Abstract
To judge the mean location (centroid) of only a set of black dots when they are spatially intermingled with white dots, one needs a mechanism that passes information carried by the to-be-attended visual feature (black) across space (all locations of black dots) onto further mental processes while blocking the passage of the information carried by the to-be-ignored feature (white). This global, selective process, Feature-Based-Attention (FBA), can be quantitatively described as an attention filter (Drew, Chubb and Sperling, Journal of Vision, 2010). The filter’s properties are derived from the observer’s responses. Here, we measured the improvement of attention filters as a function of the time-duration available for subjects to process the stimuli. Method. A brief flash of the dot-containing stimulus was followed, at various Stimulus-Onset-Asynchronies (SOAs), by a post-stimulus masking array. Stimuli were random clusters comprising 1-, 2-, 4-, 6- or 8-each of black and white dots, randomly interleaved in blocks of different SOAs ranging from 12ms to 300ms. Control trials with stimuli composed of only black dots were interleaved in each block. Subjects mouse-clicked the mean location of all the black (target) dots. Results. (1) Judgment accuracy improved as SOA increased and reached an asymptote at 80ms whether or not white (distractor) dots were present. (2) Remarkably, the improvement in accuracy as a function of SOA for black-plus-white and black-only conditions was almost identical. (3) Attention filters, at their optimum, were very effective, giving a more than 4:1 transmission advantage to black versus white dots. Conclusion. The nearly equivalent performance with black dots alone (no attention filter) and with intermingled black and white dots (attention filter required) over a wide range of stimuli and SOAs indicates that FBA takes place at a very early stage prior to the computation of the mean of multiple locations.
Meeting abstract presented at VSS 2015