Linguistic Explanation and 'Psychological Reality'

Croatian Journal of Philosophy Vol. IX, No. 25, 2009 Linguistic Explanation and ‘Psychological Reality’1 PETER SLEZAK Program in Cognitive Science University of New South Wales Chomsky’s generative approach to linguistics has been debated for decades without consensus. Questions include the status of linguistics as psychology, the psychological reality of grammars, the character of tacit knowledge and the role of intuitions. I focus attention on Michael Devitt’s critique of Chomskyan linguistics along the lines of earlier critiques by Quine, Searle and others. Devitt ascribes an intentional conception of grammatical knowledge that Chomsky repudiates and fails to appreciate the status of Chomsky’s computational formalisms found elsewhere in cognitive science. I argue that Devitt’s alternative to the psychological view – a “linguistic reality” of physical objects as the proper subject matter of linguistics – neglects the problems of tokens as opposed to types and misses the force of Chomsky’s arguments against Behaviourism. Furthermore, I suggest that Devitt’s case against intuitions misunderstands their standard, central role throughout perceptual psychology. Of more general interest, I argue that Devitt’s position exemplifies compelling errors concerning mental representation seen throughout cognitive science and philosophy of mind since the 17th Century. Key words: 1. “Original Sin” and “Rather idle controversy” When a debate persists over decades or even centuries without shift in the opposing positions, there is good reason to think that the ad1 I am indebted to Michael Devitt for his gracious responses to my criticisms in exchanges and in comments on an early version of this paper which have helped to improve it both in form and content and to correct mistakes and misattributions. Thanks also to John Collins and Georges Rey for most helpful remarks. For discussion I am grateful also to Deborah Aarons, Mengistu Amberber, Eran Asoulin, Stephen Crain, Clinton Fernandes, Iain Giblin, Peter Menzies, Nick Riemer, Benjamin Schulz, Michael Slezak, Phil Staines, and participants in the UNSW cognitive science discussion group. 4 P. Slezak, Linguistic Explanation and ‘Psychological Reality’ versaries are talking past one another. The debate about knowledge of language is prima facie a case in point. Moreover, I will suggest that the misunderstandings involved are not unique to linguistics but pervasive throughout cognitive science and philosophy of mind since the seventeenth century (See Slezak [2002a,b], [2006]). For Fodor [1968] the generative approach provided a model for psychological explanation of the sort also articulated also in Pylyshyn’s seminal work – namely, the computational view of cognition that “presupposes that we take seriously such distinctions as those between competence and performance” ([1984], 223). A decade earlier, writing on ‘Competence and Psychological Reality’, Pylyshyn ([1972], [1973]) explained “One reason why the notion of competence is particularly important is that it is the first clear instance of the influence of … mathematical imagination on the study of cognition” ([1972], 548) referring to studies by Turing, Gödel, Church and others in the foundations of mathematics and the theory of computation. Over thirty years later, it is clear that these ideas have not become less controversial. Michael Devitt [2006a,b] has mounted an extended critique of the generative enterprise and compares the situation to quantum physics where there are not only successful explanatory theories but also controversy about foundations. Devitt says in linguistics, by contrast, “There is not a similar controversy about how to “interpret” these theories but I think that there should be.” However, on the contrary, there has been just this controversy from the earliest days of generative linguistics, and we will see that Devitt is simply rehearsing the most persistent objection to the “psychological reality” of grammars as internal representations. Quine [1972] and Stich [1972] had made identical objections, repeated by Searle [1980b], Ringen [1975] and Botha [1980].2 By 1978, one philosopher remarked that “More has been written, much of it exasperatingly shallow, about the confusions surrounding the concept of competence and knowledge-as-competence than almost any other topic in recent philosophy” (Nelson [1978], 339). In light of this history, any engagement with the issues today should be accompanied by some diagnosis of the peculiar recalcitrance of the debate being rehearsed. I will suggest that the interest of Devitt’s work goes beyond the issues of linguistics with which it is directly concerned to deep and pervasive problems throughout cognitive science. Devitt’s work may be seen as a case study in what Rorty ([1979], 60) has called “the original sin of epistemology” – to model knowing on seeing ([1979], 146). In a telling metaphor, Devitt suggests “If we could look into the brain and simply “see” if there were representations of this and that, as we can look in a book and see if there are representations … then that would of course settle the matter” ([2006b], 51). Devitt’s verdict is a calamitous judgment on Chomsky’s conception of his linguistics. He writes: 2 See Slezak [1981] P. Slezak, Linguistic Explanation and ‘Psychological Reality’ 5 I urge that linguistics is not part of psychology; that the thesis that linguistic rules are represented in the mind is implausible and unsupported; that speakers are largely ignorant of their language; that speakers’ linguistic intuitions do not reflect information supplied by the language faculty … that there is little or nothing to the language faculty … (Devitt [2006b], vi) In short, on Devitt’s view just about everything in Chomsky’s conception of the status and character of the generative enterprise is wrong. Devitt’s critique rests on two pillars: his negative critique of Chomsky’s psychological realism regarding grammars, and his positive case for an alternative conception of a “linguistic reality” of physical tokens that grammars are about. Devitt’s positive alternative program, the second pillar of his book (sections 12, 13 below), entails nothing less than undoing the mentalism of the “cognitive revolution” in a return to the nominalistic aspects of Skinnerian, Bloomfieldian behaviourism. Devitt ([2006b], 27) protests that he is not a behaviourist, 3 but embracing “outputs/products” is precisely a concern with the data of performance of which Chomsky had remarked it is not an arguable matter. He said “It is simply an expression of lack of interest in theory and explanation” ([1965], 193). In this regard, we may recall Chomsky’s remarks on “this rather idle controversy” being revived by Devitt. Remarkably, it remains apt to characterise Devitt’s work in Chomsky’s words forty years ago as “a paradigm example of a futile tendency in modern speculation about language and mind” [1967]. 