Abstract
Feature-bound object representations facilitate matching of a perceptual input with a representation in visual working memory (VWM), but its underlying neural mechanism remains unclear. Using the redundant feature reviewing task (Saiki, 2016), the current study investigated the process of feature matching between a single object probe and multiple objects in VWM using MEG and EEG. A set of features was presented in a two-object memory display, followed by a linking display in which only placeholders remained visible. Then a single object probe was presented, and participants judged if it contained any features of the memory display, regardless of object correspondence. Two groups of participants in Taiwan and Japan performed this task while measuring brain activity using MEG and EEG, respectively. Both groups showed the advantage of feature conjunction relative to single features in memory matching, and this advantage was larger when target features were grouped in the memory display (intact condition) than when they were separated (recombined condition). Also, they showed the advantage of shared location between memory and probe displays, either in conjunction match or shape-only match. EEG/MEG data revealed that the difference in advantage of feature conjunction between the intact and recombined conditions was reflected by the amplitude of left frontal electrodes around 400 ms after the probe onset. Furthermore, the N3rs component reflecting retroactive search in VWM representations, defined as the mean amplitude at the midline electrodes between 340 ms and 400 ms, showed significantly smaller search costs in the intact condition than in the recombined condition, and in the two feature condition than in the one feature condition. These findings indicate that the advantage of feature conjunction in memory matching mainly reflects memory search process, and that the N3rs component reflects search cost in terms of feature redundancy, in addition to the number of memory representations.
Acknowledgement: JSPS KAKENHI Grant #JP16H01727, JSPS Bilateral Open Partnership Joint Research Projects