Poster B36, Sunday, March 25, 8:00-10:00 am, Exhibit Hall C
Neural mechanisms of precision in visual working memory
Elizabeth Lorenc1, Mark D'Esposito1; 1University of California, Berkeley
There is considerable variability, both between and within individuals, in the precision with which complex images are maintained in visual working memory (VWM). We hypothesize that precise VWM relies upon the continued maintenance of perception-related activity in stimulus-selective regions like the fusiform face area (FFA). To test this hypothesis, we collected fMRI data while participants performed a delayed-estimation task for faces. On each trial, participants were presented with a face from a continuous face space, followed by a post-cue which indicated whether to store the item through a 10s delay period (‘store”) or discard it from memory (“drop”). “Store” trials ended with a method-of-adjustment response in which a probe face was morphed to match the remembered face, and on “drop” trials, participants morphed a probe face to match a simultaneously-presented test face. We used an inverted encoding model approach to examine VWM representations in the FFA. After training an encoding model on perception-related activity patterns, we inverted this model and applied it to held-out fMRI data. First, we found that faces could reliably be reconstructed during perception, before the “store” or “drop” post-cue. Second, this model trained on perception-related activity patterns allowed successful reconstruction of faces maintained in VWM. Finally, “store” delay patterns more closely resembled perception than did “drop” delay patterns. These results indicate that the FFA is involved in maintaining the precise details of a face in VWM, and suggest that a common coding scheme underlies both perception/encoding and VWM maintenance.
Topic Area: EXECUTIVE PROCESSES: Working memory