Purchase this article with an account.
Seongmin Hwang, Andrew Hollingworth; The Reliance on Ensemble Statistics in VWM Varies According to the Quality of Item Memory. Journal of Vision 2012;12(9):352. doi: 10.1167/12.9.352.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Brady and Alvarez (2011) found that size judgments for individual items in visual working memory (VWM) were biased toward the mean size of all stimuli in the display, suggesting that ensemble statistics interact with the representations of individual objects. We investigated how the reliance on ensemble statistics is modulated by the precision of VWM representations. On each trial, a sample display of gray circles, each with a different size, was presented, followed by noise masks. To modulate the quality of the VWM representation, the number of circles (set size) and the stimulus duration were manipulated. After the mask display, participants either 1) reported the mean size of the sample stimuli (mean judgment condition) by modulating the size of a probe circle or 2) reported the size of an individual circle presented at the location of the probe circle (individual judgment condition). To measure the influence of the mean size on memory for individual sizes, we calculated the difference between the reported individual size and the mean size of the stimuli in that display (mean-referenced performance). In the conditions where array processing was highly limited by short SOA or large set size, mean-referenced performance was reliably better than actual performance in the individual judgment condition but not different from performance in the mean judgment condition. However, the difference between mean-referenced performance and performance in the individual judgment condition decreased as SOA increased or set size decreased. Furthermore, in the individual judgment condition with the shortest SOA and the smallest set size, actual performance was reliably better than mean-referenced performance. This finding indicates that the reliance on ensemble statistics depends on the quality of the VWM representation, implying that biases may be observed only when an individual item representation is unavailable or when the quality of the individual item representation is poor.
Meeting abstract presented at VSS 2012
This PDF is available to Subscribers Only