Poster D48, Monday, March 27, 8:00 – 10:00 am, Pacific Concourse
Dynamic coding in PFC, FEF and LIP during a change localization working memory task
Dante Wasmuht1, Eelke Spaak1, Timothy J. Buschman2, Earl K. Miller3, Mark Stokes1; 1Oxford University, 2Princeton University, 3Massachusetts Institute of Technology
Maintenance of task relevant information during working memory has been a matter of debate. Recent advances in population-level analysis reveal complex dynamics during working memory encoding. Here, we explore these dynamics using multi-electrode electrophysiological data recorded from non- human primates (macaca mulatta), performing a delayed change localization task. Neuronal activity from three different brain regions, lateral-prefrontal cortex (PFC), frontal eye-fields (FEF), intraparietal cortex (LIP), was recorded simultaneously. Using multivariate pattern analysis, we found that all task relevant features were decodable from each of the three brain regions even though mean activity declined to baseline levels during the memory delay. Moreover, we found that in all three brain areas, decoding was time-specific, the hallmark signature of dynamic coding. Applying a general linear model based approach to explore the temporal dynamics of single cells, we identified a large proportion of neurons with a dynamic selectivity i.e. neurons were selective to different features of the item to be remembered at different time points. Using simulations of artificial neuronal populations generated from the observed data we suggest that most of the observed dynamic coding might be explained by the dynamic selectivity of individual neurons. We also find a relationship between these dynamics and the intrinsic time-scales for individual neurons.
Topic Area: EXECUTIVE PROCESSES: Working memory