The impact of biased hypothesis generation on self-directed learning

Abstract

Self-directed learning confers a number of advantages relative to passive observation, including the ability to test hypotheses rather than learn from data generated by the environment. However, it remains unclear to what extent self-directed learning is constrained by basic cognitive processes and how those limits are related to the structure of the to-be-learned material. The present study examined how hypothesis generation affects the success of self-directed learning of categorical rules. Two experiments manipulated the hypothesis generation process and assessed its impact on the ability to learn 1D and 2D rules. Performance was strongly influenced by whether the stimulus representation facilitated the generation of hypotheses consistent with the target rule. Broadly speaking, the findings suggest that the opportunity to actively gather information is not enough to guarantee successful learning, and that the efficacy of self-directed learning closely depends on how hypothesis generation is shaped by the structure of the learning environment.


Back to Table of Contents