2. The “Natural” Interpretation of Chomsky Devitt attributes an implausible intentional, propositional attitude conception of grammar to Chomsky as the most “natural interpretation” despite Chomsky’s explicit rejection of it.4 3 Devitt (correspondence) objects to this imputation but his intentions are irrelevant since I am drawing attention to the unnoticed implications of his doctrines. Chomsky [1959] showed that Skinner was, malgré lui, up to his ears in psychological assumptions, a closet mentalist, while professing a strict behaviourism. In the same way, I am suggesting that Devitt is a closet behaviourist while professing an orthodox mentalism. 4 Devitt claims “strong evidence” ([2006b], 10), “a great deal of evidence” ([2006c], 572) and even “massive evidence” ([2006b], 7) for the attribution to Chomsky of a “strong commitment to RT” ([2006b], 71), while at the same time retreating behind a formal disclaimer on the grounds that the question concerning Chomsky’s view “is surprisingly hard to answer” ([2006b], 62). In any case, he says, “interpreting Chomsky is not my major concern” ([2006b], 7). On the exegetical question of what Chomsky means by the central term “represent” Devitt professes agnosticism and refers the reader to the exchange between Rey [2003a,b] and Chomsky [2003]. However, Rey not only construes ‘representation’ in the same way that Devitt does, but defends its attribution to Chomsky on the basis of the most uncharitable personal attributions. Devitt gives no indication that he dissents from either the substance or tone of Rey’s ad hominem remarks, including the charge that Chomsky holds the intentional view despite his denials. Devitt [2007] has correctly pointed out an error in my earlier portrayal of his attribution of RT in which I cited comments bearing on a different issue. However, my portrayal does not depend on this erroneous evidence 6 P. Slezak, Linguistic Explanation and ‘Psychological Reality’ The Representational Thesis (RT): A speaker of a language stands in an unconscious or tacit propositional attitude to the rules or principles of the language, which are represented in her language faculty. ([2006a], 482, [2006b], 4) Devitt ([2006b], 5) illustrates the doctrine with the very example of a picture that Chomsky ([2003], 276) uses to distinguish it from his own and RT is attributed to Chomsky despite the familiar merits of his computational conception found elsewhere throughout cognitive science. Devitt characterises Chomsky’s view of linguistic knowledge as “propositional knowledge of syntactic rules,” ([2003], 108) suggesting that speakers have access to the linguists’ theories ([2003], 4). 5 Devitt imputes the view that linguistics is the study of “the system of rules that is the object of the speaker’s knowledge” ([2003], 109). However, for Chomsky, the rules are not the intentional object of the speaker’s knowledge, but rather constitute this knowledge. Chomsky’s frequent comparisons with insects and bird-song could hardly make sense on any other interpretation. Devitt ([2003], 109) defends his interpretation of Chomsky because it “takes his talk of ‘knowing that’, ‘propositional attitudes’, and ‘representation’ at face value.” He asserts “The natural interpretation attributes RT to Chomsky” ([2006b], 7). It is telling that Rey [2003b] uses the same curious expression “the natural interpretation” as if we are dealing with hermeneutics of the Dead Sea Scrolls whose author’s intentions are obscure or unavailable. By any reasonable measure, the natural interpretation is clearly the one that Chomsky has repeatedly articulated and insisted upon in specific response to the very construals offered by Devitt and Rey. Stone and Davies ([2002], 278) writing of ‘Chomsky Amongst the Philosophers,’ reflect on the fact that “Philosophers object to linguistic theories, not on the grounds that these theories fail to account adequately for the empirical evidence, but because they fail in other ‘philosophical’ ways.” Devitt discovers a supposed anomaly in Chomsky’s approach: “What is puzzling about this is that a strong commitment to RT seems inappropriate in the absence of a well-supported theory of language use that gives RT a central role” ([2006b], 71; emphasis added.). That is, Devitt foists a view onto Chomsky that he doesn’t hold and is then mystified by his failure to take it seriously. This exegetical strategy leads Devitt to discover the same mystery repeatedly among other theorists such as Fodor, Bever and Garrett [1974] and Berwick and Weinberg [1984]. Devitt says again: “This raises the old puzzling question: why be so convinced about RT given this ignorance about its place in a theory of processing?” ([2006b], 79]; emphasis added). Of course, the puzzle disappears if the theorists are taken at their word and not assumed to hold RT or positing processing mechanisms. but on remarks of the kind just noted that are difficult to reconcile with professions of agnosticism. 5 B.C. Smith [2006] makes the same point. P. Slezak, Linguistic Explanation and ‘Psychological Reality’ 7 3. Theory and Object Devitt attributes implausibly naïve errors to Chomsky such as “a certain use/mention sloppiness” and a neglect of the elementary distinction between a theory and its object. Devitt says “Chomsky seems uninterested in the difference” ([2006b], 69). However, Devitt has evidently missed Chomsky’s ([1965], 25) explicit warning about the systematic ambiguity of the term “grammar” used to refer both to the speaker’s internally represented “theory” and also to the linguists’ account of it.6 On its own, this oversight fatally compromises Devitt’s critique of the generative enterprise since his ascription of the intentionalist thesis RT to Chomsky is just this mistaken attribution. Thus, it should be needless to say that Chomsky does not suggest that the formalisms of a grammar themselves are in the head.7 Attributing the conflation of theory and object follows from failure to appreciate the literal, realistic construal of abstractly specified psychological rules and representations. 