Purely endogenous capture of attention by task-defining features proceeds independently from spatial attention

Attention can be focused voluntarily and effectively on spatial locations in order to enhance the processing of task-relevant events. However, work on 'attentional capture' has demonstrated that spatial biases can be temporarily reset by transient and salient stimuli, especially if they share defining characteristics with the targets of a task goal. In the current study, we investigated whether the appearance of stimuli containing task-defining features at an unattended location was sufficient to capture attention, even when these were not perceptually salient. We used event-related-potential (ERP) markers to test whether the selection of task-defining features was modulated by top-down spatial attention, and to test whether the appearance of 'unattended targets' transiently disrupted the spatial bias. Surprisingly, the results revealed that ERP markers of selection of task-defining features were equivalent for stimuli appearing at spatially attended and unattended locations. In addition, the presentation of task-defining stimuli at the spatially unattended location induced a short-lived redistribution of the pre-established spatial attention bias toward the 'capture' side. These findings show that task-defining features of a stimulus are automatically processed independently from spatial attention, and suggest the co-existence of multiple sources of top-down biasing signals, which might in part sustain the capture mechanism.