关键词: 重复知盲, 末位优势效应, 类型标记个体化理论, 建构/归因理论, 注意资源分配最优化假设

Abstract: People are adept at identifying short lists of consecutively presented items in a rapid serial visual presentation (RSVP) paradigm. However, they often fail to detect repetitions of items (C1 and C2) when they are separated by one intervening item. This phenomenon is called repetition blindness (RB; Kanwisher, 1987). In the literature, there were mainly two views explaining why and how RB occurred, namely, “activation/inhibition” view and “construction/attribution” view (Morris, Still, & Caldwell-Harris, 2009). Neither view could explain the results from studies supporting the other view. Thus, the present study proposed an Optimization Allocation of Attention Resources Hypothesis. People can allocate attention resources according to the task to get the best recall performance. In a RSVP paradigm with repeated items, people may sacrifice deep processing of repeated items to detect non-repeated items, so RB may be caused by little attention paid to repeated items. According to the hypothesis, RB on the item repeated at the last position in RSVP would decrease because people allocated more attention to the item at the last position than to the item in the middle; no matter the context of the repeated item changed or not, RB would keep constant because people always allocate more attention to non-repeated items than to the repeated item; in a backward recall task, RB would reduce because people allocate more attention to C2 (presented later than C1) to get better recall.
The present study conducted three experiments with the RSVP paradigm to verify the Optimization Allocation of Attention Resources Hypothesis for RB. In all three experiments, the researchers used a within-subject design with two variables, manipulating Repetition (repeated and non-repeated) throughout the study and changing the other variable. Experiment 1 manipulated Position (middle and last) to investigate how the presenting position affected RB. Experiment 2 manipulated Context (changed and not-changed) to investigate how the context affected RB. Experiment 3 manipulated Task (forward recall and backward recall) to investigate how the recall task affected RB.
The accuracy rates for reporting C2 in the RSVP paradigm were compared using a two-way repeated-measures analysis of variance for each experiment. In Experiment 1, the results showed a Last Position Superiority Effect of repetition stimulus that the accuracy rate for reporting C2 at middle position was lower than that at last position in both repeated and non-repeated conditions. In Experiment 2, the interaction of Repetition and Context was not significant. And in Experiment 3, the difference of the accuracy rate for reporting C2 between the repeated condition and the non-repeated condition was not significant in the backward report.
Therefore, the results in the three experiments all supported the Optimization Allocation of Attention Resources Hypothesis. Further, the present study indicated that: 1. The last position superiority had effect on repetition stimulus because of the optimization allocation of attention resources; 2. RB occurred at the report phase, not at the perception phase; 3. The optimization allocation of attention resources hypothesis was better than the construction and attribution theory for explaining the RB.

Key words: repetition blindness, the last position superiority effect of repetition stimulus, the type-token individuation theory, the construction and attribution theory, the optimal allocation of attention resources hypothesis