Abstract

In executing any speeded movement, there is uncertainty about the outcome due to spatial and temporal motor variability; the actual outcome of the planned movement is typically not completely under our control. I will first present a Bayesian decision theoretic (BDT) model of ideal movement planning that takes into account a subject’s own spatial and temporal motor uncertainty and the rewards or penalties associated with different movement outcomes. Then I will briefly describe two experiments designed to test whether subjects take into account their own motor uncertainty in planning movement. In the first experiment, subjects are asked to touch targets within a specified time window (625 – 675 ms). Late or early arrivals result in draconian penalties and we examine how these penalties affect subjects’ planning of movement. In the second of the two experiments the subject must complete the first part of the movement with only partial knowledge of the actual location of the target represented as a Bayesian prior on possible targets. We find that subjects’ performance is typically close to that of a BDT ideal movement planner maximizing expected monetary reward. This outcome is surprising: these motor tasks are formally equivalent to decision making under risk and subjects making decisions under risk typically do not maximize expected gain. I’ll describe very recent work in which we translate classical decision making experiments (concerning the Allais paradox) into motor form and directly compare decision making under risk to “movement planning under risk” in the same subjects. The results suggest that, while individuals value rewards identically in planning movement and in making economic decisions, their use of probability is markedly different. I will also present fMRI results concerning the neural representation of motor uncertainty.