Meeting Banner
Abstract #2271

Resting-State fMRI Activity in the Basal Ganglia Predicts Unsupervised Learning Performance in a Virtual Reality Environment

Valur Olafsson1, Chi Wah Wong1, Markus Plank2, Joe Snider2, Eric Halgren3, Howard Poizner2, Thomas Liu1

1Center for Functional MRI, UCSD, La Jolla, CA, United States; 2Institute for Neural Computation, UCSD, La Jolla, CA, United States; 3Multimodal Imaging Laboratory, UCSD, La Jolla, CA, United States

Learning without feedback is often referred to as unsupervised learning. Prior work suggests that structures in the basal ganglia have a role in unsupervised learning performance. In this study we ran a resting state fMRI study on 10 subjects that had gone through an unsupervised learning task in a complex virtual reality environment and had been assessed on their performance to learn without feedback. We found that there was significant correlation between the performance measures and resting-state fMRI measures from basal ganglia structures. Our results indicate that resting-state measures could be used to predict individual differences in unsupervised learning.

Abstract PDF Poster PDF

Keywords

ability achieve acquisition activity amplitude amplitudes anatomical asked assessed basal blipped block blocks bold brain bubble camera channel coil complex computation conjugate consecutive containing corrected correlated correlation cortex courses coverage covered cross defined derivatives disappeared dividing echoes engaged entered environment environments excitation experiment exploratory eyes fast feedback felt field fixation floor fluid functional ganglia gradient grant head indicating individual isotropic iterative laboratory learn learning linear location magnetization maps matrix measures medical memory minute months motion motions nova nuisance numerous object objects occurring often opaque open part participated percent percentage performance plank positive predict predicted predicts preprocessing prior protocol real reality receive reconstruct reconstruction regression regressors regularized relation removed resolution resting richly room roughness scale score scores segmented series session shuffled significantly slices spatial spoiled start structures subject subjects subsequent supervised system thresholded touched tracked tracking trajectory type uncovering underneath unsupervised virtual visit visited volume walked whether white whole wide wore world