Human decision under risk in goal-directed movements

There is considerable evidence that human economic decision making deviates from the predictions of Expected Utility Theory (EUT) (von Neumann & Morgenstern, 1944) (e.g. Allais, 1953, Kahneman & Tversky, 1979). Recent evidence from perception and action, however, suggest that humans achieve near optimal performance in many perceptual and motor decision tasks (e.g. Trommershäuser et al., 2003a,b, 2005). Two questions have emerged after such discrepancy in findings was revealed. First, what are the boundary conditions for motor decision making? Second, how does perceptual and motor decision making differ from economic decision making? In my thesis, I used three experiments to address these two questions separately.;In the first study, we investigated the limitations in movement planning ability based on a conjecture developed from previous results (Trommershäuser et al., 2003ab) to gain a better understanding on the heuristics people might have adopted in forming their motor strategies. We found that subjects showed patterned deviations from optimality that would support the heuristic-based conjecture. In the second study, we asked if humans could trade off speed and accuracy optimally to maximize expected reward in a novel motor task that rewarded both speed and accuracy. Our results suggest that humans achieved near optimal performance by correctly trading off speed against accuracy. In the third experiment, we directly compared choice behavior between classical Decision Making under Risk (DMR) conditions (e.g. Kahneman & Tversky, 1979) and motor decision under risk conditions by resolving differences in task design. Our results revealed that subjects’ choice behavior in the motor task was more consistent with EUT.;To summarize, in two experiments we revealed the limitation and the remarkable ability of humans in planning movements. These studies suggest that even in very simple motor tasks there is a sophisticated decision making system that performs the necessary computations leading to near optimal performance. By creating equivalent task design, in the third study, we further demonstrated that humans choose more rationally in motor tasks when probability is implicit in their motor uncertainty compared to equivalent economic tasks when probability is explicitly given to the chooser.