Movement planning and decision making under risk
I’ll first describe a model of optimal movement planning based on statistical decision theory. The model allows us to predict the movement plan which maximizes expected gain in scenes where there are explicit monetary gains and losses associated with touching objects at various locations in space. The ‘ideal mover’ in these scenes must not only take into account the consequences of an intended movement, but also the consequences of their own variability in executing the movement. I’ll describe two experiments in which we compared each subject’s performance to optimal performance as predicted by the model. We found that subjects’ performance did not differ significantly from optimal. In a third experiment, we increased the subject’s effective motor variability to find out whether they would correctly compensate. They did so. I’ll show that movement in such tasks is precisely equivalent to decision making under risk. We found that subjects acted so as to maximize expected gain in our tasks, while, in contrast, humans typically fail to maximize expected gain in paper-and-pencil experiments drawn from the decision making literature. Subjects in these experiments seem to exaggerate small probabilities, transform values nonlinearly and overweight losses relative to gains. I’ll describe one final experiment in which we contrast the predictions of well-known model of human decision making under risk, Cumulative Prospect Theory (Kahneman & Tversky, 2000) to what movers actually do in our tasks. This model captures the distortions in decision making that human observers typically exhibit. we find no evidence for comparable distortions in movement under risk.