Reshaping the happy face advantage with reinforcement learning.

Area of Science:

• Psychology
• Social Psychology
• Cognitive Neuroscience

Background:

• Emotion recognition is crucial for social interaction.
• Social category biases can influence emotion recognition.
• Reinforcement learning can modify evaluative associations.

Purpose of the Study:

• To investigate if reinforcement learning can alter social category biases in emotion recognition.
• To examine the impact of reward-based learning on facial emotion processing.
• To understand how learned associations influence the perception of emotional expressions.

Main Methods:

• Experiment 1 replicated the Happy Face Advantage, showing it is influenced by social category membership (White-Dutch vs. Moroccan-Dutch faces).
• Experiments 2-3 employed a reinforcement learning go/no-go task, associating specific social categories with rewards or punishments.
• Participants completed an emotion recognition task after the reinforcement learning phase.

Main Results:

• Reinforcement learning significantly influenced emotion recognition.
• A consistent main effect of valence on emotion recognition was observed, replacing the typical interaction effect between social category and expression valence.
• Learned associations through rewards/punishments altered how emotional expressions were recognized.

Conclusions:

• Aligning actions with rewards and punishments can modify social category biases in emotion recognition.
• Reinforcement learning offers a potential method for reducing or altering implicit social biases.
• These findings highlight the malleability of social perception through associative learning mechanisms.