A Computational Cognitive Neuroscience Approach to Understanding Event Representation and Episodic Memory
Abstract
A Computational Cognitive Neuroscience Approach to Understanding Event Representation and Episodic Memory~ (Charan Ranganath, PI) to"" ONR MURI program, Topic 12 ~ Event Representation and Episodic Memory. Total costs are estimated at $7,500,000.00 over 5 years with"" 3-year base period of $4,500,000 and 2-year option period of $3,000,000. Memory for specific events (~episodic memory~) and knowl"edge accrued across multiple events (~event schemas~) play powerful roles in human cognition. Neuroscience research has strongly lin"ked these abilities to interactions between the hippocampus, ventromedial prefrontal cortex (vmPFC), and a posterior medial (PM) neo"cortical network. Gaining a mechanistic understanding of the neural processes that support representation of event schemas and episodic memory could be of enormous significance to basic neuroscience. Achieving this goal would have significant implications for DOD" capabilities, such as the development of autonomous systems capable of rapid extraction of information about temporal and spatial r""elations, causality, and intentionality from complex data streams. At present, however, there is insufficient knowledge about how ev""ents are segmented, learned, and retrieved in the human brain, and the field lacks an appropriate computational framework to explain"" existing data. To address these challenges, we propose an integrative computational cognitive neuroscience approach to understandin"g event representation and episodic memory.The centerpiece of this project is an innovative neurocomputational framework called St"ructured Event Memory (SEM). SEM integrates the strengths of three different, but related, approaches~Latent Cause Models, Event Seg""mentation Theory, and the Complementary Learning Systems Model~to explain how interactions between the hippocampus, vmPFC, and PM ne""twork support event representation and episodic memory. The architecture of SEM is scalable, it is meaningfully tied to neural mecha""nisms, and it provides a principled and parsimonious account of a diverse range of processes, including: schema formation, event seg""mentation, context in memory and language, episodic memory encoding, consolidation, retrieval, and updating, prediction, and mental"" simulation. SEM is especially innovative in that it is designed to learn about the structure of the world, in terms of temporal, ca""usal, and situational relationships within an event, and in terms of the characteristic transitions between different kinds of event""s. Building on our initial work, we will accomplish the following Tasks: (1) Develop a computational model of event representation a""nd episodic memory. (S. Gershman, Harvard), (2) Specify the neural mechanisms that support learning and application of semantic know""ledge about events. (K. Norman, U. Hasson, Princeton), (3) Specify how corticohippocampal interactions support episodic memory retri""eval and consolidation. (C. Ranganath, UC Davis), (4) Specify the cognitive and neural causes and consequences of event segmentation"" (J. Zacks, Washington University), & (5) Determine how broadband and oscillatory neural activity contributes to event segmentation"" and episodic memory (Orrin Devinsky, NYU, in collaboration with all team members).By addressing fundamental questions about how e""vent segmentation shapes the structure of semantic and episodic memory, and how the hippocampus interacts with neocortical networks,"" the proposed empirical studies go well beyond the current state of the art. Moreover, by utilizing novel data-driven machine learni""ng analysis approaches and theory-driven forward modeling of neuroimaging and electrocorticography data, every study in this project"" will break new methodological ground. At the end of the five-year period, we will deliver an integrated computational architecture"" that can translate in any direction between complex narrative or video stimuli, real-time brain activity patterns, and meaningful c
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Nov 03, 2017
- Source ID
- N000141712961
Entities
People
- Charan Ranganath
Organizations
- Office of Naval Research
- United States Navy
- University of California, Davis