CONCEPTS

Concepts are customarily regarded as intermediaries between mind and world. They are the basic elements of thoughts and the tools by which one classifies things. Concepts are central to the philosophy of mind, and they are often implicated in theories of meaning. There are also some who think that philosophical method is largely a matter of conceptual analysis. There is considerable consensus on the importance of concepts, and, to a lesser extent, on the roles that concepts play, but beyond that there is rampant disagreement. For example, philosophers disagree about the ontology of concepts, the acquisition of concepts, and the content of concepts. In the twentieth century, psychologists began to weigh into these debates, and since the 1970's, much theorizing about concepts has been informed by interdisciplinary dialogue. This entry surveys dominant theories.

What Are Concepts For?

Within philosophy concepts are most often defined as the elements or ingredients of thoughts. Concepts are for thinking. When one ascribes a thought, such as "aardvarks are nocturnal," one typically assumes that the bearer of that thought has a concept of aardvarks and a concept corresponding to the property of being nocturnal. It is sometimes said that a concept is to a thought as a word is to a sentence, but this formula can mislead, because some philosophers do not believe that thought is language-like. However, even those who resist the view that thought is language-like may be attracted to one crucial point of analogy: concepts are believed to be combinable. Those who possess a concept of aardvarks can form the thought that aardvarks are nocturnal, the thought that aardvarks are quadrupeds, or the thought that aardvarks are insectivores, assuming possession of these other concepts. Gareth Evans (1982) suggests that it is a condition on concept possession ("the generality constraint") that, if a person can have the thought that a is F, then that person should also be able to form every other thought of the form a is X, where X ranges over the concepts in that person's conceptual repertoire. Some philosophers think there may be restrictions (e.g., of intelligibility) on combination, but most agree that thought formation through conceptual combination is a central function of concepts.

A second function of concepts is categorization. Many philosophers think that concepts are the primary tools by which one determines that something falls into a category. One knows that two things are both turtles in virtue of having a turtle concept. Historically, some philosophers have reserved the word concept and closely related words for general kinds. On this usage there can be a concept of turtles, in general, but not a concept of a particular turtle, say Yertle. Other philosophers tend to say that there can be concepts of individuals and that concepts can be singular as well as general. When one identifies an individual, one can think of that as an act of categorization, broadly construed: One categorizes that individual as belonging to a class with one member. Concepts are implicated in the categorization of kinds and individuals.

Concepts are sometimes said to have a function in inference. This third function often works in concert with the second. One uses concepts to draw inferences about the things that one categorizes. If one encounters a shovel, one can infer that it is used for digging. The knowledge that shovels are used for digging is said, by many, to be contained in one's concept of shovels. Thus, when one applies the concept to some thing, one can use the concept to infer facts about that object.

Concepts are also widely presumed to play a role in linguistic meaning. For some, concepts simply are the meanings of words or components of meanings. On this view concepts are expressed when one uses words. Some philosophers' (especially those who favor reference-based semantic theories) concepts are not meanings. But these authors usually concede that concepts play a central role in the epistemology of language. One comes to understand a word by associating it with a concept. On either approach concepts and language will be closely related.

A fifth function of concepts is related to the other three, but is potentially dissociable. Concepts are said to be representations; they refer to things. Some theories of concepts encompass theories of reference. In this sense, concepts are intermediaries between mind and world.

There is controversy about what concepts are for, but the items on the preceding list are widely accepted. Concepts are usually postulated to play all or some of the preceding roles.

Some Issues of Controversy

In describing some of the functions of concepts, a few places of controversy have already been indicated. There are a number of other controversies that deserve special mention.

One issue concerns ontology. It is widely agreed that concepts are intermediaries between mind and world, but where do they reside? One possibility is that they are timeless abstracta. This view has been especially popular among those who identify concepts with word meanings. Many semantic theorists believe that meaning enjoy some autonomy from psychology. On this view the meaning of a word does not depend on the images or ideas any individual happens to possess. Others are attracted to this view because they regard concepts as a specification of the essential properties of the things to which they refer.