4. Savoir and Connaître In his Dioptrics Descartes proposes that the mind determines the distance of an object by means of an implicit triangulation or parallax calculation based on the separation of the eyes and their orientation. Descartes says “this happens by an action of thought which, although it is only a simple act of imagination, nevertheless implicitly contains a reasoning quite similar to that used by surveyors, when, by means of two different stations, they measure inaccessible places” ([1637/1965], 106). This is, of course, just Chomsky’s conception of a formal competence theory that captures our tacit knowledge – a mathematical, computational model describing what we know unconsciously and underlying our intuition or “simple act of imagination” (see Slezak [2006]). Wolf-Devine ([1992], [2000a,b]) notes Descartes’ use of the verbs savoir and connaître, suggesting that Descartes is guilty of “hopeless over-intellectualization of perception” ([2000a], 513). In the same terms Devitt is critical of Chomsky’s “highly intellectualist” account ([2006b], 60), a constant refrain in the chorus of criticism that Devitt joins concerning the “psychologically reality” of formal rules and representations. 5. Fitting and Guiding. Chomsky ([2000], 94) suggests that in its modern guise we can trace the argument about “psychological reality” of grammars back to Quine’s distinction between rules that are “guiding” and those merely “fitting” or perhaps “respected,” in the way that a planet obeys Kepler’s Laws. Above all, on such views, we must not attribute “psychological reality” to such rules. Chomsky ([1975b], 190, 198) wrote of the “singularly misleading analogy” that is frequently made between respecting math6 The same clarification was made in Chomsky ([1972], 116) and ([1975a], 37). See Devitt ([2006b], 4). Smith [2006] characterizes Devitt’s position as “wildly amiss”. 7 8 P. Slezak, Linguistic Explanation and ‘Psychological Reality’ ematical laws of physics and the rules of a grammar. However, Devitt’s main criticism of Chomsky’s grammars is just an elaboration of this Quinean distinction between “fitting” and “guiding” rules. Devitt’s version of Quine’s qualms is expressed as the following principle: Distinguish processing rules that govern by being represented and applied from ones that are simply embodied without being represented. ([2006b], 45) Devitt says that his thesis that linguistics is not part of psychology rests in part on this principle that he asserts is “not controversial” ([2006b], 46). However, apart from the fact that Chomsky has disputed it for forty years, the principle is also central to intense disputes elsewhere in cognitive science (see sections 8,9 below). 6. Simply embodied: The Birds and the Bees Devitt finds a deep paradox in the fact that Chomsky allegedly “has no worked out opinion about, or even much interest in, how that grammar in the head plays a role in language use” ([2007b], 71). However, on the contrary, Chomsky has suggested plausibly that his abstract, idealized approach to ‘competence’ is the best way to discover underlying processing correlates of grammars ([1980b], 197). Far from lacking interest in the question, Chomsky is simply responding to the obvious fact acknowledged in Devitt’s own words that “we don’t even know enough about what to look for” or, in Fodor’s ([1998], 145) words quoted approvingly by Devitt ([2006b], 52), “there isn’t one, not one, instance where it’s known what pattern of neural connectivity realizes a certain cognitive content.” Thus, Chomsky expresses exactly Devitt’s own sentiment, saying “we might go on to suggest actual mechanisms [underlying abstract rules], but we know that it would be pointless to do so in the present stage of our ignorance concerning the functioning of the brain” ([1980b], 206,7). Symptomatic is Devitt’s egregious misuse of Chomsky’s key technical term ‘competence’ in something like the colloquial sense of “a competence to produce” or “a competence to process” ([2006b], 17). Thus, Devitt identifies a bee’s “internal state of competence to produce the dance” with “the processing rules within a bee that enables it to perform this remarkable feat” – that is, “how the bee performs this dance” ([2006c], 575). This is precisely the opposite of Chomsky’s use of the crucial term and exactly what he means by its contrasting “performance.”8 On one alternative to the “most natural” reading of Chomsky’s words, he is taken to hold that language rules are merely embodied “without being represented” ([2006b], 7, [2003], 109). However, Rey, like Devitt, finds it implausible that an ant might represent “the system of vector 8 See also Smith ([2006], 449). Here Devitt [2007] seems to be correct in his response to Smith, since the issue can’t turn on whether von Frisch’s account is empirically correct or not but only on the distinctions it is used to illustrate. P. Slezak, Linguistic Explanation and ‘Psychological Reality’ 9 algebra itself” ([2003b], 153).9 Instead, Rey suggests “That system is, at best, merely implemented somehow in the ant’s nervous system.” However, given Devitt’s stipulation of how “representation” is to be used, this is simply an arbitrary terminological matter of no theoretical interest. Indeed, as we will see, from another point of view Devitt is merely re-stating Chomsky’s own competence/performance distinction. Devitt says “We need psychological evidence to show which grammar’s rules are in fact playing the role in linguistic processing, evidence we do not have.” Indeed, Devitt adds, “We need evidence that the syntactic rules of any grammar are processing rules. These rules may simply be the wrong sort of rules to be processing rules” ([2006b], 37). There could be no more explicit indication of the source of the persistent worry about the formalisms of a competence theory. Searle [1980b] made the stereotypical complaint in terms now repeated almost verbatim by Devitt, Searle explained: Additional evidence is required to show that they are rules that the agent is actually following, and not mere hypotheses or generalizations that correctly describe his behavior. … there must be some independent reason for supposing that the rules are functioning causally. ([1980b], 37) Devitt’s scepticism about rules is a refusal to accept the abstract idealizations of grammars when their underlying processing realization is unknown. However, if the term “psychological reality” is granted as referring by stipulation to processing mechanisms, then the debate collapses, Devitt’s position being no more than a restatement of the competence-performance distinction – a point also noted by Laurence ([2003], 87). This charge is easy to substantiate, as we can see from Devitt’s statements: It is not enough to know that there is something-we-know-not-what within a speaker that respects the rules of her language … We would like to go beyond these minimal claims to discover the ways in which the competence of the speaker … respect these rules. ([2006b], 38) Chomsky writes: …we are keeping to abstract conditions that unknown mechanisms must meet. We might go on to suggest actual mechanisms, but we know that it would be pointless to do so in the present stage of our ignorance concerning the functioning of the brain. ([1980b], 197) Chomsky’s phrase “abstract conditions that unknown mechanisms must meet” is precisely Devitt’s “Respect Constraint” (see section 11 below). Devitt [2007] insists that this is insufficient for “the grammar alone … does not tell us what there is in the speaker that does the respecting” and “the grammar alone gives us no reason to suppose that the formalisms are descriptive about the mind.” Chomsky [1980a] has parodied exactly this question since, unless “descriptive” concerns per9 Rey ([2003b], 157,8) responds to Chomsky with sarcasm, accusations of inconsistency or insincerity and the evidence of his colleagues’ shared incomprehension (Rey [2003b], 160 fn 19). 