Learning from imagined experiences via an endogenous prediction error.

Area of Science:

• Cognitive Neuroscience
• Neuroeconomics
• Psychology

Background:

• Preferences are shaped by experiences, especially those involving unexpected outcomes that generate prediction errors.
• Prediction errors are crucial for learning and updating beliefs about the world.

Purpose of the Study:

• To investigate whether prediction errors can arise endogenously from imagined events.
• To explore the neural mechanisms underlying preference learning driven by imagination.

Main Methods:

• Participants repeatedly chose between acquaintances and imagined interactions.
• A computational model was used to quantify prediction errors from imagined scenarios.
• Functional magnetic resonance imaging (fMRI) was employed to examine brain activity during the task.

Main Results:

• Individuals developed preferences for acquaintances with whom they imagined unexpectedly pleasant events.
• Computational modeling confirmed that prediction errors, driven by rewarding imagined experiences, explained this preference shift.
• Striatal activity, associated with prediction error signaling, was observed during the learning process.

Conclusions:

• Imagination can elicit prediction errors, driving endogenous learning and preference formation.
• The brain utilizes a reinforcement learning system, involving the striatum, to update representations based on imagined outcomes.
• This study highlights the powerful role of imagination in shaping our social preferences and cognitive processes.