Responsibility assignment in redundant systems

Abstract

Redundancy is a fundamental feature of biological motor systems. For example, when touching an object, many different combinations of movements of the shoulder, elbow, wrist, and finger joints result in the same movement at the fingertip. Exploiting this redundancy, the motor system distributes work across effectors to minimize signal-dependent noise and effort. When an error occurs, however, the motor system must assign the error to specific effectors, even though it may be ambiguous which effector caused them. Here, we studied the principles of responsibility assignment by using a bimanual task, in which the left and right hands jointly moved a visual cursor. We found that participants assigned errors, which were induced by visual rotation of the cursor, in a unified manner for correction and adaptation; the hand that corrected more for the error within the current movement also showed a bigger adaptive change in the next movement. Right-handed participants corrected errors more with their left hands, even though they corrected faster with their right hands in nonredundant tasks. Further experiments show that the motor system assigns responsibility preferentially to the hand that was previously exposed to larger errors. Our results show that responsibility assignment is a flexible process that attributes errors to the most likely cause.