Contributed by: Joel Parthemore, University of Sussex, UK

A theory of concepts is a philosophical attempt to formalize understanding of what concepts are and what conceptual mental content is. It attempts, by one means or another, to answer the deceptively simple question, “what is a concept?”, or “what do we mean when we talk about concepts?” It seeks to explain of how concepts are shared (publicity), how they can be re-used (systematicity) and re-combined with each other (compositionality), how a finite number of them can give rise to a potentially infinite number of propositions (productivity), and so on. Two of the currently hotly debated theories of concepts are the proxytypes theory of Jesse Prinz (2004) and the informational atomism theory of Jerry Fodor (1998).

An enactive theory of concepts attempts to give an account of concepts from within the enactive approach, a trend within contemporary cognitive science whose early form was given by Francisco Varela, Evan Thompson, and Eleanor Rosch (1991). Although enaction means many things to many people, a central feature of most if not all enactive approaches is an emphasis on the embodiment of cognitive agents and their embeddedness in a particular environment. Agent cannot be separated from environment nor cognition from lived experience.

An enactive theory of concepts, it shall be argued, has advantages over non-enactive accounts and in particular can be used, in the same spirit as Peter Gärdenfors' (2000) conceptual spaces theory, to bridge the apparent gap between symbolic and associationist accounts of concepts, as well as to show what is worth holding onto of classical definitionist and imagist accounts (which are both, and with good reason, out of fashion).

The classical definitionists simply took concepts to be definitions, not unlike the ones one might find in a dictionary. As language is paradigmatically symbolic, so the classical definitionist approach tended to be a symbolic approach. The imagists took concepts to be something like pictures in the mind. Pictures are paradigmatically non-symbolic, so the classical imagist approach tended to be, roughly, an associationist one: finding associations between one image and another.

What does an enactive perspective on concepts entail? By most accounts, an enactive perspective includes but goes beyond providing an account of how the concept-possessing-and-using agent is embedded in its environment and embodied in a particular physical form. It sees a first-person account of concepts (as of science) to be not just useful but a necessary part of an overall account. It takes an agent/environment, internal/external distinction to be both conceptually necessary and, at the same time, meaningful only with respect to an observer.

An enactive theory of concepts ultimately locates concepts not in the concept-possessing-and-using agent (say, as internal, a priori mental representations) nor in the agent's environment (say, as an external set of affordances). Concepts are enacted out of the dynamic interaction of the agent with the agent's environment. Nonetheless there will be occasions when it is appropriate for either agent or environment to be emphasized. When mental representations are being discussed, agent will be in the foreground and environment in the background; when affordances are being dicussed, environment will be in the foreground and agent in the background.

On the one hand, what makes mental content conceptual is that it abstracts away from the immediate perceptions of the moment, discovering regularities in the perceptual stream by finding associations between the perceptions of one moment and the perceptions of another. Paradigmatic conceptual content is content not of the moment. Paradigmatic non-conceptual mental content is content of the moment, without perceived connection to any other moment (either backward looking, through memory, or forward looking, through prediction). What makes concepts recognizably concepts is their stability. On the other hand, concepts, because of the way they are, on the enactive view, dynamically enacted by agent and environment, must logically be subject to continuous if often only incremental change, precisely because the regularities they relate to must likewise be subject to change.

An enactive theory of concepts may be seen to lead naturally to a series of these contrasting (but, arguably, mutually necessary) views on what concepts are, depending on what is being emphasized. So there is an important distinction to be made between when we, as concept-possessing-and-using agents, reflect on concepts as concepts – either introspectively or as we observe them (via folk psychology of mind) in others – and when we (as we presumably do much more often) simply employ concepts, non-reflectively. When we think of concepts as concepts, then it is useful and possibly necessary to apply symbolic and representational descriptions to them. When we think of concepts as concepts, then it is natural to associate them with words of a language, by its nature symbolic (and indeed, many philosophers of concepts, including Jerry Fodor, tie concepts very tightly to language, even to holding that non-language-using agents by definition cannot possess concepts), and to give an account of a particular concept in terms of a definition.

