Whistling "Dixie": Response to Fodor

by Fiona Cowie

California Institute of Technology

Are you gonna pull those pistols or just whistle "Dixie"?
 Clint Eastwood, The Outlaw Josie Wales (1976)

        While there is much in Fodor’s ‘Doing Without What’s Within’ that I suppose I must try to be ‘philosophical’ about, there is rather less that I feel any need to respond philosophically to. Fodor takes issue with more or less every word of What’s Within (Cowie 1999), and rebutting each of his criticisms here would in effect require that I repeat the entire argument of the book. Since I’m certainly not about to do that, I will simply offer a few brief answers to those points in Fodor’s critique that seem to me to merit further discussion. For the rest, the book – moderately priced and available, I trust, from a bookseller near you – must speak for itself.
        

1. "[W]hy…should one believe that there is any such thing [as learning]?" ¹

       Fodor argues that there’s plenty of reason to believe in innateness because Mendel (inter alia) talks about it; whereas there’s no reason to believe in learning, because only associationists talk about that, and they’re stupid.
       But these days, smart geneticists do not talk about innate traits – or not without scare quotes.² Modern biologists, working from the molecular level up, generally acknowledge the murkiness of this notion (and related ones like genetic determination and genetic coding).³ They emphasize that phenotypes develop as a result of a vast and apparently very messy array of interchanges between genotype and environment. ⁴
        One reason, then, to believe in the existence of learning is that learning is a way that genotypes and environments could interact to produce phenotypes. And one reason to be suspicious of nativist accounts of development is that they claim to have already solved a problem – that of "[u]nderstanding how such interactions work" -- that even their proponents admit "may well be the hardest problem science has ever faced."
        Another reason to believe in the existence of learning is that there is good reason to think that it is, actually, something that the brain does. While much of the grossest cerebral architecture seems more or less impervious to environmental manipulation (except of the most invasive sort), evidence is mounting that the fine-tuning – and even the not-so-fine tuning – of cortical connections is highly sensitive to incoming information.⁵ One mightn’t want to call what the brain does in response to environmental stimuli "hypothesis testing" or "association of ideas" or what-have-you. But that’s OK: one needn’t. One can just call it ‘learning’ and leave it at that. For as Fodor rightly points out – albeit well before he stipulates that learning must be hypothesis testing – "science often starts in media res, finding out what it’s ‘really’ about as it goes along."
       A final reason to believe in learning is that the reasons offered by nativists for thinking that there isn’t any are unpersuasive. A careful reading of Fodor’s commentary suggests that this might be a point on which he and I disagree. My view was defended at length in What’s Within; Fodor’s defense of his is before you (or behind you, as it might be). I leave interested readers to examine the arguments and judge for themselves.
        

2. "An empiricist who’s not also an associationist has no cognitive psychology on offer at all"

        Fodor makes much of the fact that I grant that there is no satisfactory empiricist theory of learning available "yet." However he chooses not to respond to my claim that nor is there any satisfactory nativist theory of development available "yet." Part of the burden What’s Within (§§4.3-4.4) is to argue that waving one’s hands in the direction of imprinting (Fodor 1981) or skin color or birdsong or imprinting or the Babinsky reflex and saying ‘Well, you know, concept acquisition is kind of like that’ does not amount to providing a theory of concept acquisition. Likewise, I argued in §§10.6-10.7, although nativists about language learning have (to their credit) worked much harder at spelling out the necessary details than have nativists about concept learning, their efforts, on examination, fare little better as accounts of how the acquisition process actually works than do the empiricist theories Fodor so relentlessly derides.
        What What’s Within tries to do, in other words, is to dispel the notion that nativists have the game more or less sewn up, making any other approach to explaining concept or language acquisition otiose. No-one, I argue, is within cooee of a worked-out theory of how our minds get furnished, and we’d all be better off if we acknowledged this fact and then, unencumbered by spurious ideologies, knuckled down to some serious work on understanding development – work of the kind, in fact, that people like that Mark Seidenberg, Elizabeth Bates, Jeff Elman and others are already involved in. What we need, and what we are finally beginning to get, is theories of acquisition that take seriously the inborn constraints on neural architecture and the timing of various developmental sequences, while at the same time acknowledging the developing cortex’s enormous plasticity and sensitivity to environmental inputs.⁶
        

 3. "[W]hat Chomsky proposes is a nativism of domain specific prepositional attitudes (=PAs), not a nativism of domain-specific devices."

