A unifying Bayesian account of contextual effects in value-based choice

Area of Science:

• Neuroscience
• Cognitive Science
• Decision Science

Background:

• Classical theories struggle to explain how context influences choice value.
• Existing models address either within-choice or between-choice context effects, lacking unification.

Purpose of the Study:

• To propose a unifying Bayesian framework for understanding context effects on incentive value attribution and choice.
• To reconcile empirical findings on contextual influences on decision-making.

Main Methods:

• Formulation of a normative Bayesian theory where incentive value is a precision-weighted prediction error.
• Demonstration of the theory's ability to explain diverse context effects.

Main Results:

• The proposed Bayesian scheme successfully explains within-choice context effects, including multi-attribute integration scenarios.
• The same scheme accounts for between-choice context effects, unifying past and present option influences.
• The theory provides a parsimonious explanation without ad-hoc assumptions.

Conclusions:

• The Bayesian framework offers a unified perspective on context effects in decision-making.
• This model clarifies the contextual nature of incentive value and choice.
• Potential implications for understanding decision-making in psychopathologies like addiction and pathological gambling.