Eyetracking as an Implicit Measure of Category-Based Induction

Category information is used to predict unknown properties of category members. Previous research has found that when categorization is uncertain, property predictions do not reflect integration of information across categories as normative principles and Bayesian models would suggest. Rather, people often base their predictions on only the most likely category and disregard information from less likely ones. Research in category-based induction tends to elicit explicit, verbal responses which may not readily allow for integration of information across categories. This paper explores whether changing response mode can promote more normative use of category information in induction. Experiment 1 used an implicit measure of prediction: eye movements. The results suggest that when making predictions implicitly people integrate information across categories. The results of Experiment 2 suggest that the integration of information found in Experiment 1 were not a result of explicit strategies.