Abstract
Behavioral tasks which engage visual attention ask observers to select some sensory stimuli while ignoring others. Such tasks can be split according to how observers select visual space: either by spatial location or according to stimulus features such as color or motion direction. Most task designs elicit one or the other form of selection--but how do these compare to each other? Here we designed an estimation task to measure perceptual sensitivity to visual stimuli selected by spatial location, feature dimensions, or both. Observers fixated while four patches of colored random dot motion were shown, with each patch moving in a different direction. Two of the patches overlapped on each side of fixation and within each overlapping pair one patch was colored yellow and one blue. Observers were asked to estimate the direction of the dots within a single patch, defined by color (yellow/blue) and location (left/right), using a rotating wheel. This meant that in advance observers could be cued about which side the relevant patch would be on and then post-cued about the color, or vice versa. In control conditions observers were told the exact patch to select in advance, or given no information. We also repeated the task asking observers to estimate the color of the dot patches while cueing by direction (i.e. the four patches had random colors, and moved in one of two directions on each side). A mixture of Von Mises model showed that as expected observers (n=5, >2000 trials each) were more accurate when cued in advance. But we also found that observers were equally capable of using cues about features or location prior to stimulus presentation. Our results suggest that although selection may be implemented in many different ways these all result in a common perceptual improvement.