Neural correlates of variations in event processing during learning in basolateral amygdala

Area of Science:

• Neuroscience
• Computational Neuroscience
• Behavioral Neuroscience

Background:

• Dopamine neurons signal reward prediction errors, linking animal behavior to neural reward learning.
• Existing models often treat attention as static, but some, like Pearce-Hall, suggest processing variations influence learning.
• These models propose event processing is modulated by unsigned, not signed, reward prediction errors.

Purpose of the Study:

• To investigate if neural activity in the rat basolateral amygdala exhibits characteristics of an unsigned reward prediction error signal.
• To determine the role of this signal in attention and learning when reward contingencies change.

Main Methods:

• Single-unit recordings in rat basolateral amygdala during a reward prediction task.
• Behavioral analysis of orienting responses to cues and learning rates after reward changes.
• Inactivation of the basolateral amygdala to assess its causal role.

Main Results:

• Neural activity in the basolateral amygdala showed an unsigned error signal at the time of reward.
• This signal increased after reward changes (both increases and decreases) and developed over trials.
• The signal correlated with faster orienting to predictive cues and was essential for learning from reward changes.

Conclusions:

• The basolateral amygdala generates an unsigned reward prediction error signal.
• This signal plays a critical role in modulating attention for adaptive learning.
• Findings support models where event processing is influenced by unsigned prediction errors.