Abstract
In everyday life, we often need to choose whether to move or not to move based on rapidly evolving, and potentially incomplete information. Here, we investigated how visually presented, gradually updated information influences a manual response. We used a go-/ no-go task in which we prompted observers to decide if a horizontally moving stimulus (attacker) would hit or miss a hidden goal. The vertical location of the goal was revealed gradually, by showing a sequence of six dots over the course of 1 second, providing independent samples of positions inside the goal. Observers had to touch the monitor at the goal (‘go’ response) within 1 second after the attacker started to move, but only if they thought that the attacker would intercept the goal (‘in’ trial). As a measure of uncertainty, we computed the probability of an ‘in’ trial (p'in') after each new sample. Higher probabilities of an ‘in’ trials were associated with more ‘go’ responses, and lower probabilities were associated with more ‘no-go’ responses. This pattern was also explained by the vertical distance between the samples and the position of the attacker. Comparing p'in' in correct and incorrect trials, we found that information carried by the first two samples (presented 0 and 166 ms after trial onset), differed between correct and incorrect responses. Incorrect ‘go’ responses were characterized by higher p'in' values in these time windows, and incorrect ‘no-go’ responses by lower p'in' values, compared to the correct responses in each response category. Overall, our results suggest that human observers exploit gradual updates of information for a go/no-go decision. When later samples provided evidence contradictory to former ones, responses were better explained by the later samples, indicating that an imminent decision is rapidly modified to take into account newly available information.