Implicit and explicit processes in category-based induction: is induction best when we don't think?

ABSTRACT

In category-based induction (CBI), people use category information to predict unknown properties of exemplars. When an item's classification is uncertain, normative principles and Bayesian models suggest that predictions should integrate information across all possible categories. However, researchers previously have found that people often base their predictions on only a single category. In the present studies, we investigated the possible distinction between implicit and explicit processes in CBI. Predictions of an object's motion took the form of either a catching task (implicit) or a verbal answer (explicit). When subjects made predictions implicitly (Experiment 1), they used categories as Bayesian models predict. Explicit predictions (Experiment 2) showed clearly nonnormative use of categories. This distinction between implicit and explicit processes was replicated with a within-subjects design (Experiment 3). When subjects learned categories implicitly (categories were never mentioned) in Experiment 4, their explicit predictions did not reflect integration of information across categories but again showed a nonnormative pattern of category use. These results provide support for a distinction between implicit and explicit processes in CBI and furthermore suggest that the same category knowledge may result in normative or nonnormative responding, depending on the response mode.