Stress reduces both model-based and model-free neural computations during flexible learning.

Area of Science:

• Neuroscience
• Cognitive Psychology
• Computational Psychiatry

Background:

• Flexible adaptation to environmental changes is crucial for survival.
• Stressful events are known to impair cognitive functions, including learning and decision-making.
• The neurocomputational mechanisms underlying stress-induced deficits in flexible learning remain largely unknown.

Purpose of the Study:

• To elucidate the neurocomputational mechanisms of stress-induced deficits in flexible learning.
• To investigate how stress impacts model-based and model-free learning processes.
• To identify the brain regions involved in stress-modulated flexible learning.

Main Methods:

• Combined computational modeling with functional magnetic resonance imaging (fMRI).
• Healthy participants underwent stress or control manipulation.
• Participants completed a Markov decision task with reward contingency reversals in an MRI scanner.

Main Results:

• Stress attenuated behavioral sensitivity to reward contingency reversals.
• Computational modeling revealed stress specifically impaired the use of value computations for action selection.
• Stress reduced inferolateral prefrontal cortex activity during model-free computations.
• Stress decreased posterior, but not anterior, hippocampal activity during model-based learning.

Conclusions:

• Stress specifically impairs the application of learned information in adaptive behavior.
• Stress differentially affects brain regions supporting model-free (inferolateral prefrontal cortex) and model-based (posterior hippocampus) learning.
• Stress may compromise both model-based and model-free contributions to adaptive behavior in dynamic environments.