The concept of a triangle, on this view, might be a geometric definition. Triangles had that definition before anyone discovered it. In contrast, there are philosophers who locate concepts inside the head. On this approach a concept is a mental representation, which plays a causal role in information processing. Others regard concepts as psychological, but eschew talk of mental representations. For example, one might say that a concept is a skill or ability or an operation on mental representations. Immanuel Kant (1997) says concepts are rules for constructing or organizing images. In between those who say that concepts are abstract and those who say they are psychological, there are social theories of concepts, according to which concepts supervene on human practices. What matters is not the contents of any individual's mind, but socially distributed patterns of deference, normative demands, and reason-giving behavior. There is room for uniting all these ontological perspectives into a single theory. For example, one could say that individuals have mental representations (psychological concepts) of community-enforced rules (social concepts) that dictate which timeless, essential properties their thoughts denote (abstract concepts).

Among those who think that concepts are mental representations, there are significant disagreements about representational format. Some think concepts are words in language-like mental code ("the language of thought hypothesis"), others claim they are mental images ("imagism"), and still others say they are weighted connections or patterns of activation in neural networks ("connectionism").

Those who think that concepts are mental representations also disagree about how concepts are attained. Some think that many concepts are innate, and some think that few or none are innate. There are controversies about how learned concepts are acquired. Concepts might be copied from experience, they might be abstracted, they might be learned by strengthening associations, or they might be acquired using a more deliberative procedure, such as the formation and testing of hypotheses. The innateness question is sometimes posed as a question of which concepts are primitive. Many philosophers believe that some concepts are primitive and others are assemblies or inferential networks built up from these. (When two concepts are combined to form a third, they are said to be "features" of that third concept.) Primitive concepts are often thought to be innate, so debates about this issue can sometimes be characterized as debates about how many primitives one has. Historically, however, some philosophers have assumed that many complex concepts are innate as well (such as the concept of God or of identity).

Those who think that concepts are abstract or otherwise external to individual minds sometimes talk about concepts using a definite article, "The concept of X." Those who think that concepts are mental representations are less likely to talk this way, leaving open room that different people may have different concepts of the same thing. There may be exceptions to this rule. It is natural to speak of technical concepts with a definite article ("The concept of natural selection") because there is sometimes just one correct formulation. In addition, some philosophers think concepts are individuated by their referents. On this view any concepts of the same thing will count as being the same concept. Hence, it would always make sense to talk about concepts using the definite article.

Another controversy surrounds the relationship between concepts and language. Besides the question of whether concepts are meanings (hence, whether language depends on concepts), there is a question of whether concepts depend on language. This conclusion has been defended by Ludwig Josef Johann Wittgenstein, Michael Anthony Eardley Dummett, and Donald Davidson. The arguments often turn on the claim that having concepts requires recognizing that thoughts and inferences can be mistaken, which depends in turn on belonging to a language community whose members give and demand reasons for utterances. In contrast, many think that concepts can be possessed without language, and, indeed, Jerry A. Fodor (1975) argues that language learning would be impossible without prior possession of concepts.

All these controversies are significant, but the main issue dividing competing theories of concepts has to do with content. Philosophers disagree about what information one knows in virtue of possessing concepts. One knows a great deal about many categories, but many philosophers believe that only some of this knowledge is conceptually constitutive. Some of this knowledge belongs to one's concepts, and the rest merely belongs to one's conceptions, where conceptions are thought to be more ephemeral and idiosyncratic than concepts. Theories of concepts can be distinguished by where they draw the concept-conception divide.

