Abstract
In many visual search paradigms, the target object is differentiated from distractor objects by multiple feature dimensions (i.e. conjunctive search). Process models make assumptions about whether these feature dimensions are processed in serial or parallel, but in practice it can be difficult to directly determine this temporal organization using conventional statistical methods. Systems Factorial Technology provides nonparametric statistics for distinguishing serial processing from parallel processing using selective manipulations of feature influence. In a previous study, we detected highly facilitatory, parallel processing in the target-absent condition of conjunctive search, but we failed to achieve selective influence in the target-present condition, precluding any conclusion of serial or parallel organization. Attributing this outcome to the uncertainty of the extent to which the participant must scour the display before locating the target and terminating their search, we repeated our previous experiment while eye tracking and mouse tracking in order to acquire other critical latencies, such as target and distractor fixation onsets and durations, which may be less sensitive to this confound. Participants were presented with a field of randomly positioned stimuli made from the conjunction of three possible colors (red and two shades of magenta) and three possible shapes (circle, octagon, diamond). The relative prevalence of a particular distractor changed across trials to manipulate the duration of color and shape processing within-subjects. Participants were instructed to indicate the presence of the target by moving the mouse to click on it. We replicated our previous results in terms of accuracy and response times, observing selective influence and evidence for parallel processing in the target-absent condition and failures of selective influence in the target-present condition. Furthermore, fixation onsets and durations were not selectively influenced, suggesting these intervals are not sensitive to the manipulation of stimulus prevalence. Implications for stage models of visual search are discussed.
Meeting abstract presented at VSS 2017