Crowds or photographs of crowds? Visual search through moving or static objects

The dynamic displays previously used in visual search have typically not contained low-level motion signals, but rather involved random relocation of elements. Our subjects searched for a target Gabor patch, differentiated from surrounding distractors by its carrier orientation (target/ distractor difference varied between 4 and 32 deg), in two display types: static (stationary elements) or moving (elements smoothly translating in different directions at 3.3 deg sÿ1). A target was always present, and subjects responded twice on each trial: firstly to indicate that they had found the target, at which point all elements were masked and the reaction time recorded; secondly to identify the target location. For easily identifiable targets (large orientation difference), search functions were similar for static and moving stimuli: both were efficient (shallow gradient with short intercepts). However, for less easily identifiable targets (small orientation difference), static and moving search differed: the former became less efficient, but retained short intercepts; for the latter only intercepts rose. These data present a challenge for existing search models.