When is an error not a prediction error? An electrophysiological investigation

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• A theory posits the anterior cingulate cortex (ACC) uses midbrain dopamine signals for flexible action selection.
• This theory links reward prediction error signals to the error-related negativity (ERN) component of event-related potentials.
• Previous findings contradicting the predicted monotonic relationship between ERN amplitude and outcome expectedness have raised questions.

Purpose of the Study:

• To investigate the discrepancy in previous findings regarding the ACC's role in reinforcement learning.
• To test the theory that ACC utilizes dopamine reward prediction error signals for adaptive decision making.
• To determine the conditions under which the ACC-dopamine-ERN relationship holds.

Main Methods:

• Conducted three distinct trial-and-error learning experiments.
• Measured event-related brain potentials, specifically focusing on the error-related negativity (ERN).
• Analyzed the relationship between outcome expectedness, ERN amplitude, and the learnability of optimal response strategies.

Main Results:

• All three experiments provided evidence supporting the initial theory.
• The strongest support for the theory was observed when participants could learn an optimal response strategy.
• ERN amplitude was modulated by reward prediction error signals, particularly in learnable contexts.

Conclusions:

• The findings support the theory that the anterior cingulate cortex (ACC) uses dopamine reward prediction error signals for adaptive decision making.
• The ACC's utilization of these signals is most effective when an optimal behavioral strategy is learnable.
• This research reconciles previous conflicting findings by highlighting the importance of task learnability.