Purchase this article with an account.
Stephen M. Emrich, Adam C. Riggall, Joshua J. LaRocque, Bradley R. Postle; The Representation of Stimulus Identity for Multiple Items in Short-Term Memory Revealed Using fMRI Classification. Journal of Vision 2012;12(9):174. doi: 10.1167/12.9.174.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Numerous neuroimaging studies have revealed that sustained activity in the intraparietal sulcus (IPS) and lateral occipital (LO) cortex is present during the delay period of change detection tasks, and increases as a function of memory load, suggesting that these areas play a critical role in visual short-term memory (VSTM) maintenance. In contrast, using multivariate multivoxel pattern analysis (MVPA), other studies have found evidence for stimulus-specific information in early visual cortex, suggesting that these regions also play a critical role in representing information in VSTM. In the present study, we examined this discrepancy between univariate and multivariate techniques by examining the effect of load on the classification of motion direction during a VSTM task. On each trial, subjects were presented with 1, 2 or 3 apertures of moving dots at 100% coherence (the memory sample). After a 9 second delay, a circle with a line indicating the direction of the motion was presented, and participants were told to orient the line to indicate the direction of the sample stimulus presented at that location. For each trial, the target was selected from one of 3 tangential orientations (0°, 120°, and 240°). Data from individual trial time points were used to train separate classifiers to classify each of the 3 directions of motion. A trial is considered correctly classified if the correct direction had the highest likelihood (winner-take-all classification). Results show that classification accuracy is above chance for loads 1 – 3. Overall accuracy, however, as well as target-direction evidence, decreased as a function of load, consistent with the decrease in VSTM performance with increasing set size. The results are compared to behavioral measures of accuracy and resolution.
Meeting abstract presented at VSS 2012
This PDF is available to Subscribers Only