Calibration of visually guided reaching is driven by error-corrective learning and internal dynamics.

Area of Science:

• Motor control
• Computational neuroscience
• Human sensorimotor adaptation

Background:

• Visually guided reaching relies on sensorimotor calibration.
• This calibration adjusts dynamically based on sensory feedback.
• Previous models often simplified the adaptive process.

Purpose of the Study:

• To model the trial-to-trial dynamics of sensorimotor calibration during visually guided reaching.
• To investigate the role of internal dynamics and noise in adaptation.
• To compare adaptation to random versus constant visual feedback shifts.

Main Methods:

• Developed a linear dynamical system model for sensorimotor calibration.
• Introduced an internal state variable representing calibration.
• Used random shifts in visual feedback to probe adaptation dynamics.

Main Results:

• The linear dynamical system effectively models adaptation to visual feedback shifts.
• Subjects adapt to >20% of perceived error, even when unaware of shifts.
• Identified and quantified 'state noise' contributing to adaptation variability.
• Demonstrated comparable contributions of state noise and performance noise to movement variability.

Conclusions:

• Sensorimotor adaptation is a dynamic process accurately modeled by linear systems.
• Internal state noise plays a significant role in adaptation variability.
• Findings generalize across different feedback shift paradigms.