The Classical Theory

One theory of concepts has been so dominant in the history of philosophy that it has been dubbed "the classical theory." The name is apt, because the theory is championed by Plato. In classical theory, concepts are definitions: They specify conditions that are individually necessary and jointly sufficient for the categories they designate. In his dialogues Plato tries to uncover definitions of concepts such as justice, knowledge, piety, and love. On this approach specifying a concept of justice is a matter of specifying what it is to be just. It is unclear whether Plato thinks concepts are abstract entities or mental entities. He claims that people categorize things by recalling a life in a world of ideal forms, which they inhabited before life in the terrestrial world. Possessing a concept is a matter of intuiting, through memory and reflection, the essence of these ideal forms.

Many philosophers have assumed that some version of the classical theory is correct. Kant (1997) says that concepts are rules that determine the conditions of category membership. He also suggests that many concepts contain other concepts, like houses made from bricks, and in the Prolegomena to Any Future Metaphysics That Will Be Able to Come Forward as Science, with Selections from the Critique of Pure Reason (1997) he introduces the term analytic to refer to judgments whose predicate concepts are contained in their subject concepts. These judgments are, in effect, true by definition, as opposed to synthetic judgments, which are not true by definition, but must be discovered.

Gottlob Frege (1960) uses the term concept (Begriff ) to refer, narrowly, to the concepts expressed by predicates, but he uses a more encompassing term sense (Sinn ) to refer to the components of thoughts, and each of these, he suggests, can be identified with a descriptive content that determines reference. Frege insists that senses are abstract entities; if they were in the head, he thought they could not serve as the shared meanings of words. Inspired by Frege, Christopher Peacocke (1992) claims that concept possession involves the mastery of inferences, which play a central role in determining reference.

Rudolf Carnap (1956) claims that the concepts used in ordinary thought and talk are riddled with imprecision and that they need to be replaced by concepts that are explicitly defined. Analytic truths are stipulated, and hence immune from empirical refutation.

Defenders of the classical theory disagree about how concepts are attained. Plato obviously thinks concepts are innate, and Carnap thinks explicated concepts must be learned. For many classical theorists, some are innate and others are learned.

The classical theory has been criticized in various ways. Willard Van Orman Quine (1981) argues that the distinction between analytic and synthetic truths is unprincipled, because any putatively analytic claim could be revised under empirical pressure, if, for example, such a revision would be the most conservative way to alter a prevailing theory to accommodate new evidence. Hilary Putnam (1975) argues that definitions are not essential for reference; one can think about natural kinds (e.g., tigers, gold, and water) even if no one grasps the conditions that are necessary and sufficient for falling in the categories. Ludwig Wittgenstein (1953) argues that concepts often lack catchall definitions; instead, concepts group things together on the basis of family resemblances (games are his famous example). Psychologists support Putnam and Wittgenstein by showing that people rarely know the defining features of a category. Georges Rey (1983) counters that the psychological objections presuppose that concepts are in the head and readily available to consciousness—some classical theorists are willing to deny both assumptions.

Concept Empiricism

Plato does not say much about how concepts are mentally represented. Aristotle has more to say. He says that every concept is accompanied by an image. This idea inspired subsequent empiricist philosophers to propose that concepts are perceptual in nature. This basic claim is the essence of concept empiricism. Scholastic philosophers say that nothing is in the intellect that is not first in the senses. The British empiricists, such as John Locke (1979) and David Hume (1978), say that concepts are derived from percepts. Hume says concepts are simply copies of percepts or combinations of copied percepts, and Locke proposes that many concepts are acquired through abstraction from percepts (though there is some disagreement about what he and his scholastic predecessors meant by abstraction).

Concept empiricists differ in several ways from typical classical theorists. First, many concept empiricists are imagists, whereas many classical theorists are not. Concept empiricists usually say that concepts are mental representations (the British empiricists use the term ideas ), whereas classical theorists often say they are abstracta. Concept empiricists emphasize learning, whereas traditional classical theorists assume that many concepts are innate. Concept empiricists claim that concepts refer either by resemblance or by causal relations to their referents, whereas classical theorists usually assume that concepts refer by satisfying lists of defining conditions.

