关键词: 工作记忆表征, 注意捕获, 激活状态, 干扰抑制

Abstract:

Working memory representations can guide attention toward memory-matching objects. When the memory items match the targets of a visual search task, the allocation of working memory (WM) resources contributes to the establishment of attentional templates. The memory item that receives less WM resources than others does not guide attention, even when it is stored in WM. On the other hand, two memory items that receive equal amount of WM resources guide attention simultaneously. However, it is still controversial how WM representations guide attention when the memory items match the distractors of a visual search task. Some studies found that due to cognitive control, attention cannot be guided by WM representations when they match distractors. On the other hand, other studies found that attention can be guided by even two WM representations. Could the allocation of WM resources also influence attentional guidance, when the memory items match the distractors of a visual search task?
In the present study, to answer the above question the allocation of WM resources was manipulated by varying the precision requirement of WM representations. Four experiments were carried out. Participants were asked to encode colors of items into WM and perform a subsequent memory test or a gap-location search task, which were presented randomly with an equal probability. The precision requirement of WM representations was manipulated by varying the magnitude of change between two memory test items. More precise memory representations were required to detect small changes between two memory test items than large changes. In Experiment 1, we investigated whether memory-based attentional capture was influenced by the precision requirement of WM representations. Participants were asked to memorize a color under a high or low precision requirement. In some trials, the memory color reappeared in the search task as a distractor. In Experiment 2, we investigated whether the memory-based attentional capture observed was related to the different active states of WM representations. Participants were asked to memorize two colors under a high or low precision requirement. An informative cue was presented simultaneously with the target colors to indicate which color would be tested more frequently. Each color reappeared in the search task with an equal probability. In Experiment 3, we investigated whether the precision requirement of WM representations influenced the number of WM representations that can simultaneously guide attention. Participants were asked to memorize one (memory-1) or two (memory-2) colors under high or low precision requirement conditions. Zero (match-0), one (match-1) or two (match-2) memory colors reappeared as distractors in the search task. In Experiment 4, we further explored the underlying mechanism by which precision requirement of WM representations influenced attentional guidance. The event-related potential (ERP) technique and the same experimental design as in Experiment 1 were used.
The behavioral results showed that when retaining one item in WM, the capture effect under high precision requirement was larger than that under low precision requirement. When retaining two memory items under low precision requirement, the capture effect for distractors that matched with high-priority items was larger than that for distractors that matched with low-priority items, whereas when retaining two items under high precision requirement the capture effect for distractors that matched with high- and low-priority items showed no difference. Under high precision requirement, the capture effect for the memory-2/match-2 condition was larger than that for the memory-2/match-1 and memory-1/match-1 conditions, while under low precision requirement the capture effect for the memory-2/match-2 and memory-1/match-1 conditions showed no difference, with their capture effects being larger than that for the memory-2/match-1 condition. The ERP results showed that during the maintenance phase of WM, items under high precision requirement elicited larger negative slow waves (NSW) and a larger late positive component (LPC) than items under low precision requirement. During the search task, larger N2 for distractors and smaller N2-posterior contralateral component (N2pc) for targets were elicited under high precision requirement, when the distractors matched with the memory items than when the distractors mismatched with the memory items, whereas equal N2 and N2pc were elicited under low precision requirement, when the distractors matched or mismatched with the memory items.
It can be concluded that in the contingent attentional capture paradigm, WM representations under high precise requirement can capture more attention than that under low precise requirement. Its underlying mechanism is that maintaining WM representations under high precision requirement costs more resources than that under low precision requirement, and therefore the resource for searching targets declines and the attention captured by memory-matching distractors increases.

Key words: working memory representation, attentional capture, active state, distractor inhibition