Causal inference, prediction and state estimation in sensorimotor learning.

Area of Science:

• Neuroscience
• Motor Control
• Computational Theory

Background:

• The sensorimotor system must differentiate between self-generated errors and external perturbations to effectively learn and adapt movements.
• Previous research indicates humans ignore internally generated reaching errors (motor noise) but adapt to external errors (visual perturbations).

Purpose of the Study:

• To formalize the understanding of how the brain parses movement errors into internal and external components.
• To propose and validate a novel Bayesian decision-making model for sensorimotor adaptation.

Main Methods:

• Replication of previous findings with 16 neurotypical adults.
• Development of the Parsing of Internal and External Causes of Error (PIECE) model, a Bayesian decision-making framework.
• Comparison of the PIECE model against three established hand-to-target alignment models.

Main Results:

• The PIECE model accurately captures the precise parsing of internal versus external errors observed in human behavior.
• Motor corrections are shown to reflect state estimation and belief in external perturbation.
• The PIECE model outperforms hand-to-target alignment models in explaining the observed error parsing.

Conclusions:

• Sensorimotor adaptation involves causal inference about the error source, challenging theories focused solely on hand-to-target alignment.
• The nervous system effectively discounts intrinsic motor noise while adapting to external perturbations.
• The PIECE model provides a normative explanation for maintaining finely calibrated movements.