Humans skillfully interact with objects in everyday tasks. They achieve this even though their sensory systems are inherently uncertain, and their motor systems are variable in the execution of movements. Therefore, every movement becomes a decision under risk, as actions lead to uncertain consequences. Such decisions have been studied classically with tasks involving lotteries involving uncertain gambles and explicit monetary rewards. More recently, sensorimotor decisions, specifically single movements to targets yielding rewards and losses, also have been conceptualized as decisions under risk. While human choices between gambles have long been known not to maximize expected gains, sensorimotor decisions have been well described by statistical decision theory in many tasks. However, as object interactions are inescapably governed by the laws of physics, the question arises, how people act under such circumstances.
Here, participants slid pucks to target areas, providing gains and losses within an immersive, naturalistic virtual environment. In this task, variability inherent in motor control interacts with the physical relationships governing objects’ motion, specifically the kinematics of sliding pucks under the influence of friction. We devise a series of generative models of the human sliding behavior incorporating different assumptions about how perceptual uncertainty, motor variability, and subjective gains and losses could be combined. Using model comparison, we find evidence that human sensorimotor decisions in scenarios with prospective economic outcomes take Newtonian physics into account. This result has consequences for how sensorimotor actions under uncertainty in naturalistic tasks can be conceptualized and how the mind’s internal representations of uncertainty are integrated.