Nevertheless, there are theories that straddle the border between the classical theory and concept empiricism. The verificationist theories of concepts advanced by Carnap (1956) and other logical positivists are a case in point. For a verificationist, concepts consist in conditions that are necessary and sufficient for reference, but these conditions are specified in observational vocabulary; a concept refers to that which satisfies perceivable conditions of verification.

Concept empiricism is widely believed to face serious objections. One has to do with the concepts of abstract categories. There seem to be concepts of virtue, truth, substance, cause, and being, yet none of these things has any characteristic appearances. If concepts were all derived from perception, it would be hard to explain how people think about these things. Concept empiricists reply by either arguing that people do not have concepts of these things, or by reducing these concepts to perceptual features. Both strategies are hard to pull off.

Another objection is put forward by Kant (1997). He argues that one's capacity to perceive presupposes the possession of certain concepts (including concepts of time and space), which could not be derived from experience. Contemporary psychologists also argue that there is empirical evidence for innate concepts, which are evidently in place before experience, such as the concept of physical object or of number.

Prototype Theory

When the classical theory came under attack in the middle of the twentieth century, new alternatives were sought. One alternative, already mentioned, was Wittgenstein's (1953) family resemblance account, according to which one comprehends categories by means overlapping features rather than a catchall definition. This suggestion spawned the emergence of the cluster theory, which identified concepts with features that are not individually necessary for category membership but sufficient when a sufficient number are in place. No one feature may suffice for being a game, and no one feature is necessary, but bring a few features together and one has a game. In effect, the cluster theory is a similarity theory of concepts; it says that one categorizes by looking for similarities with familiar instances.

In psychology, dissatisfaction with the classical theory and inspiration from Wittgenstein (1953) gave rise to the prototype theory. On this approach categorization is also a matter of assessing similarity to a set of features that are not individually necessary for category membership. Prototype theorists do not construe concepts as unwieldy clusters, but as summary representations capturing just those features that are most typical of the category. A prototype is a representation of features that are highly frequent, salient, and diagnostic for category membership. The prototype for the category bird might include features such as flies, has feathers, has a beak, and sings. Following Putnam (1975), philosophers sometimes use the term stereotypes for much the same thing. Psychologists, notably Eleanor Rosch and Carolyn Mervis (1975), support the postulation of prototypes by showing that people categorize prototypical category members faster, learn to recognize them earlier, and list prototypical features first when asked to describe a category.

Prototype theorists usually assume that concepts are mental representations, but they diverge on the format of these mental representations. Some say they are made up of images, some say they are patterns in connectionist networks, and some say they are lists of features coded in a language of thought. Like some concept empiricists, prototype theorists argue that concepts are often learned by abstracting from particular category instances, but prototype theorists do not always assume that concepts are grounded in perceptual experience. There can be prototypes for categories that are difficult to discern perceptually, such as a prototypical analytic philosophy paper, a prototypical democracy, or a prototypical lie.

Prototypes are often used in categorization, but some psychologists and philosophers argue that they should not be equated with concepts. One objection is that similarity to a prototype is not necessary for categorization and reference; a shaved, mute, tailless, three-legged dog is completely unlike the dog prototype but still falls under the category. Similarity to a prototype is also not sufficient for categorization and reference: a duck decoy is no duck. Another objection is that prototypes do not combine together compositionally: the prototype for a compound concept is often unlike the prototype for its parts. Pet fish prototypically live in bowls, but neither pets nor fish prototypically live in bowls. Fodor (1998) argues that concepts must combine compositionally to explain that one can generate an unbounded number of novel thoughts from a finite stock of concepts. For similar reasons, prototypes may violate Evans's Generality Constraint (1982), which implies that concepts can be freely recombined; someone might know the prototypes for red fruit and long hair without knowing the prototypes for red hair and long fruit.

