Abstract: Working memory (WM) representations can capture attention in top-down manner, even when it is irrelevant to current task performance (i.e., non-target WM representations). Non-target WM representations produce attention suppression through task relevance and cognitive control. WM representations not only suppress attention at the color relevance, but also at the abstract conceptual relevance. Previous studies have claimed that a substantial and earlier bias in orienting toward color relevance compared with conceptual relevance. However, it remains unknown whether the priority of processing will affect the attention suppression of different relevance in the non-target state, that is, whether the priority of processing leads to the separation of attention suppression. The study employed working memory and visual search dual-task paradigm. Experiment 1 and 2 adopted a single-factor, three-level, within-subject design. In baseline condition, neither the memory item nor the search items were related in terms of color or meaning. The color related condition was defined as the condition that the color of the memory item was the same as some of the search items (that were the distractors). The conceptual related condition was referred to the situation that the color of the memory item was the same as the meaning of distractors. Before the experiment, the subjects were clearly informed that "the memory item is always different from the search target" and the subjects were motivated to suppress. All the target and distractors had a gap, and participants were required to make a speeded response to the search array that indicated which of the two possible target objects was present (i.e., a square with a top gap or with a bottom gap). The main difference between Experiment 1 and Experiment 2 was whether the distractors are salient or not. In Experiment 1, only color related distractors were colored, conceptual related distractors and neutral distractors were white, however, all search items were colored in Experiment 2. In Experiment 1, the results showed that there was no significant difference between color related condition and baseline condition, while conceptual related condition led to longer response times (RTs) compared with baseline condition. On the contrary, in the early stage of a speeded response task, the RTs of color related condition were longer than baseline condition. This indicates that the color relevance is suppressed under the influence of active inhibition strategy, while conceptual relevance is not successfully suppressed. In experiment 2, the perceptual salience effect was eliminated. The results showed that the RTs of color related condition were longer than the baseline condition, while the RTs were no difference between conceptual related condition and baseline condition. This indicates that attention capture is observed in the color relevance, however, attention capture is not found in the conceptual relevance. In conclusion, the color relevance is processed earlier than the conceptual relevance. The priority of processing has an impact on the inhibition strategy, which is mainly characterized by selective suppression of different relevance in Experiment 1. Experiment 2 confirms that the perceptual salience also contributes to the formation of inhibition strategy.

Key words: non-target working memory representation, color relevance, conceptual relevance, perceptual salience, inhibition strategy

摘要： 采用工作记忆—视觉搜索双任务范式，探讨在非目标工作记忆表征下，颜色关联和抽象关联的注意捕获和抑制现象。结果显示，颜色关联条件的干扰子在快速反应阶段捕获注意，随后被成功抑制，而抽象关联条件的干扰子未被抑制，仍能捕获注意；知觉凸显消除后，对颜色关联的抑制消失，颜色关联条件重新捕获注意。结果表明，非目标下抑制过程受不同关联干扰子加工过程的影响，选择性抑制了颜色关联干扰子，而知觉凸显有助于抑制策略的形成。

关键词: 非目标工作记忆表征, 颜色关联, 抽象关联, 知觉凸显, 抑制策略