Abstract
Even in the simplest of visual searches, observers are faster if they know what they are looking for. Thus, search for a red item among green is faster if Os know they are looking for red, even though a unique red singleton would be easy to find in the uninformed case. How long does it take for top-down information about target identity to become effective in guiding the deployment of attention? Are all forms of top-down information equivalent? In order to answer these questions, Os searched amongst photorealistic objects for a target that had been cued 50–800 msec prior to presentation of the search array. There were 8 conditions, each run separately. In 3 Blocked conditions, targets were constant across a block of trials. In different conditions, all targets could be EXACTLY the same (e.g. always a specific dog), of the same TYPE (always a dog), or of the same CATEGORY (always an animal). In 5 Mixed conditions, targets varied from trial to trial. 3 of these conditions used Picture cues. If, for example, the target was a dog, the cue could match EXACTLY (the same dog), by TYPE (the same or another dog), or by CATEGORY (any animal, except dog). The other 2 Mixed conditions had Word cues that matched the target by TYPE (“dog”) or CATEGORY (“animal”). 13 Os were tested. EXACT cues were fully effective. If an EXACT cue preceded the search array with SOA >= 200 msec, the RT was the same as the Blocked condition RT. TYPE cues were never fully effective, remaining 100 msec slower than Blocked RTs after an 800 msec SOA. Picture and Word TYPE cues showed the same asymptotic performance, though words required longer SOAs to reach asymptote. Word and picture CATEGORICAL cues had identical effects and produced RTs 150 msec longer than Blocked conditions even after an 800 msec SOA. Thus, EXACT picture cues are uniquely able to optimize search. The 50 msec differences between asymptotic performance with TYPE and CATEGORY cues may be due to stimulus uncertainty.