Does the well-documented "decoy effect" emerge in decisions from experience among risky options? We conducted a series of experiments where participants made choices between gambles, and we varied whether participants learned about the options from description, experience, or both. Our results consistently showed no traces of the decoy effect when participants learned from experience. Even when participants read precise descriptions of the options, actually experiencing those options eliminated the decoy effect. Moreover, in decisions under risk (decisions from description), the decoy effect is less robust than previously thought. The decoy effect only emerged in an experimental design in which we used two decoys generating attraction for different options but did not emerge when only one decoy was used. Increasing the distance between the decoy and the target did not make the decoy effect emerge in decisions from experience but seemed to reduce the decoy effect in decisions from description. Overall, we identify two boundary conditions for the decoy effect in decisions under risk: First, it is not robust to situations that involve learning from experience; and second, the attraction of a single decoy may not be sufficient to observe a decoy effect. key words decoy effect; attraction effect; context-dependent choice; asymmetric dominance; decisions from experience; descriptionexperience gapThe "decoy effect," also known as the "attraction effect" or the "asymmetric dominance effect" (Huber, Payne, & Puto, 1982), is perhaps the most important instance of how decisions are influenced by the context in which they are made. Consider two options, A and B. The decoy effect refers to the phenomenon by which adding a third option, C, that is dominated by one of the options, say B, but not by the other (A) increases the attractiveness of the dominating option (B). The decoy effect suggests that options are not evaluated independently from one another, thus violating a central axiom in models of rational choice and a common assumption in descriptive models of choice, such as prospect theory (Kahneman & Tversky, 1979).1 It is widely taught in business schools and even presented as fact in popular media, such as the TV series Numb3rs Season 5, episode 2. 2 Huber et al. (1982) and Simonson and Tversky (1992) studied decoy effects using a variety of items and included-but did not focus on-gambles.Note. For options, the notation (p, o) describes a prospect that yields outcome o with probability p and zero otherwise. Problems "a" to "e" are the target problems and were taken from Herne (1999); problems "F1" and F2" were taken from Frederick et al. (2014); and problems labeled "nd" were fillers, in which no option was dominated.Experience and Decoy Effects 543 E. Ert and T. Lejarraga