We live in a world full of visual stimuli, but only some of them are relevant to our behavioural goals. Although we should in theory process only task relevant stimuli due to the limited attentional capacity, many studies have found that participants allocate attention to expected distractor locations both when the target and distractors appear in the same display (Tsal & Makovski, 2006) and when they appear in different displays (Makovski, 2019). This phenomenon, which is called the attentional white bear (AWB) effect (Tsal & Makovski, 2006) or the preparation effect (Makovski, 2019), is often interpreted in the framework of a process-all mechanism that guides attention to all expected stimuli regardless of task relevancy. However, it still remains unclear whether attention is allocated to expected distractor locations under all circumstances. The present study aims to examine the factors that influence the manifestation AWB effect.

In eight experiments, we used a distractor-alone paradigm developed by Makovski (2019). The paradigm consisted of a memory task for most trials and a dot detection task for the remaining trials. The main manipulation involved the presence or absence of distractors during the memory retention interval in different blocks. In the distractor-present block, while participants were holding the memory items in mind, a probe dot was occasionally presented at one of expected distractor locations, and participants made a speeded response once they detected the dot. In the distractor-absent block, the probe dot was presented at an expected empty location. Faster responses to the dot in the distractor-present block compared with the distractor-absent block indicate increased attention to an expected distractor location, ie., the AWB effect.

Experiments 1 to 4 examined how task demands and prior experience influenced the AWB effect. In experiments 1 to 3, we varied task demands by requiring participants to hold either fours colours (high memory load) or one colour (low memory load) in the memory task. To assess the effect of prior experience on the AWB effect, we counterbalanced the order of the blocks between participants (distractor-1st group vs. distractor-2nd group) and we analysed their data separately. When the memory load was high, the AWB effect was observed in the distractor-2nd group, but not in the distractor-1st group. When the memory load was low, the AWB effect was found in both groups. In Experiment 4, we again used a high memory load task but eliminated the need for attentional control by replacing the distractors in the distractor-present block with repetitions of the memory items. The AWB effect was found in both groups. These results indicate that participants normally adopt a "process-all" approach, but that top-down attentional control can overrule this approach by suppressing distractor attention if there appear to be insufficient attentional resources to process all the stimuli present.

Experiments 5 and 6 replicated the main findings of Experiment 1 with a demanding orientation memory task. Additionally, we investigated the spatial specificity of the AWB effect by presenting the dot in the distractor-present block at either an expected distractor location or an expected empty location (Experiment 5), and we examined the time course of attentional deployment to the expected distractors by presenting the dot before, simultaneously with, or after the expected distractor (Experiment 6). When the AWB effect was present, its magnitude was similar regardless of whether the dot appeared at an expected distractor location or at an empty location. Moreover, attention was deployed 400 ms before the onset of the expected distractors and lasted to at least 800 ms after their offset. These results indicated that in the present paradigm, in which the memory items and the distractors were in separate displays, the AWB effect was neither spatially nor temporally specific.

Finally, Experiments 7a and 7b explored the role that abrupt onset of distractors played in the manifestation of the AWB effect. In both experiments, equiluminant stimuli were used such that the onset of distractors was signalled by a change in colour with no change in luminance. The AWB effect was found in both experiments, indicating that the effect was caused primarily by the expectation of distractors rather than by the expectation of luminance increase that typically accompanies the onset of distractors.

Taken together, these findings challenge the notion of a mandatory "process-all mechanism" and highlight its susceptibility to prior experience and task demands. While participants generally exhibited diffuse attentional allocation within the distractor display, starting 400 ms before the expected distractor onset and lasting for at least 800 ms, they were capable of overriding this approach when the task required greater attentional resources.