Purchase this article with an account.
Ryan Ly, Hee Yeon Im, Justin Halberda; Spatial overlap of collections affects the resolution of ensemble features. Journal of Vision 2009;9(8):921. doi: 10.1167/9.8.921.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Goal: Collections of visual objects can be grouped and statistical properties of the group encoded as ensemble features (e.g., average orientation, centroid, approximate number of items). While features from multiple salient groups may be stored (Chong & Treisman, 2005; Halberda, Sires & Feigenson, 2006), spatial separation between groups may affect group selection (Watson et al, 2005). We investigated the effects of degree of spatial overlap between two groups, specified by color, on the discrimination threshold for the ensemble features of average size and approximate number.
Methods: Subjects (N=44) performed either an average size or approximate number discrimination. Trials varied the ratio of both average size and number across two briefly flashed collections (blue and yellow dots) while trial blocks varied the degree of spatial overlap between these two collections, the density of items, and the size of the display area. Accuracy and RT were recorded and the discrimination threshold for each subject was determined for each block.
Results & Conclusions: For number discrimination, analyses revealed a significant effect of overlap (p p [[lt]].01), but showed no effects for accuracy. One possibility is that subjects could adopt strategies focused on individual item sizes for computing the average size of dots (Myczek & Simons, 2008) while such strategies are not possible for computing approximate number. These results suggest that, while encoding features from more than one collection may be possible, features encoded from spatially localizable collections are more accurate than those from spatially overlapping collections.
This PDF is available to Subscribers Only