10 P. Slezak, Linguistic Explanation and ‘Psychological Reality’ formance or processing, there can be no other answer besides repeating the evidence on which the grammar is proposed. Terminology aside, Chomsky’s competence-performance distinction might well be encapsulated in Devitt’s own supposed challenge: “A grammar may have nothing more to do with psychological reality than comes from meeting the Respect Constraint” ([2006b], 37). 7. “Cognize” It is clear that the “debate” has long ago degenerated into a ritual talking past one another. Chomsky and Katz [1974], 363) replied to Stich’s [1972] “projectile” argument saying: “At best, it is an open question whether more than an uninteresting issue of terminology is involved.” Regarding the terminological issue, Devitt says “The term ‘know’ is mostly used for the propositional attitude in question but, when the chips are down, Chomsky is prepared to settle for the technical term ‘cognize’” ([2006b], 4; emphasis added). We may note the irony of Devitt’s accusation of a certain “looseness of talk of ‘knowledge.’” Devitt says “I think that linguistics would do better to avoid the talk” of knowledge ([2006b], 5). However, it was precisely because of Devitt’s sort of misapprehension arising from the misleading connotations of the term “know” that Chomsky ([1986], 265) himself suggested that it might be replaced with the neologism “cognize.” Devitt evidently misses Chomsky’s very effort to dispel confusion, adopting it as his own, while seeing Chomsky’s proposal as a compromise or implicit concession to his critics. 8. “Epiphobia” Responding to persistent complaints about “psychological reality” Chomsky describes the term as “hopelessly misleading and pointless” and arising from a methodological dualism that makes invidious distinction between linguistics and other sciences. Seeking the deeper sources of this error, we may note Fodor’s [1990] diagnosis of “epiphobia” – as he dubs the neurotic fear of the causal inertness of the mental. Evidently harbouring this fear, Devitt says “we should only posit such representations [of rules] if we can find some serious causal work that they have to do” ([2006b], 52). Chomsky’s grounds for attributing causal efficacy to rules is just that they have a place in our best explanatory theories (Chomsky [1980a,b]). Devitt regards this argument as “fast” and “dirty” ([2006c], 574), but Fodor notes that the case against the reality of psychological posits would also require epiphenomenalism with regard to all non-physical properties: “If beliefs and desires as are well off ontologically as mountains, wings, spiral nebulas, trees, gears, levers, and the like, then surely they’re as well off as anyone could need them to be” ([1990], 141). Rules are psychologically real for the same reason, even if we have no idea about the underlying realization – “what there is in the speaker that does the respecting” (Devitt [2007]). P. Slezak, Linguistic Explanation and ‘Psychological Reality’ 11 9. “But it doesn’t really” Essentially the same problem concerning the reality of rules has arisen for connectionist models too (Bechtel & Abrahamsen [2002], 120) Typical of a vast literature,10 almost at random we can pick Smolensky’s remark that in the case of a certain connectionist model, “It’s as though the model had those laws written down inside it” but he adds the curious qualification “But it doesn’t really” ([1998], 20). That is, “The system is … not really satisfying the hard rules at all.”11 In the same vein, Clark ([1990], 292) writes that in connectionist systems “the processing can hardly be sensitive to structures which aren’t there.” In these remarks we see the familiar idea that, although a system might conform to (or “respect”) rules, they are not real. Pinker and Prince note the basis for such rule-skepticism: Instead of “explicit” representations, we have weighted connections and activation levels among which “one cannot easily point to rules, algorithms, expressions, and the like” ([1988], 76; emphasis added). Theorists evidently use a certain implicit criterion for literal attribution of structure – namely, whether it is apparent to inspection for the theorist.12 However, we see the crucial point made by D. Kirsh ([1990], 351) who observes “what humans are able to see is irrelevant.”13 He gives an apt diagnosis of just this problem arising from “the bewitching image of a word printed on a page” ([1990], 350) – Devitt’s “original sin.” We will see that this is a key insight into the illusion of explanatory adequacy derived from tacit dependence on our own interpretative abilities. 10. Functionalism Fodor ([1968], ix) noted that Chomsky’s grammars illustrate the functionalist conception of mind – the modern statement of what it means to do psychology in the information processing paradigm and to attribute internal representations. Chomsky’s writes: The mentalist … need make no assumptions about the possible physiological basis for the mental reality he studies. … One would guess … that it is the mentalistic studies that will ultimately be of greatest value for the investigation of neurophysiological mechanisms, since they alone are concerned with determining abstractly the properties that such mechanisms must exhibit and the functions they must perform. ([1965], 193) Failing to appreciate Chomsky’s ([1982], 10) idealizations, like Marr’s ([1982], 28), Devitt says “it is hard to see how it [a grammar] could be a theory at the computational level” ([2006b], 66). For example, Chom10 See Van Gelder [1990]; Pinker, S. and Mehler, J. (eds) [1988]; Horgan, T. & Tienson, J. (eds) [1991]; Ramsey, W., Stich S., & Rumelhart, D.E., (eds) [1991]. 