Abstract
When observers try to identify two successive target items (e.g., letters) in a rapid stream of distractors (e.g., digits), accuracy for the second target is impaired for inter-target lags of 100–500 ms. This is the well-known attentional blink. Yet, when the same stream of items is presented more rapidly (e.g., 50 ms/item), accuracy is higher for the second target than the first when they appear in succession. To account for both of these findings, Potter et al. (2002, JEP: HPP, 28, 1149–1162) proposed a two-stage competition model in which the first target is detected in a first stage, but whose representation remains labile until it can be transferred to a more stable second stage in which conscious identification is possible. We tested this model by varying the types of distractors preceding the first target. Experiment 1 replicated the finding that second target accuracy varied directly with three different inter-target intervals (53, 107, 213 ms/item). Accuracy for the first target improved as the interval was shortened, whereas second target accuracy decreased. In Experiment 2, we omitted the stream of distractors that preceded the first target and found that first target accuracy was no longer impaired at the shortest interval. This finding is contrary to the claim that first target accuracy is decreased at fast rates of presentation because of its labile representation. Experiments 3 and 4 tested alternative explanations for these findings, including forward masking of first target by the leading stream of distractors and task switching from a “rejection mode for leading distractors” to an “acceptance mode for targets.” The findings strongly support the view that perception of items in a rapid stream is governed by multiple factors. These include attentional switching (Kawahara et al., in press, Psychological Research) and object-substitution masking (Di Lollo et al., in press, Psychological Research).