Recognition of Facial Prototypes: The Importance of Categorical Structure and Degree of Learning

Abstract

The importance of categorical structure in the recognition and classification of realistic face prototypes, following a variable number of learning trials, was investigated in four experiments. Subjects either classified or observed well-defined (Experiments 1–3) or ill-defined (Experiment 4) faces belonging to three categories for one or nine trial blocks, followed by a transfer test. Recognition performance following classification (Experiment 1) and observational (Experiments 2 and 3) training was virtually the same: Oldness judgments were highest for the category prototypes and old instances and were least for new and random instances, with these tendencies increased following nine learning trials. When features were made ill defined (Experiment 4), oldness judgments decreased for the category prototype following nine learning trials. Multidimensional scaling (Experiment 5) revealed that the category structure for the well-defined and ill-defined faces was globally similar, and therefore, the disparate recognition of the category prototype was not due to a radical reorganization of the stimulus space. These results suggest that false recognition of the category prototype is at least partially based on type of category structure, and that prior studies using false recognition of the category prototype as evidence of an earlier abstraction process are probably misguided. Rather, false recognition is more likely due to erroneous combinations of features, with category influences playing less of a role.