Abstract
It has been suggested that category learning involves changes in attention allocation based on the relevance of input to the classification. Using eye-gaze measures, Rehder and Hoffman studied changes in attention allocation during category learning in a 5-4 category structure paradigm with four features of varying diagnosticity levels. In this paradigm, participants are tasked with classifying creatures into two groups through trial and error guided by feedback. While learners' eye-gaze patterns have been studied as a function of feature diagnosticity levels throughout the learning process, they have not been evaluated in relation to performance and feedback. The present study borrowed and modified Rehder and Hoffman's category paradigm and evaluated eye-gaze behavior as a function of the diagnosticity level of features, and the valence (positive vs. negative) of the preceding feedback during learning. Our results support Rehder and Hoffman's observations that gaze on the low diagnosticity feature decreased from the beginning to the end of the task. When change in eye gaze behavior was evaluated in relation to feedback, change in fixation probability was found to be greater following negative feedback. The results indicate that in a category task that includes performance feedback, learning strategies as indicated by changes in selective attention to features are affected to some degree by the valence of the feedback on a preceding trial.