11 See also Waskan & Bechtel [1997]. 12 See van Gelder ([1992], 180), Ramsey, Stich & Garon [1991]. 13 See also Cummins [1996] and McDermott [1981]. 12 P. Slezak, Linguistic Explanation and ‘Psychological Reality’ sky’s conception of associating a sound with a meaning is misunderstood as a causal, mechanical, processing matter, whereas for Chomsky it is a purely formal, descriptive one, as Pylyshyn explains ([1973], 44). Even where Devitt ([2006b], 68) acknowledges the purely mathematical sense of the notion of “generate” he fails to appreciate the precise force of this conception according to which one says that an axiom system generates its theorems.14 Above all, this formal sense of the term ‘generate’ is not “merely metaphorical” in any sense, as Devitt seems to think. 11. “R - E - S - P - E - C - T: Find out what it means to me” (Aretha Franklin) Devitt ([2006b], 57) elaborates five possible positions on language use, some having two versions, and three versions of the idea that rules might be internally represented. He also enunciates eight numbered precepts, four methodological points, various named theses and technical distinctions. However, once we adjust for Devitt’s choice of terminology, particularly his notable misuse of the key term ‘competence’, we have only a re-statement of Chomsky’s own views, though much less perspicuous than the original. Devitt says “I emphasize that ‘respecting’ as I am using it, is a technical term” to distinguish processing rules from “structure rules” that are merely conformed with or “respected.” Thus, “something counts as a chess move at all only if it has a place in the structure defined by the rules of chess” as distinct from “the rules governing the psychological process by which [a player] … produces chess moves.” Devitt irrelevantly but tellingly characterizes the latter as “interesting,” but his analysis simply restates Chomsky’s competence/performance distinction. Devitt’s “interesting” rules are those concerning the “heuristic” aspect of problem solving of the kind exemplified in the work of Newell and Simon [1972], but must be distinguished from the ‘epistemological’ or formal approach (see Pylyshyn [1973], 22). That is, the first pillar of Devitt’s account – his critique of Chomsky (where it goes beyond the misattribution of RT) – collapses into a verbal quibble. Devitt seems to appreciate the point in a footnote: “I do not take it [a grammar] to be real simply in virtue of its meeting the Respect Constraint. But this difference may be just verbal” ([2006b], 67). 12. Unimaginative Nominalist. Gurgling and Throat Clearing. Devitt’s central thesis that linguistics is not a branch of psychology rests on his argument that “there is something other than psychological reality for a grammar to be true of: it can be true of a linguistic reality” ([2006b], 17). This “linguistic reality” is constituted by the “outputs of competence” such as “physical sentence tokens” and the “properties of 14 See Collins ([2006], 495). P. Slezak, Linguistic Explanation and ‘Psychological Reality’ 13 symbols” or “certain sounds in the air, inscriptions on paper” ([2006b], v) Surprisingly, Devitt ([2006b], 26) acknowledges that his position corresponds with the Bloomfieldian ‘nominalism’ that was supplanted by ‘conceptualism’ on the basis of Chomsky’s powerful criticisms. However, Devitt says: “Yet, so far as I can see, these criticisms are not of the nominalism of the structuralists but rather of their taxonomic methodology” ([2006b], 26). Devitt thinks that a “linguistic reality” of physical tokens has an epistemic and explanatory priority over a psychological reality, but the situation is exactly the reverse. Devitt ([2006b], 88) professes acceptance of Chomsky’s critique of Behaviourism but he sees the critique as bearing only on “a crude empiricist dislike of things unseen; an unwillingness to posit theoretical entities that explain the observed phenomena.” However, Chomsky’s remarkable review exposed not just the fear of hypotheses about “things unseen”, but Skinner’s unwitting commitment to mentalist conceptions. This fatal flaw in the behaviourist apparatus is not the same as the mistake of assuming that behaviour is under stimulus control, as Devitt seems to think ([2006b], 87). Thus, Devitt’s very identification of “outputs of a competence,” employs what Chomsky describes as a device that is “as simple as it is empty” for it “simply disguises a complete retreat to mentalistic psychology” ([1959], 32). Devitt reveals that he is victim of Kirsh’s ([1990], 350) “bewitching image” or interpretative illusion, saying: “This work and talk [by linguists] seems to be concerned with the properties of items like the very words on this page” ([2006b], 31).15 As Smith has put it, “Without the cognitive wherewithal to represent sounds as symbols they would be heard as no more than inarticulate gurgling or throat clearing” ([2006], 438). The theorist cannot help himself to the linguistic properties of these physical tokens since, as Rey [2006a,b], Laurence ([2003], 89) and Bromberger ([1989],73) have argued, there is an important sense in which they don’t exist at all independently of minds. In short, Devitt’s nominalism reduces either to a futile interest in heterogeneous physical data or an unwitting inquiry into psychological competence after all. Bromberger notes that linguists “habitually conflate mention of tokens with mention of types” but “tokens are not what linguistics is primarily concerned with. Types are” ([1989], 59). With Halle he writes: Types turn out to be rather innocuous things to which only the most unimaginative nominalists should object. ([1992], 15) Bromberger concludes “linguistics is not just about types and tokens but is also inescapably about minds.” 13. They ain’t nothin’ until I call ‘em. Devitt asks: “What makes a physical object a pawn?” taking the anal15 See also Hadley [1995]. 14 P. Slezak, Linguistic Explanation and ‘Psychological Reality’ ogy with chess to show that linguistic investigation is not psychological. However, the physical pieces and their moves do not have explanatory priority. Clearly, anything might serve as a token of a knight or bishop. Pieces of wood and their movements don’t even count as tokens of symbols or “outputs of competence” except insofar as they are classified in this way by prior conceptualisation according to mentally represented, intentional rules – a point long familiar in the philosophy of social sciences.16 Of course, we might investigate the Queen’s Gambit, Sicilian Defence and Nimzovich variation without concern for their status as human knowledge. The central issue is not the independence of the abstract rules governing these objects, but their very status. In the same way, Chomsky claims that language has no existence apart from its mental representation ([1972], 169). Without legislating what enquiry to pursue, he says that study of the outputs of competence like horseshoes and language tokens has proven to be sterile. 