Abstract
The temporal characteristics of attentive tracking were studied in two experiments in order to evaluate two models of selective attention. Episodic models of spatial attention state that attention can be resolved into discrete attentional episodes, whereas continuous models state that attention moves continuously and smoothly across visual space. We use an apparent motion stimulus consisting of an array of 10 running analog clocks, which are sequentially cued such that observers can track the cue with their attention. In an adaptation of the method used by Carlson, Hogendoorn & Verstraten (JOV, in press), observers report the earliest and latest time they can read off the tracked clock, allowing us to directly measure attentional dwell time and shift time during attentive tracking. Contrary to what might be expected from episodic models of spatial attention (e.g. Weichselgartner & Sperling, 1995), we show that the time required to shift attention is inversely related to tracking rate. This is remarkable because it means that more time is lost shifting attention at lower tracking rates than at higher tracking rates: in other words, tracking slowly is less efficient than tracking quickly. In a follow-up experiment, we manipulate the duty cycle of the tracking cue and find that during tracking, attentional dwell time is unaffected by the veridical duration of the tracking cue. Instead, the temporal dynamics remain dependent on the tracking rate, as would be predicted by a continuous model of attention. Together, our results indicate that the attentional system sacrifices temporal resolution to accommodate a smooth percept during attentive tracking, thus supporting a continuous model of attention.