Technical Program

Paper Detail

Paper: PS-1B.4
Session: Poster Session 1B
Location: Symphony/Overture
Session Time: Thursday, September 6, 18:45 - 20:45
Presentation Time:Thursday, September 6, 18:45 - 20:45
Presentation: Poster
Publication: 2018 Conference on Cognitive Computational Neuroscience, 5-8 September 2018, Philadelphia, Pennsylvania
Paper Title: Predicting memory performance using a joint model of brain and behavior
Manuscript:  Click here to view manuscript
Authors: David Halpern, Shannon Tubridy, Lila Davachi, Todd Gureckis, New York University, United States
Abstract: Understanding the links between brain and behavior is a central goal of computational cognitive neuroscience. We present a framework for simultaneous modeling of behavioral and neuroimaging data in the context of human memory acquisition and forgetting. Using a Hidden Markov Model of memory that can account for both behavioral and functional magnetic resonance imaging (fMRI) observations, we show that we can predict memory performance in held-out data at a level well-above chance and that we can surpass the predictions made by fMRI data alone as well as those made by variants of established behavioral models. This work highlights a path for better understanding the relationship between neural data and latent cognitive processes and advances a model of memory whose predictive ability could enable model-augmented learning environments.