        Based on his reading of Chomsky’s (1965) book Aspects of the Theory of Syntax, Fodor argues that Chomsky’s main purpose, at least qua nativist and proponent of poverty of the stimulus arguments, is to defend the innateness of certain propositional attitudes (viz., our ‘cognizance’ of UG), and not to defend any view about the mechanisms of language learning.
        But not only does Aspects itself not support this claim (being liberally peppered with references to the "innate human faculté de langage" (e.g., pp.37, 57), the "language acquisition system" (e.g., pp. 53, 54) and the "language-acquisition device" (e.g., pp. 55, 56)), Chomsky’s later writings make it amply clear that his is a nativism of mechanisms, and not (or not primarily) of attitudes. Take this, for example:

It seems that there is little hope in accounting for our knowledge in terms of such ideas as analogy, induction, association, reliable procedures, good reasons and justification in any generally useful sense, or in terms of "generalized learning mechanisms" (if such exist). … We should, so it appears, think of knowledge of language as a certain state of … some distinguishable faculty of the mind -- the language faculty -- with its specific properties, structure and organization, one 'module' of the mind. (Chomsky 1986:12-13)

        Or this:

This work [i.e., the Chomskyan revolution in linguistics] proceeds from the empirical assumption -- which is well-supported -- that there is a specific faculty of the mind/brain that is responsible for the use and acquisition of language, a faculty with distinctive characteristics that is apparently unique to the species… (Chomsky 1991: 628)

        Or any of the following passages, drawn (literally) at random from the Chomsky corpus, the quotation of which I’ll spare you: 1968/72:172; 1986:3; 1988:47-8; 1992:34-5. When you look at what Chomsky actually says about his nativism, it's virtually impossible to believe that it has nothing essential to do with the postulation of a domain-specific device for language learning. Fodor remarks that it is "not always Chomsky’s way to make life easy for his exegetes" (n.5). On the contrary! Chomsky has always made it perfectly clear that his is a nativism of mechanisms, and that inborn attitudes come into the picture (if at all) only to the extent that the postulated mechanisms need make use of them.⁷
        As for Fodor’s additional argument that Chomsky couldn’t be talking about innate mechanisms because his distinction between observational and descriptive adequacy doesn’t apply to them, well that’s just a non-sequitur. The fact that Chomsky thinks that there’s a fact of the matter about which grammar is represented in speakers' heads in no way entails that he doesn't also think that there's a fact of the matter about what learning mechanism children use. Chomsky, after all, is famous for his resolute commitment to realism about mental states and processes – that’s why his (1959) always gets called things like ‘ground-breaking’ or ‘seminal.’ Just as there is, on his view, a fact of the matter about what grammar speaker-hearers cognize, there's a fact of the matter too about what processes they employ to learn it. So much so, indeed, that his view about those processes has radically changed over the years: he’s moved away from his earlier hypothesis-testing conception of the LAD and now embraces a parameter-setting account (see Cowie 1999: §7.1).⁸  It's unclear to me why someone who supposedly believes that there's "nothing to choose between equivalent theories of an internal mechanism" would bother to offer even one theory of such mechanisms -- let alone two.
       I’m going on about this at greater length than promised above for two reasons. First, because I think that it’s kind of important that one represent people’s views accurately, particularly if one is going to criticize them. And second, because (as Fodor acknowledges, pp.12-13) how you read Chomsky bears on Fodor’s own nativist project. If Chomsky is positing innate knowledge of UG, then all of Chomsky’s arguments for his nativism count as arguments for Fodor’s nativism too. For, as Fodor puts it, "there can’t be innate PAs unless there are innate concepts." But if Chomsky doesn’t give a damn about whether there is innate propositional knowledge of UG – if all he cares about is that the learning mechanism, somehow or another, be constrained by UG’s principles – then the ‘trickle down’ of evidential support that Fodor so desires (and, apparently, needs) threatens to dry up.
        

