Pairwise Normalization: A Neuroeconomic Theory of Multi-Attribute Choice

We present a theory of multi-attribute choice founded in the neuroscience of perception. According to our theory, valuation is formed through a series of pairwise, attribute-level comparisons implemented by (divisive) normalization — a normatively-grounded form of relative value coding observed across sensory modalities and in species ranging from honeybees to humans. As we demonstrate, “pairwise normalization” captures a broad range of behavioral regularities, including the compromise and asymmetric dominance effects, the diversification bias in allocation decisions, and majority-rule preference cycles (among several others). In binary choice, the model also offers a potential neurobiological foundation for Cobb-Douglas preferences and other classic microeconomic preference representations.