How prediction errors shape perception, attention, and motivation.

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• Prediction errors (PE) are central to reinforcement learning, perception, decision-making, and cognition.
• PE signals are widely reported across diverse brain regions and experimental paradigms.
• Recent theories propose PE computation as a fundamental mode of brain function.

Purpose of the Study:

• To synthesize and analyze commonalities and differences in reported PE signals.
• To explore the encoding, generation, and functional roles of various PE types.
• To evaluate the hypothesis that PE computation is a core brain mechanism.

Main Methods:

• Review and synthesis of existing literature on prediction error signals.
• Comparative analysis of PE encoding across different brain regions.
• Examination of neural circuitry underlying PE generation and function.

Main Results:

• PE encoding is a widespread computational process in the brain.
• The content and functional roles of PE signals are highly variable.
• Differences in PE signals are attributable to specific neural network architectures.

Conclusions:

• While PE computation is a common neural process, its specific manifestation is context-dependent.
• Understanding PE requires considering the unique afferent and efferent connections of neural circuits.
• PE signals play diverse roles, shaped by their integration within specific brain systems.