![Sample our Humanities journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5939/TOC-Subject-Sample-2021-Humanities.jpg"})
Abstract
In this paper, I present an alternative argument for Jerry Fodor's recent conclusion that there are currently no tenable theories of concepts in the cognitive sciences and in the philosophy of mind. Briefly, my approach focuses on the ‘theory-theory’ of concepts. I argue that the two ways in which cognitive psychologists have formulated this theory lead to serious difficulties, and that there cannot be, in principle, a third way in which it can be reformulated. Insofar as the ‘theory-theory’ is supposed to replace, and to rectify the problems of, the earlier ‘classical’ and ‘probabilistic’ theories, its failure confirms Fodor's original observation. Since my critique does not rest on controversial philosophical assumptions and is readily available from within the cognitive sciences, it is a stronger argument than Fodor's.
Keywords:
Acknowledgements
I am grateful to William Seager, Ronald de Sousa, Gurpreet Rattan, and Jennifer Harker for their help and advice in writing this paper. I would also like to thank two anonymous referees of this journal for their insightful comments and suggestions.
Notes
1. The ‘theory-theory’ is also sometimes referred to as a ‘theory-based’ or ‘explanation-based’ theory of concepts. It is important to distinguish this sense of ‘theory-theory’ from what some philosophers, such as Nichols and Stich Citation(2003), have also labeled as the ‘theory-theory’, which primarily relates to self-awareness.
2. For a detailed overview of these theories, see Komatsu Citation(1992), Laurence and Margolis Citation(1999), Medin Citation(1989), Murphy Citation(2002), and Smith and Medin Citation(1981).
3. Concepts will be denoted by small capitals.
4. To see this, consider the following: For the concept bird, the following stereotypical properties may be represented under the prototype: ‘flies’, ‘sings’, ‘is brown’, ‘has a beak’, ‘is small,’ ‘has webbed feet’, etc. If correlations among properties can be represented simply by virtue of listing each of the correlated properties in the representation, then many correlations will result: small birds tend to sing, small birds tend to be brown, brown birds tend to sing, flying birds tend to sing, etc. However, many of these correlations are clearly false. Thus, representing correlations among properties requires something more than the mere listing of properties.
5. For instance, recent evidence from developmental psychology has shown that children's concepts turn out to be rather complex. Many of their concepts contain information about ontology, functions, causation, intentions, and ‘hidden’ and ‘non-obvious’ properties (such as an organism's innards or an object's ‘essence’), to which children appeal when they categorize (Carey Citation1985; Gelman and Koenig Citation2003; Gelman and Wellman Citation1998; Keil Citation1989; Medin and Ortony Citation1989).
6. Murphy and Medin explicitly state that they are not proposing a new model of conceptual representation (Citation1985, 426). Our purpose here is merely to summarize what they mean by theory in order to arrive at a manageable construal of the ‘theory-theory’.
7. Another contrast that can be drawn here is that Murphy and Medin are primarily interested in adult concepts, whereas Gopnik, Meltzoff, and Wellman are mainly concerned with concepts held by children. This contrast, however, is of minor importance in the context of our current discussion because our central purpose is merely to come up with possible ways in which to construe the ‘theory-theory’.
8. In what follows, I will assume that grandmother's shrieks serve as the only way in which James identifies mold. Of course, there will be occasions in which her shrieks will serve as additional evidence for identifying mold and still others, in which they do not come into play. Notice that it does not matter much to the argument just which role grandmother's shrieks play. So long as James' belief about grandmother's shrieks can be used on occasion to categorize mold, it would have to be included as part of the conceptual structure. After all, critics of prototype theories have insisted that correlations be included as part of the concept even though such relations only sometimes play a role in categorization. Moreover, features listed under a prototype do not always factor in the performance of categorization tasks; that is why they are probabilistic.
9. It may be tempting to refer to the reliable relationship between mold and grandmother's shrieking as a ‘lawlike’ relationship. However, recall that Gopnik and Wellman intend such relationships to involve a theoretical construct or term. Since the relationship between mold and grandmother's shrieking only pertains to observable phenomena, it is best described as an empirical generalization. Also, it is possible that there could be a lawlike relationship which exists between hyphae and his grandmother's shrieks. However, as far as James knows, this is not something entailed by his theory of mold. As such, his belief must be considered outside of theory.
