Successful generalization of conceptual knowledge after training to remember specific events.

Introduction: Categorization involves grouping information to make inferences and support novel decisions. In the laboratory, category learning tasks commonly involve trial-and-error where participants are instructed to classify stimuli and learn through feedback. Here, we tested across two experiments whether people can acquire category knowledge in an incidental manner by associating category members with other information that itself is structured, and how it compares to acquiring category knowledge directly through feedback-based classification training. Methods: Subjects were trained to remember specific associations consisting of cartoon animals paired with animal-specific background scenes. Animals presented on forest vs. mountain scenes were members of two prototype-based categories, but this was not conveyed to the participants. Spontaneous category learning was tested by asking participants to guess habitat (mountains, forests) for old and new cartoon animals without feedback. Results: We found that participants spontaneously acquired category knowledge, showing high categorization accuracy for new animals, comparable to a group that underwent a traditional feedback-based classification training with the same stimuli. Strategy analysis showed that the majority of participants in both groups abstracted the central tendency of the categories, albeit a somewhat larger proportion of subjects relied on memory for specific training exemplars after paired-associate learning. Partial evidence was found for the hypothesis that generalized knowledge emerged at the expense of memory for specific animal-scene associations. Discussion: The findings show that despite the goal to remember specific information that required differentiation of stimuli within categories, subjects can spontaneously acquire category knowledge, generalizable to novel stimuli in a way comparable to traditional supervised classification training. This work provides new insights into how category learning can proceed under more naturalistic demands.