Separate brain regions code for salience vs. valence during reward prediction in humans

Area of Science:

• Neuroscience
• Computational Psychiatry
• Decision Science

Background:

• Survival necessitates predicting rewards and avoiding aversive conditions.
• The ventral striatum is implicated in appetitive reward prediction by computational models.
• It remains unclear if the ventral striatum also mediates responses to aversive conditions.

Purpose of the Study:

• To investigate whether the ventral striatum responds to aversive prediction errors.
• To determine if the same neural system processes both appetitive and aversive prediction errors.
• To explore how the brain differentiates between appetitive and aversive stimuli.

Main Methods:

• Functional magnetic resonance imaging (fMRI) blood oxygen level-dependent (BOLD) measurements in healthy volunteers.
• Conditioning participants using both appetitive and aversive stimuli.
• Utilizing a temporal difference (TD) learning algorithm to estimate reward prediction error (RPE).

Main Results:

• Ventral striatum activation robustly correlated with RPE, irrespective of stimulus valence (appetitive or aversive).
• This suggests the ventral striatum processes salience prediction error.
• The orbitofrontal cortex and anterior insula showed differential activation patterns, coding for stimulus valence.

Conclusions:

• The ventral striatum appears to process salience prediction error, integrating information about both rewards and punishments.
• The orbitofrontal cortex and anterior insula are involved in discriminating the valence of stimuli.
• The ventral striatum's role at the limbic-motor interface may be crucial for learning about motivationally salient stimuli and guiding action selection.