14. Possible Idealized Horseshoes The ‘cognitive revolution’ and the study of generative grammar involved the very shift that Devitt unintentionally seeks to reverse – the shift described by Chomsky “from behavior or the products of behavior to states of the mind/brain that enter into behavior” ([1986], 3; emphasis added).17 Devitt explains that he “is not concerned simply with the actual outputs”, but also with “possible idealized outputs” ([2006b], 24) of a system “when it performs as it should.” That is, the theory idealizes from actual products of behaviour to concern “with any of an indefinitely large number of outputs that they might produce” ([2006b], 18). However, with this counterfactual, “modal” terminology, Devitt disguises the central idea of generative linguistics as if his project is something other than Chomsky’s own enterprise. Good horseshoes and potential horseshoes are not physical in the sense required by Devitt, but a commitment to just the idealization of a grammar Chomsky describes as “a system of rules that generates an infinite class of “potential percepts” ([1972], 168), or perhaps potential horseshoes. Pylyshyn noted that “A constant source of misunderstanding and debate over the relevance of competence theories [to psychology] has to do with the fact that they define infinite sets” ([1973], 40). Devitt’s ([2006b], 27) preferred talk of there being “no limit” to “nonactual possible sentences” is simply code for Chomsky’s familiar talk of our Humboldtian infinite capacity through finite means and the Cartesian “creativity” of language. In particular, Devitt’s talk of a grammar being “lawlike” is simply a paraphrase for the generative capacity of recursive procedures. 16 Rawls [1955]. In a famous anecdote, three baseball umpires remark in turn: ‘I call ‘em as I see ‘em,’ said the first, an empiricist. ‘I call ‘em the way they are,’ said the realist. The third, Charlie Moran, explained: ‘They ain’t nothin’ until I call ‘em.’ 17 See also Chomsky ([2000], 5). P. Slezak, Linguistic Explanation and ‘Psychological Reality’ 15 15. Intuitions: Destroying the Subject. Devitt argues that “speakers’ intuitions are not the main evidence for linguistic theories” ([2006b], 96) and do not support the claims for grammars as mentally represented. Rather than being “the voice of competence”, Devitt suggests linguistic intuitions are “opinions resulting from ordinary empirical investigation, theory-laden in the way all such opinions are” ([2006b], 98). However, Chomsky ([1982], 16) has drawn an analogy with mathematics which can also be construed as a grammar representing tacit knowledge of conceptual structures – essentially the Intuitionist view according to which mathematical formalisms have no independent existence apart from the constructions of the mind.18 Gödel’s [1944] famous reference to mathematical intuition as a kind of perception is crucial here. Despite Devitt’s animadversions, this form of evidence is commonplace and uncontroversial elsewhere throughout psychology. Devitt explicitly rejects the analogy of linguistic intuitions with perceptual experience ([2006b], 112) but he draws precisely the wrong conclusion from his own allusion to Fodor’s [1983] account of the visual module.19 Devitt suggests that perceptual judgments correspond to “what is seen” which would be analogous to the content of what is said and not syntactic properties of expressions. However, the perceptual judgments that are relevant in the visual case are emphatically not “what is seen” in the “success” or “achievement” sense of such terms that Devitt ([2006b], 114) relies upon, but in the sense of what seems to be the case.20 Devitt ([2006b], 111) refers to “correct” intuitions and expresses doubts about “Cartesian access to the truth” ([2006b], 106) as if there is some objective fact of the matter beyond the subject’s perceptions. Thus, he asks “whose intuitions should we most trust?” ([2006b], 108) and answers that of “linguists themselves because the linguists are the most expert” ([2006b], 111). However, in psychology as in linguistics, there is no relevant expertise about the data beyond the authority of the subject’s own perceptions. In this sense, we are all experts or, in Hoffman’s [1998] term, “virtuosos.” The visual module doesn’t provide the truth about the distal stimulus but only a perceptual judgment, just like Chomsky’s grammar. The two interpretations of the Necker Cube are closely analogous to the two meanings of an ambiguous sentence as percepts of a native speaker. Chomsky explains: A grammar is a system of rules that generates an infinite class of “potential percepts”, … In short, we can begin by asking “what is perceived” and move from there to a study of perception. ([1972], 168,9) The place of intuition in grammars hardly deserves to be controversial 18 See Gil [1983]. See also Pylyshyn,( [1973], 31); Parsons [1995]. Ironically, Fodor illustrates his point with Ullman’s [1979] algorithmic theory, exactly the same example used by Chomsky ([1986], 264) to illustrate the nature of his own computational theory. 20 See Ryle ([1949], 152). 19 16 P. Slezak, Linguistic Explanation and ‘Psychological Reality’ unless the whole of perceptual psychology is also open to the difficulties alleged to arise for linguistics. Thus, Chomsky explained the interest of his famous pair ‘John is easy/eager to please’ saying that introspective “data of this sort are simply what constitute the subject matter for linguistic theory. We neglect such data at the cost of destroying the subject” ([1964], 79). 