Abstract
In the typical single-stream RSVP task, two target letters (T1 and T2) are embedded among other letter distractors, with each item presented for 100 ms and then replaced by the next item. Identification of T2 is poor if it lags the first (T1) by 200–500 ms. This phenomenon, the attentional blink (AB), has been attributed to processing demands of T1. In this set of experiments, we overlay the letter stream with a 8x8 noise matrix, which was irrelevant to the task. Observers were instructed to ignore it. In Experiment 1, 3 conditions were compared: (a) canonical (no noise matrix); (b) static noise matrix; and (c) changing noise matrix. There was no difference between (a) and (b), which performance was worse than (c). When the noise matrix changed as the letters changed, performance was enhanced. In the following experiments, we show that the critical loci were the frame following the targets (i.e., the T1+1 and the T2+1 frames). In Experiment 2, the focus was the frame succeeding T2 (i.e., T2+1 frame). When the noise matrices of the T2 and the T2+1 frames were identical, performance was worse. In Experiments 3 and 4, the noise matrix of the T1+1 frame was manipulated. The results showed that when the noise matrix of the T1+1 frame was different from the preceding T1 frame, performance improved. We argue that the AB is caused by a failure in rapid disengagement from the T1 stimulus. Visual cues that inform the system that the T1 episode is over facilitates disengagement and thus modulates the blink. The changing noise matrix signals the visual system that a new frame has appeared, goading the attentional system to disengage from the previous temporal locus.
NUS Grant R-581-000-042-112