Abstract
Timed visuomotor decisions require estimates of expected performance in order to choose an appropriate action. However, predicting performance in a task with multiple options (e.g., choosing which lane to travel on a crowded highway) is difficult. By the time we make a decision conditions may change. As a result it may be desirable to rely on strategies that speed the decision process without excessive cost to performance. To investigate these decisions, a computer mouse was used to aim and fire at a moving target before a deadline. Target velocity, size, and predictability of motion varied. When only one target was present, performance (hit rate, error, RT) depended on size, target velocity, and path predictability, with size having the largest effects. Reaction time was more correlated with size than hit rate. With two targets and instructions to hit only one, subjects choice of target was mostly rational in that most (75%) choices could be predicted by the relative probability of hitting either target as estimated from single target performance. Rational decisions and shorter RTs were much more likely when the difference between the probability of hitting the two targets was greater than .5. However there were biases; for example subjects tended to pick a larger target even when it had a lower probability of being hit. As in the single target case, larger targets also tended to elicit shorter RTs which may have contributed to the bias to pick large targets. These results suggest that internal estimates of performance are accessible for making perceptual-motor decisions. However, subjects sometimes relied on heuristics based on more perceptually accessible information (size) as well as the expected duration of planning and executing the movement.
Meeting abstract presented at VSS 2014