17. Conclusion The widely shared concern about “psychological reality” of grammars appears to be a manifestation of deep errors that are not difficult to recognize in other recalcitrant debates concerning mental representation (see Slezak [2002a]). Specifically, the debates turn on a certain illusion of explanatory adequacy arising when posited representations are intelligible in a direct, intuitive sense, like the “bewitching image of a word printed on a page” – Kirsh’s echo of Arnauld’s reproach to Malebranche in the 17th century. Searle’s [1980a] criterion of intentionality in his ‘Chinese Room’ is the intelligibility of symbols to himself, the theorist. Similarly, pictorial representations have been deemed appropriate explanations of visual imagery because they are intuitively intelligible to the theorist, as Pylyshyn [2003] has tirelessly complained. For closely related reasons, Carruthers ([1996], [1998]) has argued that we think in a natural language (see Slezak [2002b]). Ryle [1968] had warned against just this kind of error arising from invoking internal representations that have their meaning because we can understand them. Devitt’s complaint that Chomsky’s grammars may be the wrong sort of internal rules is implicitly to rely on this mistake, the same mistake as his assumption that external token physical symbols might have linguistic properties. Chomsky explained that traditional grammars produce an illusion of explanatory completeness, but in fact have “serious limitations so far as linguistic science is concerned” because the success of the grammar depends on being “paired with an intelligent and comprehending reader”. That is, it is the reader and not the grammar that is doing a significant part of the work. Chomsky explains: What he accomplishes can fairly be described as theory construction of quite a nontrivial kind. … The reader is, of course, not at all aware of what he has done or how he has done it. … Reliance on the reader’s intelligence is so commonplace that is significance may be easily overlooked. ([1962], 528,9) References Bechtel, W. and Abrahamsen, A. [2002], Connectionism and the Mind, Oxford: Blackwell. Berwick, R.C. and Weinberg, A.S. [1984], The Grammatical Basis of Linguistic Performance. Cambridge, Mass.: MIT Press. P. Slezak, Linguistic Explanation and ‘Psychological Reality’ 17 Botha, R. [1980], Methodological Bases of a Progressive Mentalism. Synthese 44, 1-112. Bromberger, S. [1989], Types and Tokens in Linguistics. In A. George (ed) Reflections on Chomsky. Oxford: Basil Blackwell; reprinted in his On What We Know We Don’t Know. Chicago: University of Chicago Press, 1992. Bromberger, S. and Halle M. [1992], The Ontology of Phonology. In S. Bromberger, On What We Know We Don’t Know. Chicago: University of Chicago Press. Carruthers, P. [1996], Language, Thought and Consciousness: An Essay in Philosophical Psychology. Cambridge: Cambridge University Press. Carruthers, P. [1998], Thinking in Language? In P. Carruthers and J. Boucher (eds) Language and Thought: Interdisciplinary Themes. Cambridge: Cambridge University Press. Chomsky, N. [1959], Review of Skinner’s Verbal Behavior. Language, 35, 1, 26-58; reprinted in L.A. Jakobovits and M. S. Miron (eds) Readings in the Psychology of Language. Englesood Cliffs, N.J.: Prentice Hall. Chomsky, N. [1962], Explanatory Models in Linguistics. In E. Nagel, P. Suppes, A. Tarski (eds) Logic, Methodology and Philosophy of Science. Stanford: Stanford University Press, 528-550. Chomsky, N. [1964], Current Issues in Linguistic Theory. In Structure of Language, J. A. Fodor and J.J. Katz (eds) New York: Prentice-Hall Chomsky, N. [1965], Aspects of the Theory of Syntax. Cambridge, Mass.: MIT Press. Chomsky, N. [1967], Preface to Review of Skinner’s ‘Verbal Behavior’. In Leon A. Jakobovits and Murray S. Miron (eds) Readings in the Psychology of Language. Prentice-Hall, 142-143. Chomsky, N. [1972] Language and Mind. New York: Harcourt Brace Jovanovich. Chomsky, N. [1975a], The Logical Structure of Linguistic Theory. New York: Plenum Press. Chomsky, N. [1975b], Reflections on Language. New York: Random House. Chomsky, N. [1980a], Biological Basis of Language Capacities, in Rules and Representations, New York: Columbia University Press. Chomsky, N. [1980b], Rules and Representations, and Peer Commentary Behavioral and Brain Sciences, 3, 1-61. Chomsky, N. [1982], The Generative Enterprise: A Discussion with Riny Huybregts and Hekn van Riemsdijk. Dordrecht: Foris Publications. Chomsky, N. [1986], Knowledge of Language: Its Nature, Origin and Use. New York: Praeger. Chomsky, N. [2000], New Horizons in the Study of Language and Mind. Cambridge, Cambridge University Press. Chomsky, N. [2003], Replies. In L.M. Antony and N. Hornstein (eds) Chomsky and his Critics. Oxford: Blackwell. Chomsky, N. & Katz, J.J. [1974], What the Linguist is Talking About. The Journal of Philosophy. 71, 12, June, 347-367. Clark, A. [1990], Connectionism, Competence, and Explanation. The British Journal for the Philosophy of Science, 41, 195-222; reprinted in M. A. Boden ed. 1990, The Philosophy of Artificial Intelligence. Oxford: Oxford University Press. 18 P. Slezak, Linguistic Explanation and ‘Psychological Reality’ Collins, J. [2006], Between a rock and a hard place: A dialogue on the philosophy and methodology of generative linguistics, Croatian Journal of Philosophy, 6, 18, 469-503. Cummins, R. [1996], Representations, Targets and Attitudes, Cambridge, Mass.: MIT/Bradford. Descartes, R. [1637/1965], Discourse on Method, Optics, Geometry, and Meteorology, Translated by Paul J. Olscamp, New York: Bobbs-Merrill. Devitt, M. [2003], Linguistics is not Psychology, in A. Barber ed. Epistemology of Language, Oxford: Oxford University Press. Devitt, M. [2006a], Intuitions in Linguistics, British Journal for the Philosophy of Science, 57, 481-513. Devitt, M. [2006b], Ignorance of Language, Oxford: Oxford University Press. Devitt, M. [2006c], Defending Ignorance of Language: Responses to the Dubrovnik Papers, Croatian Journal of Philosophy, 6, 18, 571-606. Devitt, M. [2007], Dodging the argument on the subject matter of grammars: A response to John Collins and Peter Slezak, manuscript. Devitt, M. and Sterelny, K. [1987], Language and Reality: An Introduction to the Philosophy of Language, Oxford: Blackwell; (Second edition 1999). Devitt, M. and Sterelny, K. [1989], Linguistics: What’s Wrong with ‘The Right View’? Philosophical Perspectives, 3, 497-531. Fodor, J.A. [1968], Psychological Explanation, New York: Random House. Fodor, J.A. [1983], The Modularity of Mind, Cambridge, Mass.: MIT Press. Fodor, J.A. [1990], Making Mind Matter More, in A Theory of Content and Other Essays, Cambridge, MA.: MIT Press. Fodor, J.A. [1998], In Critical Condition, Cambridge, Mass.: MIT Press. Fodor, J.A., Bever T.G. and Garrett, M.F. [1974], The Psychology of Language: An Introduction to Psycholinguistics and Generative Grammar. New York: McGraw Hill. Fodor, J.A. & Pylyshyn, Z. [1988], Connectionism and Cognitive Architecture: A Critical Analysis, Cognition 28, 3-71; reprinted in Pinker and Mehler eds., [1988], Connections and Symbols, Cambridge, Mass.: MIT Press. Gil, D. [1983], Intuitionism, Transformational Generative Grammar and Mental Acts, Studies in History and Philosophy of Science, 14, 3, 231254. Gödel, K. [1944], Russell’s Mathematical Logic, In P.A. Schilpp ed., The Philosophy of Bertrand Russell, Evanston: Northwestern University Press; reprinted