The effects of transfer context on the stability of two types of conceptual structure in a function concept

The stability of rule- or exemplar-type conceptual structures was evaluated through the effects of experience, across multiple contexts, on the extrapolation of a functional relationship. The stability of extrapolation performance was assessed in Experiment 1 across three transfer tasks using experience with novel stimuli, in Experiment 2 where similarity to training stimuli was reduced for a subset of novel stimuli, and in Experiment 3 where a subset of novel Individuals were first classified to type (i.e., rule or exemplar) based on one test of extrapolation using a performance criterion. Then the stability of extrapolation behavior was appraised along the dimensions of performance characteristics and empirical stability, both between and within the two types. Performance characteristics of each type were defined as extrapolation of the function relationship for rule use and limited extrapolation for exemplar use. Empirical stability was defined statistically by comparisons of performance within types and differences between types, after classification.; Across the three experiments, the rule type, with limited exceptions, extrapolated the function. The exemplar type exhibited a mixture of extrapolation performance. These results were evaluated in terms of the possible performance outcomes of: type stability across experience, instability across experience, or adaptability across contexts.