Acquisition of knowledge about uncertainty in the outcome of sensory motor decision tasks

Abstract

Recent years have seen a growing number of experiments in which decision making is studied in the context of visuo-motor decision tasks. In these visuo-motor experiments, the outcome of a hand or eye movement is associated with explicit gains and losses (see, e.g. [1]). The results obtained from these studies indicate that human participants typically select surprisingly efficient strategies that come close to maximizing expected gain. For example, under conditions in which the end point variance of speeded hand movements is artificially increased by perturbing the final finger position, humans quickly generate a representation of their modified end point variance and rely on this representation in planning their movements so as to maximize gain [2]. Furthermore, human subjects have been found to implicitly learn the stochasticity of spatially stochastic rewards and losses in only a few hundred trials [3]. These results indicate that humans are able to implicitly estimate the parameters of noisy movement outcomes and of stochastic environmental variables and can use this knowledge to plan their hand movements such as to maximize expected gain. A comparison of performance across these tasks indicates that knowledge about movement uncertainty is learnt within fewer trials than knowledge about the uncertainty in the final target position. This suggests that visual feedback about the hand position is processed and integrated by different mechanisms than visual information about the final target position.