Human Performance in Timing of Discrete Actions

An experiment on human timing performance is described. A scenario of street crossing without the aid of traffic lights was presented on a computer screen. The subjects' task was to let a pedestrian to cross the street between successive vehicles by pressing a key. Independent variables were the size of the gap between vehicles and the speed of the traffic stream. The subjects' performance was evaluated by the distributions of the crossing initiation times and the proportions of two types of timing errors in the experimental conditions: Too early and too late releases of the pedestrian. Increasing the accuracy requirement by reducing the duration of the window of opportunity resulted in a shift to earlier release times, smaller standard deviations of the release time distributions, and increased proportions of the “too early” errors. Our results show a way to include also the tails of the response distributions in the analyses.