Preceding Options Affect Subsequent Listwise but Not Pairwise Choice, Even for Experts

Recent theories of decision-making, such as Decision by Sampling, suggest that people lack stable preferences. Instead, preferences are malleable and constructed in the moment by comparisons of target attributes to small samples of attribute values active in working memory. Manipulating the distribution of attribute values observed before a choice has therefore been suggested to affect subsequent choices. In a series of four experiments, we investigate if prior exposure to different distributions of attribute values affect subsequent pairwise, two-alternative forced choices and listwise choices between multiple options. We also investigate if these suggested effects are attenuated by domain expertise. We typically find that listwise choices are affected by prior experience of attributes in the predicted manner but that the pairwise choices are not. This occurs even when we hold range constant, and the effect is reduced but not eliminated by substantial domain expertise. We propose that this format dependence of the malleability of choices is an important challenge for any theory of their cognitive origin.