Improving long-term retention of generalized knowledge and detailed memory by shaping neural representations during learning
University Of Texas At Austin, Austin TX
Investigators
Linked publications & trials
Abstract
PROJECT SUMMARY Optimal learning methods often differ based on task goals, whether it be enhancing generalized knowledge that can be flexibly applied to new contexts or remembering specific details of a single event. Research has identified different neural codes that support these two different types of learning: integrated representations stored in anterior hippocampus (aHPC) and medial prefrontal cortex (mPFC) promote knowledge that is generalizable, whereas pattern separated representations stored in posterior hippocampus (pHPC) and lateral prefrontal cortex (lPFC) promote descrimination and specificity. However, there is a general lack of understanding of how integrated and separated neural representations may compete or interact, and how different learning experiences can shape the formation and retention of different neural representations to drive subsequent behaviors. Thus, the overarching goal of this proposal is to identify the mechanisms and training conditions that promote the successful acquisition and retention of different types of knowledge. We propose a neuro-computational approach to examine category learning, which allows us to simulteneously measure and assess trade-offs associated with the acquisition and retention of generalized (category-level) and detailed (exemplar-level) knowledge as a function of different learning experiences. In Aim 1, we will use model-based fMRI analyses to characterize how attention biases formation of neural representations in aHPC via interactions with mPFC to improve generalization at the cost of specificity. In Aim 2, we will manipulate stimulus presentation order to determine how different learning experiences can bias individuals to form general (integrated) vs. specific (pattern-separated) knowledge representations in HPC and PFC using representational similarity analysis (RSA) techniques. In Aim 3, we will use connectivity measures and RSA to identify neural patterns within the HPC-PFC circuit that predict successful consolidation of general vs. specific information and determine the conditions associated with successful long-term retention of knowledge. Findings from these proposed studies can help uncover ways to improve training paradigms and target learning interventions to individualize and optimize knowledge retention. This research is particularly relevant to patients with Alzeimer?s disease, epilepsy, and dementia, who experience profound deficits in learning and memory retention.
View original record on NIH RePORTER →