Relational feature acquisition: The interaction of two learning systems

Studies of expertise have indicated that experts have different representations from novices. Moreover, part of expert knowledge consists of perceptual relations, which could not have been acquired entirely through formal instruction. It is possible that the development of expertise also depends on developmental learning mechanisms which are part of the cognitive architecture. In this study, a theoretical framework is presented for the interaction of two systems--representational and strategic--which covers both processing and developmental time scales. It is hypothesized that each type of learning increases the developmental capacity of the other, and that only one learning system at a time is active. These hypotheses are tested in a set of four experiments, and the results support both hypotheses. Implications for the course of cognitive development over a longer time frame are addressed, and avenues of research in cognitive development and performance which might benefit from the interactive systems approach are proposed.