Categorization of Prototypical Stimulus Classes by Pigeons

Previous findings have demonstrated the pigeon's ability to categorize sets of schematic faces defined by a polymorphous feature rule. We present here an experiment that was conducted to evaluate the influence of the category structure determined by the distribution of features across stimulus classes. The task was a middle-value discrimination, involving a band of positive stimuli surrounded by a band of negative stimuli. Two large sets of Brunswik faces were conceptualized as variations of a standard stimulus (prototype) and occupied highly overlapping regions in the feature space. The classes could be separated in terms of a distance-from-prototype rule. Only three of five animals were successful in classification training. Analyses of both the transfer performance and of the effects of feature values on the variance of responding revealed the classical prototype effect-that is, performance determined by the overall similarity to the central tendency of the classes. A theoretical evaluation of the data examines the possibility that pigeons discriminated between the classes by virtue of a prototype-distance rule, although exemplar models offer more parsimonious accounts.