BEHAVIORAL AND BRAIN SCIENCES (1998) 21, 55±100 Printed in the United States of America A common structure for concepts of individuals, stuffs, and real kinds: More Mama, more milk, and more mouse Ruth Garrett Millikan Department of Philosophy, University of Connecticut, Storrs, CT 06250-2054 Electronic mail: millikan^uconnvm.uconn.edu Abstract: Concepts are highly theoretical entities. One cannot study them empirically without committing oneself to substantial preliminary assumptions. Among the competing theories of concepts and categorization developed by psychologists in the last thirty years, the implicit theoretical assumption that what falls under a concept is determined by description (ªdescriptionismº) has never been seriously challenged. I present a nondescriptionist theory of our most basic concepts, ªsubstances,º which include (1) stuffs (gold, milk), (2) real kinds (cat, chair), and (3) individuals (Mama, Bill Clinton, the Empire State Building). On the basis of something important that all three have in common, our earliest and most basic concepts of substances are identical in structure. The membership of the category ªcat,º like that of ªMama,º is a natural unit in nature, to which the concept ªcatº does something like pointing, and continues to point despite large changes in the properties the thinker represents the unit as having. For example, large changes can occur in the way a child identifies cats and the things it is willing to call ªcatº without affecting the extension of its word ªcat.º The difficulty is to cash in the metaphor of ªpointingº in this context. Having substance concepts need not depend on knowing words, but language interacts with substance concepts, completely transforming the conceptual repertoire. I will discuss how public language plays a crucial role in both the acquisition of substance concepts and their completed structure. Keywords: basic-level categories; categorization; child language; concepts; externalism; names; natural kinds; Putnam; theory of meaning. 1. Introduction Frank Keil observes mildly, ªit is difficult to design and entirely negative. Moreover, the tentative positive views motivate empirical studies on concept acquisition without offered have concerned not the nature of concepts (some- first committing oneself to a set of assumptions about what thing in the mind) but rather the extensions of words in a concepts are and how they are representedº (Keil 1989, public language. Putnam and Kripke left obscure the na- p. 25). Indeed so! Concepts, taken as items that the psyche ªacquires,º are highly theoretical entities. Clearly it is not possible to study them empirically without committing Ruth Garrett Millikan is Pro- oneself to substantial preliminary assumptions about their fessor of Philosophy at the University nature. of Connecticut. She returned to pro- fessional philosophy in 1983 after rais- One aim of this target article is to show how, throughout ing children. Since then she has pub- the changing variety of competing theories of concepts and lished Language, Thought, and Other categorization developed by psychologists in the last half Biological Categories (MIT Press), century, the theoretical assumption that the extensions of White Queen Psychology and Other concepts (the set of things that fall under the concept) Essays for Alice (MIT Press) and thirty five papers in the are determined by descriptions has managed to go un- philosophies of psychology, language, and biology. She challenged. This is true despite the fact that Putnam's has been a fellow at the Center for Advanced Study in (1975a) and Kripke's (1972) famous arguments against de- the Behavioral Sciences and President of the American scriptionism (or at least their conclusions) have been re- Society for Philosophy and Psychology. She delivered hearsed numerous times in the core psychological litera- the Gareth Evans Memorial Lecture for the University of Oxford in 1991 and the Romanell Lecture for the ture, and despite the fact that there have been a number of American Philosophical Association in 1997. Currently brave attempts to integrate these insights into the psycho- she is Visiting Research Professor at University of Stock- logical tradition (Gelman & Coley 1991; Keil 1989; Ko- holm and is preparing a book on Concepts for Cam- matsu 1992; Lakoff 1987; Markman 1989; Neisser 1987, bridge University Press. Ch. 2). The difficulty is that these insights were almost Q 1998 Cambridge University Press 0140-525X/98 $12.50 55 Millikan: A common structure for concepts of individuals, stuffs, and real kinds ture of the psychological states or processes that would the acquisition of substance concepts and in their com- constitute an understanding of the meanings of the words pleted structure. they discussed, thus offering no aid to psychologists. I will I will begin with a positive statement of what I take try to help remedy that situation. substances and substance concepts to be (sects. 