The attribute selection process in pattern perception: The effect of constraint redundancy and stimulus exposure time on the classification of spatially represented Markov patterns

Abstract

A transfer paradigm was used to study the effect of constraint redundancy (Rc) and stimulus exposure time (ET) on the "same-different" classifications of pairs of spatially represented Markov patterns (Vargus 7 stimuli; Evans & Mueller, 1966). In the training phase, each S classified, without feedback, pairs of Vargus 7 stimuli generated as deviations from three prototypes (i.e., most probable sequences). Each S received stimuli generated at one of three levels of Rc and exposed for one of three durations. In the test phase, all Ss classified different deviations from the same three prototypes at a single Rc level and a single ET level. Unsupervised classification performance was above chance in both training and transfer and increased slightly over trials. Significant Rc and ET effects were found. These results supported two models of attribute selection and utilization; discussions of these models and their relationships to the findings were directed toward the analysis of the "schema plus correction" hypothesis.