Successful reinforcement history suppresses explicit and implicit error corrections

Area of Science:

• Cognitive Neuroscience
• Reinforcement Learning
• Motor Control

Background:

• Understanding how reinforcement history impacts error correction is crucial for explaining human learning and adaptation.
• Previous research has not clearly distinguished the roles of reward prediction errors versus expected value in modulating explicit and implicit error corrections.

Purpose of the Study:

• To investigate whether reward prediction errors or expected value drives explicit and implicit error corrections.
• To determine how different reinforcement histories (long-term vs. immediate) influence error correction mechanisms.

Main Methods:

• Three reaching experiments were conducted to compare two hypotheses regarding error correction.
• Computational modeling was employed to analyze the experimental data and test the proposed hypotheses.

Main Results:

• The expected value derived from a long reinforcement history modulates explicit error corrections.
• An immediate reinforcement history directly influences implicit error corrections, independent of task error.
• A history of successful reinforcement leads to suppressed explicit and implicit error corrections.

Conclusions:

• Reinforcement history, particularly expected value and immediate rewards, plays a significant role in shaping both conscious and unconscious error adjustments.
• Successful past performance can lead to reduced error correction, potentially impacting adaptability in dynamic environments.