Item

Abstract

A lot is known about the neural basis of directing attention based on explicit cues. In real life however, attention shifts are rarely directed by explicit cues but rather generated implicitly, for example on the basis of previous experience. Here, we aimed at studying attention shifts dependent on recent trial history. We asked observers to detect targets in a stream of visual stimuli with three feature dimensions: colour, shape and motion. Critically, target occurrence probability was always higher in one stimulus dimension than in the others, and probabilities switched between dimensions over blocks of trials. After each probability switch, target detection times decreased exponentially for high-probability targets and increased for low-probability targets, compatible with gradual shifts in attention dependent on trial history since the switch. BOLD signal in left prefrontal and intraparietal sulcus regions was higher in the early phase after the switch, while anterior cingulate, cuneus, precuneus, temporal and more anterior frontal regions showed more activation later after the switch. These findings are compatible with expectation about engagement of regions involved in the establishment and maintenance of attentional sets. BOLD signal in left intraparietal sulcus correlated with the size of the performance changes consecutive to the detected targets, suggesting that it reflects the size of attention shifts induced by updating target probabilities over recent trial history.