4. The psychological reality of grammars -- redux all over again.

        Fodor credits me with arguing that since there’s no reason to think that grammars are ‘psychologically real,’ there’s no reason to think that they’re learned, hence there’s no reason to think that UG is innate. This characterization of my argument is simply false. Hence most of Fodor’s §2.2, in which he defends the psychological reality of grammars, is beside the (or at least my) point.
        To be sure I do, in §§10.2-10.3, take a quick glance at the old arguments about psychological reality. But this is solely in the context of establishing the following point: the claim that speakers know (/cognize) a grammar is an empirical hypothesis about the mental processes underlying linguistic competence. Period.⁹ And while some philosophers have questioned this claim, I don’t. Rather, and for much the same reasons as Fodor belabors, I take it completely for granted in What’s Within that grammars are psychologically real and that mastering a language is (at least in part) a matter of internalizing one.
        Why then, do I mention the old debate and bother to make such a boring and uncontentious point? Because if you thought that the claim that grammars are psychologically real were not an empirical hypothesis – if, for instance, you thought that it followed, a priori, from consideration of what linguistics is about – then you could argue that it’s also knowable a priori that the language-learning mechanism is constrained by UG, on the grounds that this claim follows from consideration of what Linguistic Theory is about. That is, you might argue that Chomskyan nativism is true a priori. But since this is (i) a really bad argument; and (ii) not consistent with linguistic’s claim to be an empirical scientific enterprise; and (iii) not Chomsky’s most persuasive argument for his nativism, I wanted to set it aside. I wanted to concentrate, rather, on the good and potentially convincing argument for Chomsky’s nativism, namely, the inference to the best explanation that goes like this:

1. The hypothesis that there is an inborn language learning mechanism constrained (somehow) by the principles of UG offers the best explanation of how linguistic competence/knowledge of grammar is acquired.
2. Therefore, there is such an inborn mechanism or ‘special faculty

        What I really argued in What’s Within, then, is not that a whole bunch of philosophizing has to be done before we can tell whether Chomskyanism is true (pace Fodor); but rather that, as a matter of empirical fact, the first premises of the argument just displayed is false. §§10.5-10.7 argue that neither hypothesis-testing nor parameter-setting models of language acquisition provide adequate explanations of how children become competent in their language. A fortiori, they do not provide the best explanations of how children acquire language. Thus, I conclude, an inference to the existence of a UG-based language faculty from the explanatory successes of UG-based theorizing is unsound.¹⁰ And thus, I conclude, the evidence for Chomskyan nativism is unconvincing. Fodor says: "One vindicates the ontology and methodology of a science by appeal to the work they [sic] do, NOT THE OTHER WAY AROUND." And I say: precisely.
        

5. "I think Cowie’s failure to understand how best explanation arguments work undermines quite a lot of her book."

        Fodor argues that none of this undercuts the real abductive argument in Chomsky’s favor – the one I’m supposed to have missed and which according to Fodor goes like this:

What else but grammars being mentally represented could explain their empirical successes [i.e., in explaining how we speak and understand language]? What else but UG’s being innate could explain the child’s ability to assimilate the grammars whose predictive/explanatory successes the story about grammars being mentally represented is supposed to account for?

        Fodor says that all I’ve argued is that Chomsky’s nativism might not explain childrens’ ability to assimilate their grammars, and that this doesn’t undercut the claim that Chomsky’s nativism is true because it’s really good at explaining a bunch of other stuff: " 'T è D and maybe not D' simply does not rebut, or even get a leg up on rebutting, 'T because it explains E.' "
        First, and apropos inferences to the best explanation, there is considerable disagreement among philosophers of science, both about the merits of inference to the best explanation arguments and about how the notion of a 'best explanation' is to be understood. However, defenders of such inferences do tend to agree that theories that are candidate best explanations of the phenomena in some domain must at least be consistent with those phemonena. That is, in order to run an inference to the best explanation argument for T based on its ability to explain E, T must not imply other claims (D) about the domain it professes to explain that are thought to be false. Thus, and contra Fodor, if Chomskyan nativism did make false predictions (D) about phenomena within its domain, then that would put it outside the reach of an inference to the best explanation, and it would do so notwithstanding the fact that Chomskyan nativism, were it true, could explain other phenomena (E) within that domain.
        Second, note that the structure of the ‘real’ argument for Chomskyan nativism – the argument Fodor gives in the passage quoted above – is not as Fodor’s schematic characterization suggests. The ‘real’ argument involves two separate abductive inferences, and goes like this:

1. G (grammars are psychologically real) because this explains L (linguistic behavior)
2. UG (Chomskyan nativism) because this explains A (acquisition of a grammar)

   with the additional premiss that knowing a grammar necessitates that one somehow acquire one:

3. G requires A

being enthymematic in Fodor’s presentation.

        As we’ve already seen (§4 above), I’ve nothing against (1). As to (2), Fodor takes me to be arguing that that inference is unsound because UG implies some other proposition, D, that might be false. But that’s not what I’m doing! In §§10.5-10.7 I attack (2) directly, and argue that UG does not explain A. My point, in other words, is that Chomskyan nativism does not explain how children learn the grammar of their language, and if that is so, then the inference in (2) fails.
        