The Theory Theory and Holism

Unconvinced by prototype theory, some psychologists developed an alternative, which is associated with the following basic tenets. First, not all concepts are alike; one must distinguish animal concepts, artifact concepts, psychological concepts, mathematical concepts, concepts of physical objects, and so on. Each of these classes is governed by different "folk theories" that comprise small collections of basic principle; for example, folk biology explains that animals have hidden genetic essences, and folk physics explains that solid objects cannot pass through each other. Second, some folk theories lead one to postulate defining essences (as in the case of folk biology), but, unlike classical theorists, psychologists do not assume that these essences are known to those who postulate them; this is called psychological essentialism. Third, each of the concepts within one of these classes may contain causal and explanatory features besides prototypical features; for example, a concept of birds may contain the belief that wings enable flight. Together, these tenets suggest that concepts are like scientific theories: they divide into domains, they postulate hidden features, and they play a role in explanation. The approach has been dubbed the theory theory.

Most theory theorists assume that some rudimentary folk theories are innate, but they disagree about which ones. They also disagree about whether one's innate theories remain intact over development, or whether they undergo significant transformations, akin to conceptual revolutions in science. On the latter view adult concepts may be incommensurable with the concepts of children and infants.

The theory theory has been primarily developed by psychologists, but related ideas can be found in philosophy. Quine's (1981) critique of the analytic-synthetic distinction has led some to believe that the basic units for understanding any given category is an entire theory. Quine differs from most psychologists in three respects: He does not claim that theories are insulated from each other (perturbations in one may have ripple effects); he does not claim that theories are mentally represented in the head (Quine is a behaviorist); and he assumes that theories are learned (one's initial sorting behavior is driven by superficial similarities). Still, one might appropriate Quine's ideas into a psychological theory by proposing that each concept is a mental representation individuated by its place in a complete network of mental representations. This would be a holistic theory of concepts.

Critics of the theory theory and holism worry that these approaches entail that concepts are rarely shared. If two people have different theories, then they have different concepts, and their ability to communicate and to obey the same psychological laws becomes difficult to explain. It is also unclear whether these approaches can explain how concepts are combined to form thoughts, because theories are too cumbersome to easily combine together.

Informational Atomism

The theory theory and holism pack a lot of information into concepts. Some philosophers prefer the opposite strategy. Fodor (1998) argues that just about every lexical concept (a concept expressed by a single word) is primitive: a primitive concept is one that is not individuated by its relation to any other. This is called atomism. Instead, concepts are individuated by their referents, and concepts refer by falling under the nomic control of properties; roughly, a cow concept refers to cows because it is a law that encounters with cows and causes cow concepts to be tokened. This is called informational semantics.

Informational atomism is unlike all the theories considered so far, because all the others assume that most lexical concepts are complex. A primary advantage of informational atomism is that it can explain how concepts are recombined compositionally. If concepts are primitive symbols, then they can retain their identity when combined, just as words retain their shape when placed into sentences. Concepts can also be easily shared on this view: Two people have the same concept if they have symbols that are under nomic control of the same properties regardless of any difference in their beliefs.

These advantages come at cost. If lexical concepts are primitive, then they cannot be used to explain the inferences one draws or the way one categorizes. For a thoroughgoing atomist, someone could possess a concept of bachelors without knowing that they are male or unmarried. Atomism has also been associated with radical concept nativism. Many philosophers assume that primitive concepts are innate and that complex concepts are learned; if all lexical concepts are primitive, then all are innate. Fodor (1981) used to embrace this consequence, and Fodor (1998) now argues that primitive concepts can be learned.

Philosophy as Conceptual Analysis

Beginning with Plato, one of the dominant methods for doing philosophy has been philosophical analysis. Practitioners begin with a specific concept and reflect on its content. In so doing, they hope to reveal not only how one thinks about the referent of that concept but also what the essence of that referent is. By reflecting on the concept of virtue, for example, one might reveal what it is to be virtuous. The viability of this method depends on which theory of concept is correct. If concepts are definitions, conceptual analysis can reveal the essence of things. But if concepts are merely assemblies of typical features, incomplete and revisable theories, or semantically primitive symbols, then conceptual analysis cannot reveal the essence of things. There is as yet no consensus on which theory of concepts is right, but at stake is the methodology of philosophy itself.

