Risk-sensitive optimal feedback control accounts for sensorimotor behavior under uncertainty

Area of Science:

• Motor control
• Computational neuroscience
• Robotics

Background:

• Optimal feedback control models explain human motor behavior.
• Current models are risk-neutral, ignoring movement cost variability.
• Risk-sensitive control incorporates cost variance (risk-averse or risk-seeking).

Purpose of the Study:

• To test the hypothesis that human motor control is risk-sensitive.
• To model human motor behavior under uncertainty using risk-sensitive optimal control.
• To differentiate between risk-sensitive and risk-neutral control strategies.

Main Methods:

• A sensorimotor task involving controlling a virtual ball with Brownian motion.
• Minimizing a cost function combining positional error and control effort.
• Manipulating levels of Brownian motion noise and cost weightings.

Main Results:

• Human subjects exhibited risk-averse behavior when facing increased movement uncertainty.
• Movement strategies shifted pessimistically with higher uncertainty.
• Results align with predictions of a risk-averse optimal controller.

Conclusions:

• Human motor behavior demonstrates risk-sensitivity.
• Optimal feedback control models should incorporate risk-sensitivity.
• Risk-sensitivity is a fundamental aspect of motor control under uncertainty.