Knowing you may miss the goal: Integration of response variability into the movement plan

[2] Vision Science Program, School of Optometry, University of California at Berkeley, Minor Hall 360, Berkeley, CA 94720-2020, USA

We recently carried out a series of experiments studying human movement planning in video game-like situations, where there are explicit monetary rewards and penalties associated with touching objects at various locations in space (Trommershäuser et al., 2003, JOSA, 20, 1419). Under these conditions, an optimal movement strategy, which maximizes the expected gain of the movement, not only has to account for the monetary consequences of the intended movement, but also for the consequences of possible errors which may carry the movement into "costly regions."

Method. We performed two experiments to test the assumption that humans use an estimate of their own variability in planning their motor responses. In both experiments, subjects had to rapidly touch a target region while trying to avoid hitting an adjacent or overlapping penalty region. Hitting the target scored a monetary reward, hitting the penalty a monetary penalty. In the first experiment, we compared performance in two stimulus configurations, one of which was a copy of the other reduced in scale, whereas the subjects' motor variability did not change from one configuration to the other. In the second experiment the subjects' variability was manipulated using a visuo-haptic force-feedback device (PHANToM™) to perturb the visual feedback of the motor response. Rewards and penalties were scored based on the perturbed (and noisier) representation of the finger, forcing subjects to generate a representation of "their" new response variability.

Results. In both experiments, subjects shifted their movement end points further away from the penalty when their response variability increased, correctly allowing for the relative (Experiment 1) and added (Experiment 2) increase in variability. Overall, subjects' performance did not differ significantly from optimal performance. The results suggest that humans generate a representation of their own response variability and take it into account in movement planning.