Optimal Compensation for Changes in Task-Relevant Movement Variability

Abstract

Effective movement planning should take into account the consequences of possible errors in executing a planned movement. These errors can result from either sensory uncertainty or variability in movement planning and production. We examined the ability of humans to compensate for variability in sensory estimation and movement production under conditions in which variability is increased artificially by the experimenter. Subjects rapidly pointed at a target region that had an adjacent penalty region. Target and penalty hits yielded monetary rewards and losses. We manipulated the task-relevant variability by perturbing visual feedback of finger position during the movement. The feedback was shifted in a random direction with a random amplitude in each trial, causing an increase in the task-relevant variability. Subjects were unable to counteract this form of perturbation. Rewards and penalties were based on the perturbed, visually specified finger position. Subjects rapidly acquired an estimate of their new variability in Ͻ120 trials and adjusted their aim points accordingly. We compared subjects' performance to the performance of an optimal movement planner maximizing expected gain. Their performance was consistent with that expected from an optimal movement planner that perfectly compensated for externally imposed changes in task-relevant variability. When exposed to novel stimulus configurations, aim points shifted in the first trial without showing any detectable trend across trials. These results indicate that subjects are capable of changing their pointing strategy in the presence of externally imposed noise. Furthermore, they manage to update their estimate of task-relevant variability and to transfer this estimate to novel stimulus configurations.

Causal Model Comparison Shows That Human Representation Learning Is Not Bayesian

“…These findings stand in contrast to multiple demonstrations of Bayes-optimality (Doya 2007) in perceptual decision-making (Gold and Shadlen 2002;Knill and Pouget 2004), motor control (Trommershäuser et al 2003(Trommershäuser et al , 2005, multimodal integration (Körding and Wolpert 2004), reasoning (Oaksford and Chater 1994), and even setting metalearning parameters for reinforcement learning (Behrens et al 2007;Yu 2007). However, whereas Bayesian inference may be computationally feasible (and indeed, simple) in scenarios that can be reduced to a several-alternative forced-choice decision (Gold and Shadlen 2002) or a choice between lotteries (Wu et al 2009), representation learning in natural environments places much heavier computational demands on the learning system.…”

How the required precision influences the way we intercept a moving object

“…The actual performance is obviously poorer than this, but the difference is not very large. This confirms that near-optimal performance can be achieved through adjustments that are primarily based on recent experience (Brenner and Smeets 2011a;Narain et al 2013;Trommershäuser et al 2005;van Beers 2009van Beers , 2012.…”

Inside the brain of an elite athlete: the neural processes that support high achievement in sports

“…People are able to implicitly estimate the magnitude of their own variable error and use it to modify their movements in the light of the experiment's reward context 66,67 .…”