Dagmar Zeithamova, PhD - US grants

DESCRIPTION (provided by applicant): A powerful aspect of episodic memory-memory for individual events-is the ability to flexibly apply and combine information from past experiences to guide new behavior. Such flexibility to combine experiences in novel ways to infer new relationships is essential to behavior in ever-changing environment. The structures of the medial temporal lobe (MTL) are critical for the formation of new episodic memories that bind information within single episodes. However, how MTL supports the flexible combination of information gained across distinct episodes is not well understood. The research presented in this proposal will use functional magnetic resonance imaging (fMRI) to investigate the mechanisms that enable binding of information across distinct episodes, or cross-episode binding. We will further investigate how individual regions of the MTL support these mechanisms. Experiment 1 will test the hypothesis that cross-episode binding relies on reactivation of prior memories that share elements with the current experiences. A novel method will be used that allows inference about brain states, such as inferring a general content of a recalled memory, without an external behavioral response. We will examine whether encoding of new events that overlap with prior experience triggers reactivation of the prior experience, leading to a flexible use of memory. Experiments 2 and 3 will use high- resolution fMRI to investigate how individual MTL subregions support the processes critical for cross-episodic binding. Experiment 2 will test whether the CA3 field of the hippocampus supports reactivation within MTL cortex, as predicted by the role of CA3 in pattern completion-recollection of stored memories from a partial cue. The overlap between current and prior experience may serve as such a partial cue. The proposed research will provide the first direct evidence that MTL mediates reactivation of past events during overlapping experiences and that such reactivation supports cross-episode binding to promote the flexible use of memory. Experiment 3 will test how spatial contexts that are either the same or different across two experiences affect the likelihood of combining information across the two episodes and investigate how comparison between current and prior experience promotes cross-episode binding. A comparison between recalled memories and current events may serve to detect a change from a stored memory and trigger new encoding, leading to an integration of the new information into existing memories. This comparator function is thought to rely on the CA1 field of the hippocampus. This research will elucidate how MTL subregions support the abstraction of new knowledge from the relationship between distinct episodes and provide new evidence for the role of context in eliciting cross-episode binding. Knowing how MTL subregions combine information across events will provide a means to both predict and promote the occurrence of flexible learning and potentially enable the development of tools to enhance these processes in specific populations characterized by deficits in flexibility, such as individuals with schizophrenia, mild cognitive impairment, Alzheimer's disease and depression. PUBLIC HEALTH RELEVANCE: The proposed research investigates how the structures of the medial temporal lobe support memory formation and flexible use of memory. Understanding the mechanisms that allow the flexible use of experience will provide a means to both predict and promote the occurrence of flexible learning and potentially enable the development of tools to enhance these processes in specific populations characterized by deficits in flexibility, such as individuals with schizophrenia, mild cognitive impairment, Alzheimer's disease and depression. !

Project summary Memory-based cognition depends on both the ability to remember specific details of individual experiences (specificity) and the ability to combine information across experiences to derive new knowledge (generalization). It is well established that the hippocampus rapidly encodes specific events, but the mechanisms of memory generalization are less clear. Traditionally, a division of labor has been postulated between the hippocampus, learning rapidly to support specificity, and other memory systems, learning slowly and incrementally to support generalization. More recent research has also shown a role of the hippocampus in rapid generalization, where new knowledge is derived by combining information across a small number of events. It is currently debated whether the hippocampal contribution to generalization is fully explainable by its role in storing specific experiences or whether the hippocampus integrates information across experiences to form generalized knowledge. Furthermore, it is unclear whether generalization learning proceeds at the expense of memory for specific details or whether multiple types of memory representations form and co-exist to serve multiple forms of cognition. The current project tests the hypotheses that representations at distinct levels of specificity form in parallel along the long axis of the hippocampus through interactions with distinct cortical regions. More anterior generalized representations inform generalization. More posterior specific representations support memory specificity but can also inform generalization decisions under circumstances when generalized representations have not been formed. To test these hypotheses, the proposed studies use functional MRI in humans during concept-learning tasks in which both specificity and generalization learning is indexed using (1) behavioral methods, (2) cognitive modeling and model-based fMRI, and (3) neural pattern similarity analyses. The studies will determine the mechanism(s) of memory generalization and how it relates to memory for specific events, the degree to which the formation of distinct representation is under strategic control, and the hippocampal-cortical interactions that contribute to specificity and generalization learning. The results will reveal whether specificity and generalization are two behavioral expressions supported by a single neural representation, a single neural system that can form different types of memory representations under different task demands, or distinct neural systems that form distinct representations. The results will inform current theories of memory function and help reconcile a decades-long debate regarding the nature of concept representations. The basic science knowledge obtained here will help shed light on the memory and generalization mechanisms that may go awry in many neuropsychiatric disorders.