6. "Cowie has nothing against POSAs as a form of argument; she just doubts that, in the case of language acquisition, its empirical assumptions are true."

        Right, and what’s wrong with that? It’s generally accepted, even within philosophy, that you can’t cite a fact as evidence for a theory unless that fact is itself well attested (or at least, not in serious dispute). Hence, you can’t cite the poverty of the stimulus as evidence for linguistic nativism unless that poverty is – if not well attested, then not seriously open to question.
        What I tried to show in What's Within is that the poverty of the stimulus is currently in serious dispute (Pullum’s tongue-in-cheek references to Lady Bracknell notwithstanding – see his (1996) and the rest of the discussion in §§8.3 and 9.5) and that nativists have done nothing – beyond thumping the table and saying it again, louder (cf. Rey 2000) – to defend their crucial premiss.
        Look, I do understand that a lot of people (Fodor clearly included) think that it’s just obvious that children are not 'likely' to hear sentences that would enable them to figure out that they should use the structure-dependent rule for forming yes-no question or when to contract want to to wanna. I understand too that a lot of people think that it’s just obvious (maybe even a priori) that there is no (or not nearly enough) negative evidence around to enable learners to recover from overgeneralizations. "Pas devant les enfants," as Fodor puts it.
        But I don’t find any of this obvious at all, for the reasons I outline in What’s Within (chs. 8, 9). Tout devant les enfants is how it seems to me (and a bunch of other people, too). Being that my intuitions about the sorts of grammatical structures that kids between ages (say) 0 and 8 have access to are presumably no more or less probative than Fodor's or Chomsky's, what we need are a few data to settle the matter one way or the other. And that, as I point out in What's Within, is precisely what we don't have. All that skeptics about the poverty of the stimulus are asking for is a bit of actual evidence (by preference something a bit younger than Brown and Hanlon (1970)) to make it plausible that children in fact don't hear and/or can't process and/or don't hear enough of even one structure that they allegedly couldn't learn a language without.
        

7. "[P]hilosophical arguments are supposed to be knock-down; or better, lethal…"

        In perhaps the most strained of his critical sorties, Fodor castigates me both for allegedly failing to realize that Chomsky's nativism is an empirical hypothesis that's not susceptible of knock-down proof, and for failing to provide a knock-down 'philosophical' argument against it (or for an empiricist alternative). Complaining about my (supposed) repudiation of the idea that grammars are internally represented, he writes in defense of Chomsky's inference to the psychological reality of UG and grammars: "Well, what else if not ‘what else’ arguments would you expect to ground an empirical inference from data to theory? Empirical inferences are ipso facto not demonstrative." Quite. I’m all for empirical, non-demonstrative inferences – so much so, indeed, that What’s Within is full of them. And there, apparently, is the rub. Fodor rejects my 'empirical inferences' from the data to the falsity of Chomsky's theory on the grounds that "Cowie needs a principled reason for doubting that the problem about how UGs function in language acquisition can be solved; but all she's got is that, to date, nobody has solved it."
        It seems that Chomsky (the linguist) is allowed to make empirical, nondemonstrative "what else?" inferences in favor of his position, whereas Cowie (the philosopher) isn't allowed to make empirical, nondemonstrative "not yet" arguments against it. For "philosophical arguments are supposed to be knock-down; or better, lethal" but "[l]inguistics…is different. Like any other empirical discipline, it appeals to a balance of plausibility."
        Three short points in response to all this. First, I repudiate this characterization of the philosophical method. Wherever did philosophers get the idea that they were equipped -- let alone required -- to provide 'knock-down' answers to empirical questions? Second, and even if Fodor were right that my argument is not a "philosophical" argument, it's still an argument, and as such needs to be addressed. The fact that I might be confused about my proper job description does not invalidate the points I make. Finally, it is particularly ironic that Fodor should be willing to dismiss my argument on methodological grounds such as these, given that earlier in his paper he was the one castigating me for (allegedly) belonging to a "tradition of trying to settle issues [in linguistics] by taking sides on issues about … methodology…"
        

8. "There's a dumb joke about an enlightened empiricist who could count sheep very fast."

