Page 83 - Encyclopedia of Philosophy of Language

listeners, and these latter environmental categories are different in different speech communities, say Spanish and English. Moreover, although it is unclear whether children form relevant constitutional categories, their discrimination of stops varying in voice onset time is certainly sharpest around environmental category boundaries. So one may say that the environmental categories formed have an established constitutional basis, at least.
3. Concepts
It is essential to distinguish between the mental struc- ture which represents a category and the category itself. Within psychological discussions, these mental structures corresponding to and representing cat- egories have been called 'concepts.' (This usage is different from that found in most philosophical dis- cussions, following Fregean practice, in which con- cepts are taken to be abstract entities specifying the intension of a category.) Two important varieties of concept are, or ought to be, distinguished: 'individual- concepts' and 'type-concepts.' Concepts may rep- resent categories in different psychological functions. Perhaps the simplest such function is 'recognition.' Preliminary definitions would then be:
(a) An individual-concept is a mental structure that enables recognition of the same individual, encountered at different times and places;
(b) A type-concept is a mental structure enabling recognition of different individuals as being of the same type (i.e., belonging to the same category).
In the late twentieth century, most psychologists outside the behaviorist and some cognitive-science traditions insist on rather more by way of definition than what is offered above. The definitions given above are minimal in several senses. For instance, recognition can be based on a very partial specification of the recognized individual or type, provided indi- viduals and types are well-separated in the world in question. To take an example from Dennett (1987:290), in a particular country, a coin-operated device may distinguish the desired type of coin from others on the basis of a partial specification of weight and shape, ignoring, say, embossed or engraved marks and inscriptions. As an instantiated type-concept such a device is very defective, though it may work well enough, since the objects inserted form well-separated clusters. If the device were improved in conceivable ways, then the danger would arise of its rejecting per- fectly good coins because of surface imperfections, etc.,soitmaybeseenthatattainmentofafullyeffec- tive instantiated type-concept is a difficult goal. In fact, it is an impossible requirement. A fully effective type-concept for a given coin specifies a history for the coin—that it was minted in a particular place by certain machines. Exactly the same conclusion follows for individual-concepts: ideal individual-concepts will
distinguish 'indiscernible' individuals (pennies, twins) and will not be diverted by 'disguise' changes. Ideal individual-concepts will therefore require the speci- fication of a history as well. Though such ideal con- cepts perhaps cannot be attained, it is a common enough notion that concepts should not be ascribed to creatures or devices, unless they can pick out some- thing like the correct category in most circumstances. This requirement of additional functionality may be characterized as a demand that concepts should be 'computationally effective.'
A second notion of desired functionality, additional to that specified in definitions (a) and (b), is that con- cepts should be 'representationally effective': they should allow their possessors to hold the target indi- vidual/type in mind when it is absent or competing for attention with other categories. This notion of concept coincides more or less with Piaget's.
A third suggestion for additional functionality is that concepts should be susceptible to combination, so that novel properties, relations, and relational properties may be constructed from familiar concepts. There is no doubt that this sort of additional func- tionality is highly desirable: the creative and imagin- ative capacities of individuals depend on the possession of such 'productively effective' concepts.
These three different characterizations of the additional functionality required need not lead to three different theories of concepts. Instead, it may be reasonable to attempt a theory of concepts that sat- isfies all three requirements simultaneously. After all, the requirements answer to capacities that work to- gether developmentally.Individualsbecomecreatures that (a) are not easily fooled (concepts become com- putationally effective), (b) can think about remote objects, etc. (concepts become representationally effective), and (c) show some capacity for rep- resentational novelty (concepts become productively effective).
Only the nature and origins of representationally effective concepts are discussed here. However, as noted, it is hoped that attainment of such concepts is at least associated with the other two sorts of additional functionality.
The definitions (a) and (b) above clearly do not define concepts in any of the sensesjust described, but they are useful notions nonetheless. The structures defined there will be referred to instead as 'individual-' and 'type-detectors.'
3.1 Formation of Early Concepts
Traditionally,followingVygotsky(1962)andInhelder and Piaget (1964), it has been assumed that the free- sorting task, in which children form a large collection of diverse objects into groups that share a similar property, depends upon the ability to hold the shared property in mind across the several sorting operations and despite constant change in the other properties.
Language Acquisition: Categorization and Early Concepts
61