When we reflect on our concepts, they assume a representational or symbolic role. But symbols and representations, according to Harvey (1992), require someone for the symbols and representations to be symbols and representations to. So on the one hand, we cannot step outside of our role as observers, and when we give an account of concepts, we do so as observers of concepts; on the other hand, we require means to talk about contexts in which there logically is no observer (there is no homunculus!). When one employs a concept, no one need be watching one do it (and indeed no one, including oneself, directly can).

In line with this, as Andrew Woodfield (1994) has pointed out, there is a distinction to be made between concepts as private entities or as public entities (i.e., existing in the shared space of public discourse). Woodfield writes: “The thesis that a personal concept can change during its owner's lifetime must be sharply separated from the claim that a public concept can change over time.” Without needing to commit oneself to Woodfield's views on conceptual change, one can see him as emphasizing a difference that many philosophers prefer to skate over. While implicitly acknowledging concepts to be both, they emphasize either the public or the private aspect. So when Fodor (1998) says that concepts must be precisely the same between agents in order to ensure publicity, he is tying them to language and emphasizing the public aspect. When Prinz (2004) says that concepts only need to be sufficiently similar and have the same core (prototype) features in order to ensure publicity, he is emphasizing the private aspect.

There are other distinctions to be made: e.g., between concepts as implicit knowledge how and explicit knowledge that, which is where some might prefer to draw the line between non-conceptual and conceptual mental content. Yet there seems no obvious reason why knowledge how cannot, to some significant extent, meet Gareth Evans' (1982) Generality Constraint on conceptual content: i.e., that it be redeployable systematically and productively. (That is, a finite set of concepts can be redeployed systematically across contexts, and they can be recombined productively into potentially infinitely many complex structures.) On the one hand, the examples Evans gives to illustrate the Generality Constraint reflect propositional thought. On the other, Evans is keen to emphasize concepts as abilities rather than objects – and abilities seem more in keeping with knowledge how than knowledge that. (Note that Alva Noë (2005) likewise is keen to emphasize concepts as abilities, specifically sensorimotor ones.) Prinz (2004) in offering his updated version of imagism argues that the idea of concepts as necessarily consciously introspectible entities needs to be abandoned. If some concepts are not available to introspection, then they cannot, presumably, form part of our explicit knowledge but only contribute to our general conceptual abilities.

All of these distinctions – between concepts as internal vs. external, concepts as abstracting from context vs. being defined by context, concepts as reflected upon vs. concepts as used non-reflectively, concepts as symbolic entities vs. concepts as associational entities, concepts as private vs. concepts as public, concepts as abilities vs. concepts as objects, and so on – bear a strong relation to each other. If an enactive theory of concepts can show how concepts are not, ultimately, internal or external, then it is reasonable to think it might also show how each of these other distinctions, while conceptually necessary, may not ultimately be logically maintainable. Gärdenfors wishes to locate his theory of concepts between a symbolic and an associationist account, in order to bridge the two. An enactive theory of concepts must, it would seem, do the same.

Bibliography
Evans, Gareth (1982). VARIETIES OF REFERENCE, Oxford University Press.

Fodor, Jerry (1998). CONCEPTS: WHERE COGNITIVE SCIENCE WENT WRONG. Oxford University Press.

Gärdenfors, Peter (2000). CONCEPTUAL SPACES: THE GEOMETRY OF THOUGHT. MIT Press.

Harvey, Inman (1992). “Untimed and misrepresented: connectionism and the computer metaphor”, Cognitive Science Research Paper 245, University of Sussex, UK.

Noë, Alva (2005). ACTION IN PERCEPTION. MIT Press.

Prinz, Jesse (2004). FURNISHING THE MIND: CONCEPTS AND THEIR PERCEPTUAL BASIS. MIT Press.

Varela, Francisco, Evan Thompson and Eleanor Rosch (1991). THE EMBODIED MIND: COGNITIVE SCIENCE AND HUMAN EXPERIENCE, MIT Press.

Woodfield, Andrew (1994). “Do Your Concepts Develop?” In Hookway, C., Peterson, D. eds. , PHILOSOPHY AND COGNITIVE SCIENCE, Cambridge University Press.