Abstract
Ensemble statistics are a means by which our brains can take the average across features such as color (De Gardelle & Summerfield, 2011), size (Chong & Treisman, 2003), and texture (Alvarez & Oliva, 2009) to help us retain the gist of a scene without relying on effortful processing of every object. Other research has shown that features such as color appear to be mixed together when attention is split (Golomb et. al, 2014). We wanted to examine how focused attention versus distributed attention would influence the use of ensemble statistics and feature mixing. To test this, subjects fixated on the center of a computer screen and three squares appeared in their periphery. Either all three of the squares became bold (a distributed-attention trial) or just one became bold (a focused-attention trial), and subjects were asked to direct their attention to the bold square(s). Each square was then briefly filled with a different color, and subjects were asked to report the color of one of the squares (instructed with a post-cue) by clicking along a color wheel. Using probabilistic modeling, we found that subjects tended to report a color shifted toward the ensemble mean (the mean across all colors) in both focused-attention and distributed-attention trials, and these conditions were not significantly different. Focused-attention trials showed a decrease in overall noise, with lower standard deviations and guessing rates. Focused-attention trials also showed a decreased frequency of "swap" errors (reporting one of the distractor colors instead of the target color). These results suggest that ensemble statistics can bias the perception of features by drawing them toward the ensemble mean, and focusing spatial attention does not appear to modulate this effect.
Meeting abstract presented at VSS 2017