Making probabilistic relational categories learnable

Abstract

Kittur, Hummel and Holyoak (2004) showed that people have great difficulty learning relation-based categories with a probabilistic (i.e., family resemblance) structure. In Experiment 1, we investigated interventions hypothesized to facilitate learning family-resemblance relational categories. Changing the description of the task from learning about categories to choosing the “winning” object in each stimulus had the greatest impact on subjects’ ability to learn probabilistic relation-based categories. Experiment 2 tested two hypotheses about how the “who’s winning” task works. The results are consistent with the hypothesis that the task invokes a “winning” schema that encourages learners to discover a higher-order relation that remains invariant over members of a category. Experiment 3 reinforced and further clarified the nature of this effect. Together, our findings suggest that people learn relational concepts by a process of intersection discovery akin to schema induction, and that any task that encourages people to discover a higher-order relation that remains invariant over members of a category will facilitate the learning of putatively probabilistic relational concepts.