in P. Benacerraf and H. Putnam eds, Philosophy of Mathematics, 2nd edition, 1983, Cambridge: Cambridge University Press. Hadley, R.F. [1990], Truth Conditions and Procedural Semantics. In Philip Hanson ed., Information, Language and Cogntion. Vancouver: University of British Columbia Press. Hadley, R.F. [1995], The “Explicit - Implicit” Distinction, Minds and Machines, 5, 219-242. Hoffman, D.D. [1998], Visual Intelligence: How We Create What We See, New York: Norton and Company. Horgan, T. & Tienson, J. eds., [1991], Connectionism and the Philosophy of Mind, Dordrecht: Kluwer. P. Slezak, Linguistic Explanation and ‘Psychological Reality’ 19 Kirsh, D. [1990], When is Information Explicitly Represented? in Hanson, ed., Information, Language and Cognition, Vancouver: University of British Columbia Press. Laurence, S. [2003], Is Linguistics a Branch of Psychology? in A. Barber ed. Epistemology of Language, Oxford: Oxford University Press. Lerdahl, F. and Jackendoff, R. [1983], A Generative Theory of Tonal Music, Cambridge, Mass.: MIT Press. Marr, D. [1982], Vision, New York: Freeman. Matthews, R.J. [2006], Knowledge of Language and Linguistic Competence, Philosophical Issues, 16, 200-220. Nelson, R.J. [1978], The Competence-Performance Distinction in Mental Philosophy, Synthese, 39, 337-381. Newell, A. and Simon, H.A. [1972], Human Problem Solving. Englewood Cliffs, N.J.: Prentice-Hall. Parsons, C. [1995], ‘Platonism and Mathematical Intuition in Kurt Gödel’s Thought’. The Bulletin of Symbolic Logic, 1, 44-74. Pinker, S. and Mehler, J. eds., [1988], Connections and Symbols, Cambridge, Mass.: MIT Press. Pinker, S. and Prince, A. [1988], On Language and Connectionism: Analysis of a Parallel Distributed Processing Model of Language Acquisition, in S. Pinker and J. Mehler eds. [1988]. Connections and Symbols, Cambridge, Mass.: MIT Press. Pylyshyn, Z. [1972], Competence and Psychological Reality, American Psychologist, June, 546-552. Pylyshyn, Z. [1973], The Role of Competence Theories in Cognitive Psychology, Journal of Psycholinguistic Research, 2, 1, 21-50. Pylyshyn, Z. [1984], Computation and Cognition, Cambridge, Mass.: MIT Press. Pylyshyn, Z. [2003], Seeing and Visualizing: It’s Not What You Think, Cambridge, Mass.: MIT Press. Quine, W.V. [1972], Methodological Reflection on Current Linguistic Theory, D. Davidson and G. Harman eds. Semantics of Natural Language, Dordrecht: D. Reidel. Ramsey, W., Sitch, S. and Garon, J. [1991], Connectionism, Eliminativism and the Future of Folk Psychology, in W. Ramsey, S. Stich and D. Rumelhart, eds., 1991 Philosophy and Connectionist Theory, Hillsdale, N.J.: Lawrence Erlbaum. Ramsey, W., Stich S., & Rumelhart, D.E. eds., [1991], Philosophy and Connectionist Theory, Hillsdale: Lawrence Erlbaum. Rawls, J. [1955], Two Concepts of Rules, The Philosophical Review, 64, 3-32; reprinted in S. Gorovitz ed., Utilitarianism: John Stuart Mill, New York: Bobbs-Merrill. Rey, G. [2003a], Chomsky, Intentionality, and a CRTT, in L.M. Antony and N. Hornstein eds., Chomsky and his Critics, Oxford: Blackwell. Rey, G. [2003b], Intentional Content and a Chomskian Linguistics, in A. Barber ed. Epistemology of Language, Oxford: Oxford University Press. Rey, G. [2006a], The Intentional Inexistence of Language – But Not Cars, in R.J. Stainton ed., Contemporary Debates in Cognitive Science, Oxford: Blackwells. 20 P. Slezak, Linguistic Explanation and ‘Psychological Reality’ Rey, G. [2006b], Conventions, Intuitions and Linguistic Inexistents: A Reply to Devitt, Croatian Journal of Philosophy, 6, 18, 549-569. Ringen, J.D. [1975], Linguistic Facts: A Study of the Empirical Scientific Status of Transformational Generative Grammars, in D. Cohen and J.R. Wirth eds., Testing Linguistic Hypotheses, Washington, D.C.: Hemisphere Publishing Corporation. Rorty, R. [1979], Philosophy and the Mirror of Nature, Princeton: Princeton University Press. Ryle, G. [1949], The Concept of Mind, London: Hutchinson. Ryle, G. [1968], A Puzzling Element in the Notion of Thinking, in P.F. Strawson ed., Studies in the Philosophy of Thought and Action, Oxford : Oxford University Press, 7-23. Searle, J. [1980a], Minds, Brains and Programs, Behavioral and Brain Sciences, 3, 417-24; reprinted in in M.A. Boden ed., [1990], The Philosophy of Artificial Intelligence, Oxford: Oxford University Press. Searle, J. [1980b], Rules and Causation: Reply to Chomsky, Behavioural and Brain Sciences, 3, 37-38. Simon, H.A. and Newell, A. [1964], Information Processing in Computer and Man, in Z. Pylyshyn ed., Perspectives on the Computer Revolution, New York: Prentice-Hall. Skinner, B.F. [1957], Verbal Behavior, NY: Appleton-Century Crofts. Slezak, P. [1981], Language and Psychological Reality, Synthese, 48, 3, 427440. Slezak, P. [2002a], The Tripartite Model of Representation, Philosophical Psychology, 13, 3, 239-270. Slezak, P. [2002b], Thinking About Thinking: Language, Thought & Introspection, Language & Communication, 22, 353-373. Slezak, P. [2006], ‘Cartesian ‘Ideas’ and the First (C17th) Cognitive Revolution.’ Ron Sun (ed.) Proceedings of the 28th Annual Conference of the Cognitive Science Society, Vancouver, BC, Canada, July 26-29, Mahwah, New Jersey, 798-803. Smith, B.C. [2006], Why We Still Need Knowledge of Language, Croatian Journal of Philosophy 6, 18, 431-456. Smolensky, [1988], On the Proper Treatment of Connectionism, Behavioral and Brain Sciences, 11, 1-74. Stich, S. [1972], Grammar, Psychology, and Indeterminacy, The Journal of Philosophy, 69, 22, December, 799-818. Stone, T. and Davies, M. [2002], Chomsky Amongst the Philosophers, Mind & Language, 17, 3, 276-289. Ullman, S. [1979], The Interpretation of Visual Motion, Cambridge, Mass.: MIT Press. van Gelder, T. [1990], Compositionality: A Connectionist Variation on a Classical Theme, Cognitive Science, 14, 355-384. van Gelder, T. [1992], Making Conceptual Space, S. David ed., Connectionism: Theory and Practice, Oxford: Oxford University Press Waskan, J. & Bechtel, W. [1997], Directions in Connectionist Research: Tractable Computations Without Syntactically Structured Representations, Metaphilosophy, 28, 1&2, 31-62. Wolf-Devine, C. [1993], Descartes on Seeing: Epistemology and Visual Perception, Carbondale: Southern Illinois University Press. P. Slezak, Linguistic Explanation and ‘Psychological Reality’ 21 Wolf-Devine, C. 2000a. Descartes’ theory of visual spatial perception, in S. Gaukroger, J. Schuster, J. Sutton, eds., [2000], Descartes’ Natural Philosophy, London: Routledge, 506-523. Wolf-Devine, C. [2000b], The role of inner objects in perception, in S. Gaukroger, J. Schuster, J. Sutton, eds., [2000], Descartes’ Natural Philosophy, London: Routledge, 557-568.