Submission 485
Understanding the Computational Mechanism Underlying Context Effects in Food Bundles
SymposiumTalk-03
Presented by: Jennifer March
According to classic economic theory, rational decisions should obey a set of choice axioms (Von Neumann & Morgenstern, 1947). One such axiom, the axiom of independence, states that choice preferences should not be affected by common consequences. While violations of the independence axiom in decisions under risk and uncertainty go back as far as the Allais Paradox (Allais, 1953), our goal is to investigate whether and to what extent violations of the axiom can also be established in choice settings with multi-attribute bundle options. Using a food choice task, we examine whether preference for A over B in binary choice changes when choosing between bundles A + C and bundle B + C, where we systematically vary the multi-attribute similarity of C compared to A and B. Simulations of different choice models, including normalization, expected utility and similarity-based inhibition show attribute-dependent shifts in choice behavior. Specifically, the models make different predictions, when C has very extreme attribute values. We use simulations to optimize our experiment to facilitate differentiation between models. Our results will determine the precise mechanisms that give rise to context effects in multi-attribute bundled choice. We discuss to what extent the results from our food choice task apply to other choice domains and how this adds to the contextual nature of decision-making.