The Argument for Concept Splitting from Language

In our forthcoming paper, “Splitting Concepts,” Sam Scott and I argue, among other things, that the notion of concept may need to be split into linguistic representations (responsible for cognition that involves language) and nonlinguistic representations (responsible for the rest of cognition). Roughly, the reason is that linguistic cognition appears to require representations with more expressive and inferential power than the rest of cognition. Another way of putting the point is, creatures that can learn and master languages are much smarter than creatures that cannot. We need an explanation for this fact, and a reasonable explanation might involve concepts of radically different kinds.

In his comments on the version of the paper that we presented at last year’s SPP, Dan Ryder suggested that the argument from language at most shows that syntactic linguistic representations are special, whereas semantic representations (i.e., concepts) may be left unaffected.

In the paper, we have a multi-pronged response to this worry.

First, in so far as syntactic representations are needed to explain linguistic cognition, they belong in the theory of concepts, broadly construed. If you will, the concepts in question are concepts of syntactic categories, rather than concepts of kinds and properties in the domain of discourse. Nevertheless, they are concepts in the same sense in which other representations are concepts, and the fact that they are usually not called concepts in the literature is only a terminological point.

Second, the exact relationship between semantic and “syntactic” representations are controversial. Depending on what they are, the argument might affect semantic representations too. (E.g., perhaps there is no sharp distinction between syntactic and semantic representations.)

Finally, even if semantic and syntactic representations are sharply distinct, it remains possible (though we don’t argue for it) that semantic linguistic representations are different in kind from non-linguistic ones.

In a comment to a previous post, Dan Ryder expresses skepticism about our response to his worry. He writes:

“First: surely the psychologists’ and linguists’ use of a different term here indicates that they think syntactic processing involves a different kind of representation, i.e. some non-conceptual kind of representation – and isn’t the paper supposed to be about the scientists’ notion of concepts? And second, the theory of concepts is not “the theory of the representations that explain phenomena (1) to (6).” There’s no such thing as *the* representations that explain (1) to (6). (1) to (6) will involve all sorts of perceptual representations, for instance, and most scientists doubt those belong to the same kind of representation that concepts do. (Note that many think that syntactic representations are more like perceptual representations than conceptual ones.)”

Dan’s comments are relevant and helpful, but they do not affect the important point underlying our argument.

With respect to the terminological point, I agree that people typically use different terms to mean different things. The question is whether the difference is relevant for present purposes. Language is used to represent the domain of discourse, and in that respect, there is a useful distinction between semantic representations (which represent objects and properties in the domain of discourse) and syntactic representations (which do not). But syntactic representations still represent: they represent properties of linguistic structures (which are still aspects of the world, by the way). So in so far as “concept” means, roughly, representation of some aspect of the world, both semantic and syntactic representations are concepts. Also, “concepts” as psychologists use the term are representations postulated to explain certain cognitive capacities. So in so far as both semantic and syntactic representations are needed to explain the same capacity, they belong in the same psychological theory. Bottom line: given the way the term “concept” is used in the literature, there is one respect (here not very important) in which syntactic representations do not count as concepts, but there are other respects (here relevant) in which they do. And by the argument from language, linguistic representations (syntactic, “semantic,” or both) are different in kind from nonlinguistic ones.

With respect to Dan’s second point (contrasting perceptual and conceptual representations), I am skeptical of the traditional constrast between perceptual and conceptual representations. I think all representations are “perceptual”, at least in the minimal sense that they originate with the brain’s processing of perceptual information. And I think all representations are “conceptual,” at least in the minimal sense that they discriminate between what falls under them and what doesn’t. Perhaps the perceptual-conceptual dichotomy constitutes a continuum rather than a sharp divide. (BTW, I’m taking no stance with respect to the nativism-empiricism debate.) But this way of putting things is inadequate, because it uses the ambiguous term “concept”. The whole point of our paper is that there is no single notion of concept: there are many. In one sense, concepts are linguistic representations. In another sense, they are representations undelying nonlinguistic cognition. (And there may be other ways that concepts split.)