Stimulus Generalization as a Mechanism for Context Effects in Choice

Context effects---preference reversals produced by adding or removing options from a choice set---have been documented across consumer goods, medical decisions, perceptual judgments, and non-human animals. It remains less clear whether these effects reflect separate decision heuristics or a common process operating over the geometry of the choice set. The present paper examines the possibility that they arise from stimulus generalization, the Pavlovian principle whereby value acquired by one stimulus transfers to similar stimuli in proportion to their similarity. Formalizing this principle within a discrete-choice framework yields the Similarity-Weighted Utility (SWU) model. In simulations using a single fixed parameter set, the model reproduces a wide range of reported phenomena. A limited extension is considered for boundary cases in which options are difficult to individuate and similarity produces interference as well as generalization. Because SWU retains the random-utility form, its parameters can be estimated by maximum likelihood and tested against empirical choice data.