        I admit that the position I called Enlightened empiricism – the potential alternative to nativism that What’s Within argues had been unfairly ignored by nativists and empiricists alike – was not adequately discussed in the book. This was a mistake, and if I had it to do over again, I'd do it differently. In any case, I now have a chance to say a little more about how Enlightened empiricism is supposed to work, and I trust that by doing so, I can make clear why Fodor's objections to the position in his §2.3.3 are premature.
        So: What we all should believe – what the poverty of the stimulus and logical problem arguments and good old common sense tell us – is that a language-learner (call her Sarah) needs to be able to make use of what she already knows about language when she is updating her grammatical hypotheses in light of incoming data. So, for instance, suppose that yesterday, I said to Sarah: ‘Don’t you know what that is on there? Don’t you realize what that’s doing on there?’ Sarah's problem (one of them, anyway -- the other was that she'd pulled the ‘child-proof’ cover off an electrical outlet and thus was in serious trouble) is to figure out what rule is governing is-contraction here – for one time I contracted it, and the other time I didn’t. There are, as we all know, lots of rules she could try, and most of them would be wrong.¹¹ Clearly, since she will end up getting the rule right, something must be constraining her theorizing about this topic. And clearly too, since waffling on about her preferring ‘simple hypotheses’ and the like has been so thoroughly discredited, what’s constraining her in her theorizing about this rule must be what she already knows about how language works – domain-specific linguistic information, in other words.
        Thus far are the Enlightened empiricist and the Chomskyan nativist in accord. Where their views diverge is, first, in their accounts of the aetiology of the domain-specific information young Sarah has access to. The nativist assumes that Sarah could only be making use of domain-specific information to learn the rule for is-contraction if that information were innately encoded as a constraint on the learning mechanism. That is, the nativist infers from 'Sarah would need to know that p in order to learn rule R' to 'p is in inborn constraint on the acquisition mechanism.'
        The Enlightened empiricist resists this inference. He argues that Sarah could have gotten the information she needs from her prior experience -- presumably of language. Thus, for example, a Bayesian might argue that how Sarah changes her theory of grammar in response to my utterance depends on her current state of linguistic belief, as reflected in her priors; which in turn depend on her previous experiences and beliefs. If this Bayesian also thought that the initial probabilities in the grammatical domain were set randomly, then she would be an Enlightened empiricist: she'd be positing a learning mechanism that is both general purpose and able to make use of domain-specific knowledge aquired along the way.¹²
        A second point at which an Enlightened empiricist and a Chomskyan would almost certainly disagree is in their accounts of the kinds of domain-specific information Sarah's theorizing is being constrained by. That is, while Fodor is right to say that "’Enlightened Empiricism’… allows that language acquisition may crucially require prior knowledge of the domain specific sort that UG provides," (emphasis added) no Enlightened empiricist in his right might would actually propose this, and for precisely the reasons that I discuss at length in What's Within §10.1 and Fodor goes on to elaborate. It is absurd to think that children first learn UG, and then learn language. God wouldn't be that stupid. But what that shows, according to the Enlightened empiricist, is that whatever domain-specific information children are invoking, it isn’t knowledge of UG.¹³
        BUT counters the nativist (and Fodor): it must be UG that is constraining Sarah's theorizing, else we have no explanation of how she could acquire a grammar that conforms to it! To which the Enlightened empiricist responds as follows. ‘You're right that I don't have a "'detailed' alternative to nativism 'on hand' 'yet'."’ (p. 29) But you're wrong if you think that that is a reason to believe in Chomskyan nativism. Since Chomskyans don't have a workable theory "‘on hand’ ‘yet’" either -- since neither of us can explain how Sarah learns a grammar confirming to UG -- the score remains love all.
        

9. The Parable of Anon.

        In what is surely the most bizarre move in a paper not otherwise lacking in polemical eccentricity, Fodor offers a "parable" by way of arguing that the very idea of Enlightened empiricism is bunk. He argues that postulating a general purpose learning device to account for language acquisition is like postulating a general purpose question answerer (GPQA) to account for a certain fellow's truly remarkable facility for answering questions about opera. The latter hypothesis is very silly (it’s this guy's knowledge of opera that’s doing the job, stupid); and therefore, so is the former.
        But the case of 'Anon.,' as Fodor calls his opera buff, is not analogous to the case of someone's learning language. Rather, it's analogous to the case of a person who is competent in her language. Just as it’s Anon.’s knowledge of opera that’s enabling him to answer questions about opera, it’s that person’s knowledge of grammar that enables her to speak and understand her language.¹⁴ So far, so good. Fodor and I, as we’ve seen before, agree about the plausibility of positing knowledge (/cognizance) of grammar to explain linguistic competence, just as we agree about the plausibility of positing knowledge (/cognizance) of opera to explain Anon.'s remarkable ability.
        Trouble is, none of this speaks to the point that I take to be at issue at this stage in the dialectic, which concerns the question: how do people learn a language? If Anon. is to demonstrate the implausibility of a non-UG-based (or general-purpose) approach to language learning, then his story would have to continue thus:

