Timing is a crucial component of movement planning and control. We examined human performance in timed reaching tasks where there were asymmetric penalties for arriving early or late. An ideal movement planner based on statistical decision theory takes into account rewards, penalties and its own temporal uncertainty in movement to maximize expected gain. Such a planner responds to changes in rewards/penalties by shifting its mean arrival time by predictable amounts. We compare each subject's performance in a timed reaching task to that of an ideal movement planner with the same timing uncertainty.

Methods: Subjects pointed at spatial targets (1 cm radius circles) on a screen. Target location was chosen from a set of positions ~43 cm from the starting point of the reach. On each trial, subjects were given a brief temporal target zone (center: 650 ms), indicated by a green region on a timer bar, and a temporal penalty zone preceding or following the target (red region on timer bar). Touching the spatial target within the temporal target zone earned a reward; touching within the temporal penalty zone incurred a loss. Each of three subjects completed 256 trials across four different reward/penalty conditions. Subjects were instructed to earn the greatest reward possible.

Results: The standard deviation of reach duration increased linearly with mean arrival time. The regression line relating mean reach durations and predicted optimal durations had a slope of .98 and r² of .90. Human performance was not distinguishable from optimal performance maximizing expected gain.