The main goal of the present research was todemonstrate the interaction between category and causalinduction in causal model learning. We used a two-phaselearning procedure in which learners were presented withlearning input referring to two interconnected causal rela-tions forming a causal chain (Experiment 1) or a common-cause model (Experiments 2a, b). One of the three events(i.e., the intermediate event of the chain, or the commoncause) was presented as a set of uncategorized exemplars.Although participants were not provided with any feedbackabout category labels, they tended to induce categories inthe first phase that maximized the predictability of theircauses or effects. In the second causal learning phase,participants had the choice between transferring the newlylearned categories from the first phase at the cost of sub-optimal predictions, or they could induce a new set ofoptimally predictive categories for the second causal rela-tion, but at the cost of proliferating different categoryschemes for the same set of events. It turned out that in allthree experiments learners tended to transfer the categoriesentailed by the first causal relation to the second causalrelation.
展开▼