Perceptual learning and decision-making in human medial frontal cortex.

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• The traditional view posits perceptual learning modifies early sensory representations.
• Recent evidence challenges this, suggesting higher-level cognitive processes may be involved.

Purpose of the Study:

• To investigate if perceptual learning can be explained by reinforcement learning mechanisms.
• To identify the neural correlates of perceptual learning, particularly changes in higher decision-making areas.

Main Methods:

• Subjects underwent an orientation discrimination task with feedback over 4 days.
• Functional magnetic resonance imaging (fMRI) was used to acquire brain activity data on days 1 and 4.
• Behavioral data was modeled using a reinforcement learning framework.

Main Results:

• Behavioral improvements were accurately predicted by a reinforcement learning model.
• Stimulus orientation was encoded in both early visual and higher cortical regions (e.g., lateral parietal cortex, anterior cingulate cortex).
• Activity patterns in the anterior cingulate cortex (ACC) specifically correlated with changes in decision variables during learning.

Conclusions:

• Perceptual learning involves changes in higher-order brain regions, specifically the ACC.
• Reinforcement learning provides a viable model for explaining perceptual learning.
• Perceptual and reward learning may share common neurobiological underpinnings.