Abstract
We examined the detection of events within scenes composed of many events. Previous results indicate that attentional set is critical: the detection of target events is more accurate when observers focus on a single event type than multiple event types. The present experiments indicate that detecting one event has a negative impact on detecting subsequent events, and that the decrements (blinks) last over 2 sec. After the blink, an optimal attentional set is re-established in favorable conditions.
Method. Observers monitored 60 sec streams composed of 144 asynchronous, 4 sec events. Blocks of trials were either single event (1 event type) or multi-event (4 event types). Event types were visually distinctive from each other and had their own critical dimension (color, shape, location, or motion). Multi-task blocks were grouped by event type or distributed (event types intermixed). Detection accuracy for targets was the main measure (false alarm rates <1%).
Results. As previously, single-event performance (81% hit rate) exceeded grouped multi-event (64%) which exceeded distributed multi-event (48%). Of most interest were functions relating hit rate to amount of time after an initial target. In single-event and grouped multi-event conditions, there was a 20% drop for targets occurring 0 - 1 sec after an initial target -- a blink. Performance then recovered in a linear manner, reaching asymptote at 3 sec. The asymptote indexes recovery of an optimal attentional set. Complex responses caused less recovery (lower asymptote) -- a less effective set. Complex responses combined with distributed multitasking to eliminate reinstatement of set. These data support a detailed interpretation of attentional set in the present situation, in terms of the use and switching of search templates for dynamic event types. And the data indicate that some conditions prevent setting of set.
Meeting abstract presented at VSS 2012