Every new experience in our life takes place within the context of familiar environments and situations. However, most research on memory has focused on the artificial memorization of word lists, symbols or pictures; these studies do not meaningfully address how structured prior knowledge about the world (e.g., in the form of a familiar spatial map, or knowledge of how restaurant meals unfold over time) can scaffold new learning. In the proposed studies, I aim to precisely characterize how and where prior knowledge and new information are represented, how they get linked at encoding, and how they interact at recall to allow memories to be retrieved. In the first proposed study of my F99 phase, I test the hypothesis that hippocampal engagement at event boundaries during learning binds new information (i.e. objects) to the scaffold of existing knowledge (i.e. knowledge of a familiar location), and that hippocampal activation during recall mediates the successful retrieval of the bound object from the location in which it was stored. I also test the hypothesis that distinctive representations of spatial locations in the brain will reduce interference between objects stored in those locations. There is a potential downside to using prior knowledge as a scaffold: When there is too much information attached to one part of the scaffold, old and new memories will interfere with each other. How, then, could someone prioritize the retrieval of new memories over older (now-irrelevant) memories that were linked to the scaffold? Recent research on intentional forgetting suggests a solution to this limitation. Specifically, in my second proposed study, I test the hypothesis (supported by neurophysiological evidence, prior neuroimaging results, and computational models) that previously encoded memories can be weakened by moderately activating their neural representation, thereby ?cleaning? the scaffold and reducing interference. In the K00 phase, I will extend my research to identify pathologies in how clinical populations use prior knowledge to interpret and remember their experiences, using tools from computational psychiatry; I also plan to design new technological tools to address these issues. Overall, the proposed project makes use of naturalistic and ecologically valid stimuli (in the form of continuous stimuli and immersive virtual reality) paired with advanced machine learning tools applied to brain imaging data, to study the fundamental nature of how new and old information are linked to allow learning. In the long-term, the findings from this project regarding how prior knowledge can be optimally leveraged to support new learning will lead to the development of tools to help memory-impaired individuals make better use of prior knowledge to support new learning, as well as remedies for groups where deficiencies in prior knowledge prevent them from learning properly.
In my dissertation, I will use newly-developed machine learning techniques to study how the brain uses prior knowledge (about the spatial structure of an environment, or how certain types of events unfold in time) to scaffold new learning. By precisely characterizing how this scaffolding process works, my research will help to identify ways in which prior knowledge can be more optimally leveraged to support learning. This will lead to the development of tools to help memory-impaired individuals make better use of intact prior knowledge to support new learning, as well as remedies for groups where deficiencies in prior knowledge prevent them from learning properly.