…And then all the clever people began to wonder how Anon. acquired his remarkable store of operatic knowledge. Clearly, only stupid people would think that Anon. was just a guy who listened to a lot of LPs and went to the Met at lot and read a lot of scores. For (the smart people said) it’s very unlikely that the operatic inputs to which Anon. has had access would suffice to enable him to memorize the mass of recondite facts that would enable him to do things like naming all the composers of operas with 'La' in their titles. Anon., the clever people said, must be possessed of a special inborn faculty -- an 'opera faculty' or 'OKAD' (‘Opera Knowledge Acquisition Device’) – that enabled him to acquire the competence in question. Just as only stupid people would deny that we all are possessed of a special inborn faculty -- a 'language organ' or 'LAD' -- that enables us to acquire language, only an idiot would deny the appropriateness of nativism as an explanation of Anon.'s feat.

Away, away, away down south in Dixie…
        

10. "Cowie has two objections to Impossibility arguments.

        Fodor says that I am committed both to the view that there couldn’t be innate concepts and to the view that there must be some innate concepts because some concepts are learned (and learning any concepts requires that you have some of them to start with). "I really don’t think," Fodor says, that " Cowie can have this both ways." And he’d be right, if I were arguing what he says that I am, which I’m not.
        I don’t argue that there couldn’t be innate concepts. What I argue (§§4.3-4.4) is that it’s hard to see how there could be innate concepts given Fodor’s picture of what concepts are. Fodor believes (or at least, used to believe circa his (1981)) that a concept’s semantic properties are exhausted by its reference: all there is to a concept is its ability to refer to something in the world. Fodor also believes (/used to believe) that a concept’s reference is determined by its relations (causal, informational, nomic, whatever) with the world and that these relations begin when the concept is, as he says, ‘triggered’ by experience. But this raises a problem, given that Fodor also believes (/believed) that most concepts are innate. For if triggering supplies reference, and concepts’ reference exhausts their semantic properties, then what was the thing that got triggered, and in what sense could it be said to be intentional (i.e., in what sense could it be an innate concept)?¹⁵ I raised this question, and, after trying a bit to figure it all out, failed to answer it. I couldn’t see how, on Fodor’s ‘minimalist’ view of concepts, you could make sense of the idea that they were innate.
        So instead, I supposed (§4.5) that what Fodor must mean when he says that there are innate ideas, is that there is some or other kind of inborn brute-causal mechanism which, in response to a given stimulus, S, delivers a concept of S-ness as output. (The brute-causality or non-intentionality of the mechanism is needed to differentiate it from – horrors! – a mechanism of learning. Cf. Fodor 1981.) To put it in more up-to-date-terms, what he must have been saying, I thought, is that what is innate is a brute-causal mechanism in virtue of which certain states of our brain covary with (or resonate to) instantiations of S-ness in the world in the ways that, on Fodor’s view, enable us to satisfy the possession conditions for having the concept S.
        Given the best understanding I could manage of what Fodor’s claims about the innateness of concepts might mean, I then went on to examine his further contention that "most" of our concepts are innate. Are most the mechanisms in virtue of which we resonate to the referents of our concepts (i) brute-causal and (ii) inborn? After much arguing too and fro, §6.1-6.3, I concluded that this last claim was implausible. While some of these mechanisms must clearly be both inborn and brute-causal (or non-intentional ¹⁶) – while there are clearly some ‘primitive’ or ‘innate’ concepts – some of them are neither. Some resonations, I argued, are intentionally mediated (e.g., perhaps, those allowing me to resonate with carburettors); and some of the mediating mechanisms are products of a learning process (e.g., again, those underpinning my possession of CARBURETTOR).
        So here’s what I think I was arguing: You can’t have innate concepts as Fodor conceives of them, for there’s nothing in a Fodorean unstructured protoconcept that could be inborn. But if you’re prepared to allow that some concepts have internal structure (the ‘structure’ at issue being the mediating mechanisms that get you resonating to things in the world and fix your concepts’ reference) then you can think that concepts are innate – even if you also think, as I do, that "most" of them are not.¹⁷
        

11. "[E]mpiricists and rationalists are both partly right about where concepts come from…"

The quotation heading this section continues:

The acquisition of DOORKNOB, for example, has two (not necessarily successive) phases: One of them maps from (eg.) doorknob experiences to (something like) a doorknob prototype. Since prototype formation is generally held to be a species of statistical inference, this phase of concept acquisition approximates to being a rational process, just as empiricists would like. But, as we’ve seen, prototypes don’t compose, so they aren’t the right kind of mental representations to be concepts; or even to be components of concepts…So there has to be another phase of DOORKNOB acquisition; namely a mental representation that is ‘locked’…to an extension that includes all and only doorknobs.