See also Content, Mental.

Bibliography

Carey, Susan. Conceptual Change in Childhood. Cambridge, MA: MIT Press, 1985.

Carnap, Rudolf. "Empiricism, Semantics, and Ontology." In Meaning and Necessity: A Study in Semantics and Modal Logic. 2nd ed. Chicago: University of Chicago Press, 1956.

Davidson, Donald. "Thought and Talk." In Mind and Language, edited by Samuel Guttenplan, 7–23. Oxford, U.K.: Clarendon Press, 1975.

Dummett, Michael. The Interpretation of Frege's Philosophy. Cambridge, MA: Harvard University Press, 1981.

Evans, Gareth. Varieties of Reference edited by John McDowell. New York: Oxford University Press, 1982.

Fodor, Jerry A. Concepts: Where Cognitive Science Went Wrong. New York: Oxford University Press, 1998.

Fodor, Jerry A. The Language of Thought. Cambridge, MA: Harvard University Press, 1975.

Fodor, Jerry A. "The Present Status of the Innateness Controversy." In Representations: Philosophical Essays on the Foundations of Cognitive Science. Cambridge, MA: MIT Press, 1981.

Frege, Gottlob. "On Sense and Reference." In Translations from the Philosophical Writings of Gottlob Frege, edited by Peter Geach and Max Black.b Oxford, U.K.: Blackwell, 1960.

Hume, David. A Treatise of Human Nature, edited by Peter H. Nidditch. New York: Oxford University Press, 1978.

Kant, Immanuel. Prolegomena to Any Future Metaphysics That Will Be Able to Come Forward as Science, with Selections from the Critique of Pure Reason, edited and translated by Gary Hatfield. New York: Cambridge University Press, 1997.

Keil, Frank C. Concepts, Kinds, and Cognitive Development. Cambridge, MA: MIT Press, 1989.

Locke, John. An Essay concerning Human Understanding, edited by Peter H. Nidditch. New York: Oxford University Press, 1979.

Margolis, Eric, and Stephen Laurence, eds. Concepts: Core Readings. Cambridge, MA: MIT Press, 1999.

Medin, Douglas, and Anthony Ortony. "Psychological Essentialism." In Similarity and Analogical Reasoning, edited by Stella Vosniadou and Anthony Ortony. New York: Cambridge University Press, 1989.

Murphy, Gregory, and Douglas Medin. "The Role of Theories in Conceptual Coherence." Psychological Review 92 (1985): 289–316.

Peacocke, Christopher. A Study of Concepts. Cambridge, MA: MIT Press, 1992.

Prinz, Jesse J. Furnishing the Mind: Concepts and Their Perceptual Basis. Cambridge, MA: MIT Press, 2002.

Putnam, Hilary. "The Meaning of 'Meaning.'" In Language, Mind, and Knowledge, edited by Keith Gunderson. Minneapolis: University of Minnesota Press, 1975.

Quine, Willard Van Orman. "Two Dogmas of Empiricism." In From a Logical Point of View: Nine Logico-philosophical Essays. 2nd ed. Cambridge, MA: Harvard University Press, 1981.

Rey, Georges. "Concepts and Stereotypes." Cognition 15 (1983): 237–262.

Rosch, Eleanor, and Carolyn Mervis. "Family Resemblances: Studies in the Internal Structure of Categories." Cognitive Psychology 7 (1975): 573–605.

Smith, Edward E., and Douglas L. Medin. Categories and Concepts. Cambridge, MA: Harvard University Press, 1981.

Spelke, Elizabeth. "Initial Knowledge: Six Suggestions." Cognition 50 (1994): 431–445.

Wittgenstein, Ludwig. Philosophical Investigations. Translated by G. E. M. Anscombe. Oxford: Basil Blackwell, 1953.

Jesse Prinz (2005)