2 and 3). I will present a nondescriptionist theory of the nature of From this nondescriptionist vantage it will be easier to see concepts of what (following Aristotle's Categories) I will call just how descriptionism continues to be an ingredient in ªsubstances.º The category of substances includes (1) contemporary experimental work on concepts (sect. 4). things we would ordinarily call ªsubstances,º namely, stuffs Then I will discuss the nature of concept development from such as gold, milk, and mud, (2) things designated ªprimary a nondescriptionist perspective (sect. 5) and finally the substancesº by Aristotle, namely, individuals such as Bill crucial involvement of language in the acquisition and use Clinton, Mama, and the Empire State Building, along with of substance concepts (sect. 6). (3) things designated ªsecondary substancesº by Aristotle, namely, real (as opposed to nominal) kinds. Real kinds include, paradigmatically, both ªnatural kindsº and the 2. Substances correspondents of what Eleanor Rosch called ªbasic levelº categories (Rosch 1975) ± those intermediate level catego- The bulk of a child's earliest words are concrete nouns, ries such as shoe and mouse and house that children in all including names of individuals, names of basic-level kinds, cultures learn first (Angelin 1977; Mervis & Crisafi 1982; and some names for stuffs (milk, juice). These are acquired Nelson 1974b). My claim will be that these apparently quite in a rush by the dozens between about one-and-a-half and different types of concepts have an identical root structure, two years of age: ªthis vocabulary spurt is often called the and that this is possible because the various kinds of naming explosion to reflect the large preponderance of ªsubstancesº I have listed have an identical ontological nouns that are learnedº (Markman 1991, p. 81; see Gentner structure when considered at a suitably abstract level. That 1982 and Ingram 1989 for reviews, Dromi 1987 for some is, surprisingly to us moderns, the Aristotelian term ªsub- reservations.)1 Adjectives come later and more slowly and stanceº is univocal (having one meaning only). Unlike the abstract nouns later still. This suggests that the ability to Aristotelian tradition, in modern times concepts of stuffs distinguish concrete individuals in thought and the ability and real kinds have traditionally been treated as predicate to distinguish basic kinds and stuffs may have something in concepts. That is, to call a thing ªgoldº or ªmouseº has been common, and that concepts of properties and of other taken to involve saying or thinking that it bears a certain abstract objects may not be required for these tasks. There description. One understands it as being gold or a mouse by is much independent evidence that children come to appre- representing it as having a certain set or appropriate sample ciate separable dimensions, such as color, shape, and size of properties, or certain relations to other things, or a only after a considerable period in which ªholistic sim- certain kind of inner nature or structure, or a certain origin ilaritiesº dominate their attention (see Keil 1989 for discus- or cause. I will argue that, on the contrary, the earliest and sion). Thus concepts of properties again appear as less most basic concepts we have of gold and mouse and so forth fundamental than those expressed with simple concrete are subject concepts. Their structure is exactly the same as nouns. I propose that individuals, basic-level kinds, and for concepts of individuals like Mama and Bill Clinton. stuffs have something in common that makes them all To call a person ªMamaº is not to attribute to her any knowable in a similar way, and prior to properties. properties, relations, or inner or outer causes. It is not to We can begin with kinds. In recent years, a number of classify her but to identify her. Similarly, Putnam argued, to researchers have been interested in the structure of con- call a thing ªgoldº or ªmouseº is not to describe it. Neither cepts of ªnatural kindsº and in the development of chil- concept consists of a representation of properties. Rather, dren's understanding of these kinds (e.g., Carey 1985; the extensions of ªgoldº and ªmouse,º like the extension of Gelman & Coley 1991; Keil 1989; Markman 1989). Natural ªMama,º are natural units in nature, units to which the kinds are said to be distinguished in part by the fact that concepts gold and mouse do something like ªpointing,º and many true generalizations can be made about them. Con- to which they can continue to point despite large changes in cepts of natural kinds thus provide an indispensable key to the properties the thinker represents them as having. For the acquisition of inductive knowledge. According to Gel- example, large changes can occur in the way a child identi- man and Coley (1991), people develop natural kind con- fies gold, hence in the things the child is willing to call cepts ªgold,º without affecting the extension of the child's word with the implicit .