Purchase this article with an account.
Anastasia Belinskaia, Igor Utochkin; Is there a general "statistical module" in visual perception? . Journal of Vision 2016;16(12):49. doi: 10.1167/16.12.49.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Observers are good at rapid extraction of summary statistics at a very brief glance at multiple objects. Although this ability is well documented for item numerosity, the average feature or variance in separate domains, less is known about the general cognitive architecture of statistical processing in vision (Haberman, Brady, & Alvarez, 2015). We studied individual differences in the ability to judge different statistical properties to establish whether there can be a single factor affecting the precision of statistical judgments ("statistical module"). Our participants (N = 93, 51 women, mean age = M=19,53) performed three tests requiring them to compare two sets of briefly presented dots to determine which one (1) was more numerous, (2) had larger mean size, or (3) had greater size variability. As size was a relevant dimension, we also evaluated rapid size discrimination in a popout visual search task which required detection of a size singleton in a display similar to our ensemble tasks. Serial visual search for a size target was also added to evaluate whether "statistical abilities" are related to attention. We found in the result a moderate correlation between the accuracy of numerosity and mean judgments (r=.385) but not with variance. All statistics turned out to be correlated with the accuracy of detecting a size singleton (r = .287-.534) and one of serial visual search (r = .238-.438). Overall, these extensive correlations suggest that it can be a low-level factor of individual differences (size discriminability) potentially explaining the observed correlation between number estimation and averaging. It does not require postulating a single "statistical module" in vision. However, additional research is required for more certain claims about this.
Meeting abstract presented at VSS 2016
This PDF is available to Subscribers Only