ARTIFACTUAL KINDS AND FUNCTIONAL DESIGN-FEATURES - WHAT A PRIMATE UNDERSTANDS WITHOUT LANGUAGE

Of several domains of knowledge, humans appear to be born with an inna tely structured representational system for making sense of objects, w hat properties individuate them, how they move in space, and what caus es them to move from one location to another. They also appear to make simple conceptual cuts between artifactual kinds and living kinds. Th e basis for this distinction seems to be a combination of crucial func tional properties, together with a teleological (i.e., historical/inte ntional) stance, one that asks 'What was this object designed for?'. A lthough non-human primates also appear to have considerable understand ing of objects, and often use objects as tools, it is not clear whethe r they draw a distinction between artifactual acid living kinds, and i f so, what factors guide this distinction. As a step in addressing thi s problem, I present experiments on a small New World monkey, the cott on-top tamarin (Saguinus oedipus oedipus), designed to reveal their un derstanding of the functional properties of tools using a procedure as sociated with minimal training. Specifically, the experiments explored whether tamarins distinguish between relevant and irrelevant properti es of a tool, and further, understand that some features can be transf ormed with little cost to functionality. The first experiment was a me ans-end task and involved using a cane-like object (a tool) to access a piece of food. In this experiment, there were always two choices: ei ther the food was immediately accessible because it was located on the inside of the cane's hook or less readily accessible because it was l ocated on the outside of the hook. Most of the tamarins reached criter ion on this task within a few sessions, consistently picking the cane with the most accessible food. Subsequent experiments (2-4) involved p roperty changes (i.e., its color, texture, size and shape) that had ei ther significant or relatively insignificant effects on the tool's fun ction. In general, the tamarins appeared tolerant of all property tran sformations as evidenced by the fact that they selected each object at least once. However, clear preferences also emerged suggesting that s ome properties had a more significant impact on the tool's functionali ty. Thus, in head-to-head competitions, tools with color or texture ch anges were selectively preferred over tools with shape or size changes . This makes sense since color and texture do not effect the tool's fu nction, whereas shape and size do. The final experiments involved both novel and familiar objects that, based on their current configuration , couldreadily be used as tools, in contrast with objects that require d considerable manipulation to convert into a tool. Consistently, the tamarins preferred possible over convertible tools, and when two conve rtible tools were presented at the same time, they preferred the tool that required the fewest changes to the required motor response. Resul ts suggest that the tamarins distinguish between relevant and irreleva nt properties of a tool and this distinction is based on functionality , on having good design. This ability is especially surprising given t he fact that tamarins do not naturally use tools, and infrequently com e into contact with artifacts. Results are discussed in light of curre nt theories concerning the representational foundations of natural kin ds, and in particular, artifactual kinds. (C) 1997 Elsevier Science B. V.