Excepting only the disclaimer that prototypes aren’t parts of concepts (and this issue, I suspect, is largely verbal), this is exactly what I argued must be the case in §6.4 of What’s Within, and I’m just tickled pink to find that Fodor shares my view.¹⁸, ¹⁹

Notes

^1_ Unless otherwise noted, all quotations are from Fodor 2000

^2_ See Wimsatt (1999) for an account of the conceptual messes that the not-so-smart can get themselves into when they use the term ‘innate.’ Wimsatt distinguishes no fewer than 28 different senses of ‘innate’ at work in the philosophical and biological literatures. See also his (1986).

^3_ Cf. Godfrey-Smith (1999, 2000).

^4_ Cf. Gerhart, Kirschner and Moderbacher (1998).

^5_ Cf. the first 250-odd pages of Gazzaniga (2000); or see Elman et al. (1996: ch.5) for a review.

^6_ Cf., e.g., Elman and Bates (1996), Bates et. al (1999) and Bates and Dick (2000).

^7_ I have to admit that it makes me nervous that Chomsky is one of the people whom Fodor thanks for comments on his paper (n.46). Did Chomsky agree with Fodor’s characterization of his views? If so, how is that he could have said repeatedly, over 40-odd years, that p, all the while meaning not p, but rather q? (This, I guess, is a cry for help.)

^8_ Unless otherwise noted, all future references to sections (§) are to Cowie (1999).

^9_ A psychological hypothesis, a linguistic hypothesis – I don’t care what you call it, so long as you think it’s empirical.

^10_ Note: I don’t deny that some future UG-based account of the language faculty might enjoy the kind of success that would license Chomsky’s abduction here.  My point is just that, as of now, there isn’t one that does the job.  Thus – and pace Professor J.D. Fodor’s private assurances to her husband that “some pretty good stuff has turned up” (Fodor, n.16) – this argument speaks to my ‘no-one has a theory of acquisition’ theme, mentioned in §2 above.

^11_ One version of the right rule is: “Don’t contract is to –‘s when there’s a wh-trace intervening between the is and the material to its right.”

^12_ One hesitates in the present context to mention that many connectionist learning algorithms are also Enlightened emipricist in tenor. So one hastens to add that, in addition to Bayesian models, so is Glymour-style causal inference, which infers causal relations (represented as directed graphs) from correlational data. Although this model makes use in the first instance only of domain-neutral constraints (the Causal Markov and Faithfulness conditions) in inferring causal structure from the data, it counts as an Enlightened empiricist model because it also allows for the imposition of additional, more domain-specific constraints on certain inference problems. For instance, one may impose the Causal Sufficiency condition, which reflects the (domain-specific) fact that all the common causes of the measured variables in the domain have themselves been measured.  See Glymour, Scheines, Spirtes and Kelly (1987) and Glymour (1997) for discussion of the theory. See Scheines (1997) for a comparatively non-technical overview. It would be interesting to see how well a Glymour-style model works in inferring grammatical relations from distributional data. There's a concrete empiricist research project for you.

^13_ E.g., our Bayesian, mentioned earlier, would presumably think that it’s the probability distribution over the entire body of linguistic knowledge, not any specific principle, that is constraining how Sarah updates.

^14_ Actually, this reading is probably overcharitable. What the case of Anon. is really analogous to is the case of a person who is able not to speak and understand a language -- Anon., we assume, is no Bryn Terfel -- but rather to answer questions about her language. But since people aren't much good at doing this, it's unclear what moral, if any, one should draw from Fodor's fable. (N.B.: Apart from recommending that linguistics departments be emptied forthwith of Ph.D.'s and populated with 3-to-8 year olds, I shall draw a veil over Fodor’s claim that “Normal human children are…quite extraordinarily good at answering questions of the form ‘what grammar underlies the language of which the following corpus is a sample (insert PLD here)’” I take this to be in line with the aforementioned principle of charity.)

^15_ Remember, “the argument about innate ideas…was an argument about whether there are innate ideas” (p.5).

^16_ I called them ‘cognitive reflexes’ in the book.

^17_ See the discussion of mediating mechanisms in §6.4.1

^18_ For a convincing account of how all the details of such a picture might work, see Prinz (in press).

