Feedback-related EEG dynamics separately reflect decision parameters, biases, and future choices

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Decision Science

Background:

• Optimal decision-making relies on dynamic learning from environmental changes.
• Learning can be biased by irrelevant information, hindering optimal performance.
• Understanding these biases is key to improving learning strategies.

Purpose of the Study:

• To investigate how task-irrelevant factors bias learning and decision-making.
• To explore the neural mechanisms underlying learning biases using electroencephalography (EEG).
• To model the computational processes involved in learning from biased feedback.

Main Methods:

• A modified probabilistic choice task was designed with task-irrelevant factors (variable payouts, surprising feedback).
• Computational modeling was used to analyze learning behavior and identify biases.
• Electroencephalography (EEG) recorded neural activity during the task.

Main Results:

• Participants' learning performance was significantly biased by the irrelevant factors.
• Distinct neural activity patterns in EEG reflected the processing of these biasing factors.
• A central to centroparietal positivity in later feedback processing indicated a convergence of signals influencing learning.

Conclusions:

• Task-irrelevant information, even when uninformative, can substantially bias learning and decision-making.
• Neural processes dynamically represent and integrate biasing information, affecting downstream learning.
• These findings offer insights into the neural basis of learning biases and potential targets for intervention.