Temporal cues derived from statistical patterns can overcome resource limitations in the attentional blink

Abstract

Previous studies have shown that humans are sensitive to statistical patterns indicating the likely locations, identities, and timings of visual targets. Here we tested whether participants can also use this kind of information to ameliorate the attentional blink (AB)-a reduction in accuracy for the second of two targets (T1, T2) presented at brief intertarget intervals (lags). In particular, we asked whether participants can use patterns arising from differential distributions of intertarget lags across trials to predict the arrival of T2. We tested this by comparing the ABs in an aging versus a nonaging distribution of trials, where aging refers to the increased likelihood of T2, given that it has not yet occurred, when lags occur with equal frequencies. Experiments 1 and 2 showed that the aging condition yielded greater T2 accuracy at longer lags than did the nonaging condition. In Experiment 3, we used a more sensitive response time measure to show faster T2 discrimination at shorter lags in the nonaging condition. These results demonstrate that participants can predict the likely onset of T2 by using statistical patterns present in the AB task, and that they can use this ability to more effectively direct limited processing resources.

Expectancy-based modulations of lag-1 sparing and extended sparing during the attentional blink.

“…This more sensitive measure also revealed benefits linked to target probability-however, these benefits were relatively small and only occurred when probability varied by approximately .25 across conditions. The present work replicates the findings of Visser et al (2015) using T2 accuracy as a dependent measure. This provides additional evidence that probability can serve as strong temporal cue as long as differences in probability are sufficiently salient.…”

Comparing the effects of implicit and explicit temporal expectation on choice response time and response conflict

“…Other studies, however, have demonstrated that the temporal information that reduces T2 detection deficits can be learned from experience under appropriate conditions (Shen & Alain, 2012; Visser et al, 2015). Shen and Alain (2012) used an auditory version of the attentional blink task and block-wise manipulated the relative probability (80% or 20%) of a short T2 lag (lagging 2 items from T1) and a long T2 lag (lagging 8 items from T1).…”

“…In general, when the timing of T2 relative to T1 was predictable, attentional blink was attenuated (e.g., Martens & Johnson, 2005; Shen & Alain 2011; 2012; Visser, Ohan, & Enns, 2015; Visser, Tang, Badcock, & Enns, 2014). In particular, Martens and Johnson (2005) compared the effects of the learning and the explicit knowledge of the likely T1-T2 interval on the attenuation of attentional blink.…”

“…Nevertheless, Visser et al (2015) suggest that even when temporal information is visually presented, the establishment of temporal expectation through statistical learning can be demonstrated using a sensitive measure of T2 detection deficits. By manipulating the relative probabilities of different T1-T2 lags, they generated two temporal structures, one in which the conditional probability of T2 increased with longer lags—the “aging” condition—and the other in which the probability of T2 was greatest at the shortest lag—the “non-aging” condition (Näätänen 197 ).…”

“…If temporal expectations were to be established from learning these temporal structures, for the longest lag, T2 detection deficits should be reduced to a greater degree in the aging compared with the non-aging condition, whereas for the shortest lag T2 detection, deficits should be reduced to a greater degree in the non-aging compared with the aging condition. When Visser et al (2015) used accuracy to measure T2 detection deficits, they demonstrated only the superiority of the aging condition for the longest lag. However, when they made the task speeded and used response times as a more sensitive measure of T2 detection deficits (and also increased the sample size), they were able to demonstrate both the superiority of the aging condition for the longest lag and the superiority of the non-aging condition for the shortest lag.…”