^19_ Thanks to Alan Hájek, Steve Quartz, Kayley Vernallis and Jim Woodward for their help and comments on this material.

Bibliography

Brown, R. and Hanlon, C. (1970): 'Derivational Complexity and Order of Acquisition in Child Speech,' in J. R. Hayes ed., Cognition and the Development of Language, John Wiley and Sons, New York.

Cowie, F. (1999): What’s Within? Nativism Reconsidered (Oxford University Press, New York).

Chomsky, N. (1959): Review of Skinner's Verbal Behavior, Language, 35, pp. 26-58.

_____ (1965): Aspects of the Theory of Syntax, MIT Press, Cambridge, Mass.

_____ (1968/1972): Language and Mind, 1st pub'd 1968. 'Enlarged Edition'pub'd 1972, Harcourt Brace Jovanovich, Inc., NY. References are to the 1972 edition.

_____ (1986): Knowledge of Language, Its Nature, Origin and Use, Praeger,NY.

_____ (1988): Language and Problems of Knowledge, The Managua Lectures,MIT Press, Cambridge, Mass.

Chomsky, N. (1991): 'On the nature, use and acquisition of language,' inLycan (1991:627-45).

_____ (1995): Language and Thought, Wakefield : Moyer Bell.

Bates, E. and Elman, J. (1996): ‘Learning Rediscovered,’ Science, 274, pp.1849-50.

Bates, E., Elman, J., Johnson, M., Karmiloff-Smith, A., Parisi, D., & Plunkett, K. (1999): ‘Innateness and Emergentism.’ In W. Bechtel & G. Graham (Eds.), A Companion to Cognitive Science (pp. 590-601). Oxford: Basil Blackwell.

Bates, E., & Dick, F. (2000) ‘Beyond Phrenology: Brain and language in the next millennium,’ Brain and Language, Vol. 71, No. 1, Jan 2000, pp. 18-21

Elman, J.L., Bates, E.A., Johnson, M., Karmiloff-Smith, A., Parisi, D., and Plunkett, K. (1996): Rethinking Innateness: A connectionist perspective on development. Bradford/MIT Press, Cambridge Ma.

Fodor, J.A. (1981): 'The Present Status of the Innateness Controversy' in RePresentations: Philosophical Essays on the Foundations of Cognitive Science, MIT Press/Bradford Books, Cambridge, Mass.

Fodor, J.A. (2000): ‘Doing without What’s Within: Fiona Cowie’s critique of nativism’ to appear also in Mind, (2001)

Gazzaniga, M.S. (ed) (2000): The New Cognitive Neurosciences, 2^nd Edition, (Bradford/MIT Press, Cambridge, Ma).

Gerhart, J., Kirschner, M., and Modernacker, E.S. (1998): Cells, Embryos and Evolution: Toward a cellular and developmental understanding of phenotypic variation and evolutionary adaptability, Oxford: Blackwell

Godfrey-Smith, P. (1999): ‘Genes and Codes: Lessons from the Philosophy of Mind?’ in V. Hardcastle (ed.), Where Biology Meets Psychology: Philosophical Essays. (MIT Press, Cambridge, Ma).

_____ (2000)" ‘On the Theoretical Role of 'Genetic Coding,"’ Philosophy of Science 67, 26-44.

Prinz, J. (in press): Impressions and Ideas, Cambridge, Ma., Bradford/MIT Press.

Pullum, G.K. (1996): ‘Learnability, Hyperlearning and the Poverty of the stimulus,’ J. Johnson, M.J. Juge and J.L. Moxley (eds.), Proceedings of the 22^nd Annual Meeting: General Session and Parasession on the Role of Learnability in Grammatical Theory (Berkeley Linguistics Society, Berkeley) pp.498-513.

Rey, G. (2000): ‘Going Native,’ Times Literary Supplement, August 11, 2000. pp.26-7.

Seidenberg, M.S. (1999): ‘Do infants learn grammar with algebra or statistics?’, Science, 284, pp.434-5, discussion 436-7.

Wimsatt, W. C. 1986; Developmental constraints, generative entrenchment, and the innate-acquired distinction, in W. Bechtel. ed. Integrating Scientific Disciplines, (Martinus-Nijhoff) pp.185-208.

_____ (1999) ‘Generativity, Entrenchment, Evolution, and Innateness,’ in V. Hardcastle, ed., Where Biology Meets Psychology: Philosophical Essays. Conjectures (MIT Press): XX-YY

top | back to symposium index | Fodor's paper