10. Some may object that my notion of categorization is not the same as that employed by cognitive psychologists. For instance, their notion pertains to spontaneous, quick, or ‘intuitive’ acts of subsuming things under a category, whereas mine concerns a much broader phenomenon, including categorization which results from deliberation or inferences. In response, I point out that there is no reason to think that James' categorization judgments involving his belief about his grandmother's shrieks cannot be of the spontaneous or the intuitive kind; in fact, James' belief may have been an implicit one all along, never enjoying a moment of conscious reflection. Moreover, it has become ‘second nature’ for him to associate grandmother's cries with mold. If this is true, then my argument against the ‘literal’ interpretation holds: he would still be appealing to something outside of theory to categorize (in the same sense of categorization employed by cognitive psychologists).
11. For example, Keil Citation(1989), Murphy and Medin Citation(1989), and Rips Citation(1989) have suggested formulations of theory that incorporate, among other things, empirical generalizations.
12. Incidentally, this avoids another objection: an empirical generalization could still be regarded as theoretical because the requirement for theory is not that the generalization itself has to satisfy the three criteria set forth by Gopnik and Wellman, but that the overall theory, of which the generalization is a part, satisfies them. Thus, the empirical generalization could be incorporated under theory solely on the basis of it being a law-like correlation. In my view, this objection fails because my reformulated thought experiment no longer relies on such a correlation.
13. Here is another way in which the thought experiment can be modified. Suppose that James has witnessed only one occasion in which grandmother shrieked at the presence of mold. However, it was such a memorable and vivid experience (e.g. he has never seen his grandmother shriek like that before) that there is a strong association in his mind between her shrieks and the presence of mold. So, the next time he hears her scream in a similar fashion, he cannot help but to think that she is in the presence of mold.
14. There are two points to note here. First, as with the original thought experiment, whether or not James is correct to arrive at this conclusion is not important; he is perfectly willing to acknowledge the fallibility of his belief. Second, that the belief itself is based on a probabilistic relationship between shrieks and mold should not disqualify it from being used for categorization. After all, prototype theories made use of features and properties that were probabilistic.
15. At this point, one could object that in such cases, James would not be categorizing mold; instead, he would merely be appealing to these beliefs to identify it. However, it is unclear whether the ‘literal’ view is committed to the idea that a concept must include all information that might be relevant for identifying its members. If it is not, then it could not be faulted for excluding information pertinent to identification. In response, I would like to point out that identification is (a type of) categorization: to identify x as y is simply to apply the concept y to x. It is unclear how a person can be said to identify something as mold if she had not already applied the concept mold to the entity appearing before her. If this is correct, then the ‘literal’ view would be committed to the idea that a concept has to include any information relevant for identifying its members, for it is already committed to the idea that a concept has to include information relevant for categorization. As we have already seen, this latter commitment was a primary motivation behind the adoption of the ‘theory-theory’.
16. For instance, this construal of the ‘literal’ view not only would encompass items (1)–(5) (see the discussion on the ‘liberal’ view's account of conceptual structure in the third paragraph below in the text), but also some ‘trivial’ information (provided that it is relevant for categorization purposes).
17. Just how much information conceptual structure will need depends, of course, on the theory itself (whether the ‘liberal’ or ‘literal’ interpretation).
18. For instance, Prinz argues that if concepts were construed as ‘long-term-memory networks’, then they could not be activated by our finite and limited working memory (2002, chaps 4, 6).
19. This is perhaps unsurprising given that the only way to ensure that all such information would be included in the theory is to subscribe to the ‘liberal’ view.
20. I have hitherto been assuming that the ‘theory-theory’ is a theory of conceptual structure which holds for all kinds of concepts. Thus, the argument presented here shows, at the very least, that there are certain concepts of which the ‘theory-theory’ cannot be true, and therefore, that the ‘theory-theory’ is untenable. Not all psychologists, however, make this assumption about the ‘theory-theory’. For instance, Gopnik and Nazzi hold that the ‘theory-theory’ ‘does not seem to be as naturally applicable to other types of knowledge, for example, purely spatial knowledge, syntactic or phonological knowledge, musical knowledge or mathematical knowledge’ (2003, 307). Instead, they (along with other cognitive psychologists) maintain that the ‘theory-theory’ holds at least for natural object concepts, concepts like fish, grass, skunk, etc. The important point to note here is that our critique of the ‘theory-theory’ can still be applied to such concepts.
21. It is worth emphasizing that it is not my requirement that concepts should embody all information that is relevant for categorization. As I have noted, this requirement was set forth by cognitive psychologists when they abandoned similarity-based theories in favor of the ‘theory-theory’. My point is that such a requirement would rule out the possibility of a theory that posits a minimal structure (see Note 22).
22. We may, of course, want to devise a criterion other than categorization for what to include under conceptual structure. However, adopting this path will essentially undermine the rationale for subscribing to the ‘theory-theory’.