Behavioral and Neural Mechanisms of Spontaneous Generalization

Memory generalization is the process by which we extract commonalities across our individual experiences to form new knowledge that can guide future decisions. Studies examining generalization have traditionally employed tasks, like category learning, that emphasize learning categorical information via extraction of commonalities among stimuli. Generalization is then explicitly assessed via transfer of category knowledge to new examples. Separately, memory for individual experiences, or memory specificity, has been studied through episodic memory tasks that emphasize differences between stimuli. However, real-world experience rarely puts us in situations where learning goals prioritize specificity or generalization at the expense of the other. Rather, circumstances often require us to extract the commonalities across our experiences while also maintaining memory for the specific details. Thus, the goal of the dissertation was to evaluate the behavioral and neural mechanisms that support spontaneous memory generalization during learning that emphasizes memory specificity. Using a novel, paired associates learning task where blended faces were paired with full-name labels, we provided an opportunity for participants to form category knowledge based on shared surname labels. Unlike traditional category learning tasks, learning goals in the current task explicitly required participants to differentiate all faces, even those with shared family membership. Across 3 studies, using behavioral measures of perceived similarity and neural pattern analyses during encoding, we found that the mere presence of a shared label produced behavioral and neural evidence for category-biased representations during learning. Notably, neural evidence for category-biased representations extended beyond hypothesized memory generalization regions to include widespread aspects of the brain including higher-order visual cortex. Further, we found evidence that the hippocampus may support generalization and specificity simultaneously via differential connections with other hypothesized memory generalization and specificity regions. Together, our results inform our understanding of current theories of memory generalization by demonstrating conditions under which memory generalization proceeds spontaneously during learning