No selective attentional shift despite prefrontal activation during a working-memory task with unconscious stimuli

A key process for successful working memory is to prioritize task-relevant information over distraction, i.e., to control attentional deployment. Here we investigate to what extent attentional control during a delayed match-to-sample task can be achieved when to-be-remembered items were presented unconsciously together with distracting information, and with a pre-stimulus cue that indicated whether the target was likely to appear on the left or right side of the screen. This expectation was sometimes violated (20 % of trials), requiring reorienting of attention to successfully solve the task. Moreover, the cue was uninformative of the exact location of the target, which could appear on the top, middle, or bottom part of the screen. Participants performed better than chance on unconscious trials only when the cue correctly indicated target side, suggesting an inability to reorient attention when the cues were invalid. Neural activity (fMRI BOLD signal change) in medial and lateral prefrontal cortex was significant for unconscious valid-cue trials, and remained significant for invalid-cue trials only in the lateral prefrontal cortex, although neither region was significantly modulated by cue validity. Parietal regions did not show significant activation for valid or invalid unconscious stimuli. Thus, even though activity in brain regions associated with cognitive control reached significant levels for unconscious stimuli, there was no evidence for adaptive deployment of selective attention based on unconscious information, which was the case for conscious stimuli. The ability to control attentional deployment appears to differ between conscious and unconscious working memory. Significance statement A key process for successful working memory is controlling attentional deployment. Here, we investigated how attention is controlled during a working-memory task with unconscious stimuli, by testing participants' ability to shift attention from an expected target location to an alternative location when necessary. Despite prefrontal cortex activation, participants did not reveal shift of attention when cues were invalid, suggesting an inability to adaptively deploy selective attention based on unconscious information. Additionally, no activation was observed in parietal attentional areas for either valid or invalid unconscious stimuli. These results indicate that, unlike conscious working memory, unconscious stimuli do not trigger flexible attentional shifts, highlighting a key distinction in how attention is controlled based on awareness.