Biased competition theory suggests that stimuli matching the contents of visual working memory can always capture attention. However, the tests of this hypothesis failed to product consistent results. Some researches indicated that items matching the contents of visual working memory can automatically capture attention (i.e., memory-based attentional capture). Others argued that the contents of visual working memory are not necessary to capture attention. They can be rejected when they were irrelevant to the goal of the current task (i.e., memory-based attentional rejection). One of the remained questions is that whether or not the memory-based attentional capture effect is obligatory during the early deployments of visual attention, or in other words, whether or not the memory-based attentional rejection effect can be found during the early attentional deployments.

Present study conducted an ERP experiment to investigate the time course of memory-based attentional guidance. Participants performed a visual search task while concurrently maintaining an item in visual object working memory. The search task included one target and one distractor. In matching trials, the search distractor shared the same color with the memory item. In control trials, neither the search target nor the distractor matched the color of the memory item. Meantime, the proportion of matching trials (20%/50%/80%) was manipulated. Higher proportion of matching trials elicits higher level of cognitive control incentive.

Behavioural results showed search RTs were faster in matching trials than in control trials when the proportion of matching trials was set to be 50% and 80%, suggesting that items matching the contents of working memory can be rejected. However, no effect of trial type was found on search RTs when the proportion of matching trials was 20%. ERP results firstly revealed N2pc components at approximately 200~300 ms post-stimulus in all three proportion conditions (20%/50%/80%), suggesting the memory-based attentional capture effects. Furthermore, N2pc amplitudes were reduced when the proportion of matching trials increased from 50% to 80%. N2pc latencies were speeded with higher proportion of matching trials. Secondly, the positive difference waves corresponding to the inhibition of the memory-matching item were then observed after N2pc (since approximately 300 ms post-stimulus) in 50% and 80% conditions, suggesting the memory-based attentional rejection effects. Such positive difference waves were enhanced when the proportion of matching trials increased.

In sum, our results indicate the guidance effect of visual working memory on attention has two phases, i.e., the memory-based attentional capture on the early phase and the memory-based attentional rejection on the later phase. The early memory-based attentional capture effect is involuntary, though the effect-size and time course of it can be varied by the level of cognitive control incentive. The later memory-based attentional rejection effect is voluntary and more in evidence with higher cognitive control incentive.