
32 Frege already complained that psychologistic logicians exploit such fudge-operators in order to defuse the absurd consequences of their claims: ‘This “as such” is an excellent invention of authors who want to say neither yeah nor nay. But I do not brook such oscillation between the two, …’ (Grundgesetze der Arithmetik, op. cit. note 11, XXIII–XXIV).
33 Fodor cannot turn the tables on the Fregean by accusing her of the following equivocation: First Premise: Concepts (in the sense of types) are shareable; Second Premise: Mental particulars are not shareable; Conclusion: Concepts (in the sense of tokens) are not mental particulars. For neither the First Premise nor the Conclusion feature in her argument. The Fregean is not advancing any claims about concepts qua tokens or types. Rather, she accepts that concepts—the things with which we credit people in ascriptions of propositional attitudes and in intentional explanations–must be shareable–in line with Fodor's own professions. From this in conjunction with the Second Premise she validly infers that concepts (i.e. the aforementioned things which can be assumed to be shareable) are not particulars. At most one could reciprocate the charge of equivocation if the non-negotiable constraint on concepts (the things with which we credit people, etc.) included that they are particulars. As mentioned above, this is what Fodor suggests in chapter 2 of Concepts…op. cit. note 7. But he does not use this constraint as a premise in criticizing objectivism; advisedly so, since such a line of reasoning would be blatantly question-begging.
34 Armstrong, D.M., Universals: An Opinionated Introduction (Boulder, Colorado: Westview Press, 1989)Google Scholar.
36 Even if sound, these remarks do not constitute a definitive refutation of Armstrong's theory of universals. For Armstrong himself wavers on the question of whether universals can really be located in space and time. On occasion he suggests instead that they constitute space and time by being constituents of states of affairs, with space-time being the conjunction of all states of affairs (Universals…op. cit. note 36, 98–9). I shall not assess this proposal, since it does not lend succour to the idea that concepts might be types and nonetheless concrete occurrences in the heads of individuals. My aim was merely to show that the rejection of the Fregean argument cannot be based on a literal interpretation of Armstrong's text.
37 Hume Variations, op. cit. note 2, 136–142.
38 Furthermore, even if the opponents of universals are right, their conception of particularized properties may be cold comfort for RTM. Consider first the reaction of a realist about universals to our particular example: if the two believings in the respective heads of Anne and Sarah resemble each other in their role or function (rather than their owner or their spatio-temporal location), then they must still have the same role or function, that is, fall under a universal. Their opponents resist the introduction of such a universal by insisting that similarity is a primitive concept. But even if similarity is primitive, i.e. a concept which cannot be further explained, it prima facie remains similarity in a certain respect. As a result opponents of universals like Heil (From an Ontological Point of View (Oxford: Clarendon Press, 2003), ch. 14) are driven to a further contention: while objects are similar or dissimilar in certain respects, this does not hold for (particularized) properties. Whatever the merits of this further move, however, it takes us beyond the board on which the debate of this paper is being played out. For the mental particulars of RTM are precisely objects rather than properties in Heil's sense.
39 Concepts…op. cit. note 7, 6.
40 Concepts… op. cit. note 7, 25.
43 Concepts… op. cit. note 7, 2.
44 Hume Variations, op. cit. note 2, 13.
45 Consistency cannot be restored by pleading that the first claim refers to concept-tokens and the second to concept-types. For sign-types are not the contents or meanings of sign-tokens. The token ‘dog’, for instance, does not mean or express the sign type ‘dog’, it instantiates that type. What could mean the sign type are signs like ‘“dog”’ or ‘the word “dog”’.
46 Like many contemporaries, Fodor equates meaning and content, e.g. when he describes believing that p as standing in a computational relation to a mental representation that ‘means that’ p. In my view, this equation is untenable. A proposition, propositional content or thought is something that is or could be said or believed—a sayable or thinkable. It is something expressed or conveyed by the use of a sentence, not the meaning of a sentence. Unlike the meaning of a sentence, what is said (believed, etc.) can be true or false, implausible or exaggerated (Cartwright, R., ‘Propositions’, in his Philosophical Essays (Cambridge, Mass.: MIT Press, 1987), 33–53Google Scholar; White, A., Truth (London: Macmillan, 1970), 14Google Scholar; Künne, W., Conceptions of Truth (Oxford: Oxford University Press, 2003), 368–72CrossRefGoogle Scholar). Furthermore, far from being identical with sentence-meanings, what is said on a particular occasion depends on sentence-meaning and context of utterance. In the present context, however, the difference can be ignored. The important point is that concepts are situated not at the level of signs but at the level of what signs mean or have as content when they are used.
47 Concepts…op. cit. note 7, 29; see 7.
48 A Study of Concepts (Cambridge, Mass.: MIT Press, 1992), 3.
50 ‘Concepts and Cognitive Science’, op. cit. note 30, 76.
51 To be sure, in aesthetics we encounter an extension of the semiotic distinction which might accommodate Laurence and Margolis. It treats a literary work such as Anna Karenina as a type, of which, e.g. Solzhenitsyn, Bloom and I possess different tokens, notwithstanding the fact that these tokens are spelled differently. But this terminological extension of the notion of a type in no way alters the crucial fact: what is common to the envisaged representations of cities lies at the semantic rather than the typographic or syntactic level.
52 Contrary to Glock, H-J., ‘Neural Representationalism’, Facta Philosophica 5 (2003), 105–29Google Scholar.
53 Concepts…op. cit. note 7, 28.
54 I deliberately leave open the precise nature of this semantic relation of representing. My preference is to adopt Künne's (Conceptions of Truth, op. cit. note 48, ch. 1) distinction between application, signification and expression: The concept-word ‘dog’ applies to dogs, signifies the property of being a dog, and expresses the concept of being a dog. Of course there are alternatives to this terminology, and there may be superior accounts of concept-words. But this does not affect my point against RTM, namely that concepts are not themselves symbolic representations, but something which stands in a semantic relation to such representations.
55 At the same time, the Fregean position preserves a connection between concepts and representation. Concepts are neither symbols nor subjective representations of a different (e.g. iconic) kind, but they are ‘modes of presentation’, i.e. ways in which different subjects can think of something. This opens up the possibility of treating concepts both as things represented—namely things expressed by concept-words—and as representations—namely ways of presenting properties (see Künne, W., ‘Properties in Abundance’, in Strawson, P. F. and Chakrabarti, A. (eds.), Universals, Concepts and Qualities (Aldershot: Ashgate, 2006), 263-6Google Scholar. My criticism of RTM does not rule out this option. Alas, it is difficult to spell out, especially if one seeks to avoid both the Scylla of subjectivism and the Charybdis of Platonist myths about a ‘third realm’.
56 For comments I should like to thank Peter Hacker, John Hyman, Wolfgang Künne, Christian Nimtz, Oliver Petersen and Mark Textor, as well as audiences at Reading, Edinburgh, Leeds, Berlin, Bern, Geneva, Essen and Konstanz. I am particularly grateful to Javier Kalhat for suggestions and editorial help.