Computational and neurophysiological research has highlighted neural processes that accumulate sensory evidence for perceptual decisions . These processes have been studied in the context of highly simplified perceptual discrimination paradigms in which the physical evidence appears at times and locations that are either entirely predictable or exogenously cued (e.g., by the onset of the stimulus itself). Yet, we are rarely afforded such certainty in everyday life. For example, when driving along a busy motorway, we must continually monitor the movements of surrounding vehicles for events that call for a lane change. In such scenarios, it is unknown which of the continuously present information sources will become relevant or when. Although it is well established that evidence integration provides an effective mechanism for countering the impact of noise , the question of how this mechanism is implemented in the face of uncertain evidence onsets has yet to be answered. Here, we show that when monitoring two potential sources of information for evidence occurring unpredictably in both time and space, the human brain employs discrete, early target selection signals that significantly modulate the onset and rate of neural evidence accumulation, and thereby the timing and accuracy of perceptual reports. These selection signals share many of the key characteristics of the N2pc component highlighted in the literature on visual search [3, 4] yet are present even in the absence of distractors and under situations of low temporal and spatial uncertainty. These data provide novel insights into how target selection supports decision making in uncertain environments.
Trinity College Dublin ->