Abstract
We present three experiments that test the range of validity of a novel motor planning model (MEGaMove) based on statistical decision theory. In these experiments subjects had to rapidly touch a target region on a computer screen. Hitting the target within a prescribed time limit gained them a monetary reward. There were also one or more penalty regions on the screen that could partially overlap the target region. Hitting these regions incurred a specified monetary penalty. Late responses were also penalized. The model predicts that subjects Maximize Expected Gain in their average MOVEement end points (hence, “MEGaMove”).
In the first experiment the amount of penalty associated with a penalty region and the position of the penalty region relative to the target region were varied. Subjects followed the predictions of our model. Subjects shifted their mean movement end points away from the penalty region. This shift was larger for closer penalty circles and higher penalty values.
The second experiment provided a direct test of the assumption that subjects use an estimate of their own motor variability in motor planning. The size of the target and penalty regions was varied, while the subjects' motor variability remained constant. As predicted, subjects shifted their movement end points farther away from the penalty region for the smaller stimulus configuration (when scaled in units of the target size).
In the third experiment, stimuli included four rotated versions of a penalty/target configuration from the first experiment and four more complex configurations consisting of one target and two penalty regions. Subjects altered movement end points in direction and magnitude as predicted by our model in all configurations. Furthermore, end point variability remained as low as in the previous experiments. Thus, subjects were able to use an estimate of motor variability in planning their response in this novel and more complex situation.