Effects of choice task on attribute memory

Winnowing process models of multiattribute choice in which alternatives are successively eliminated in stages based on partial information predict differential memory for minor attributes as a function of an alternative's relative desirability. This prediction, which differs from the results of judgment studies finding little or no relationship between evaluation and memory, was tested in two experiments in which the choice task (Choose 1 vs Choose 3) and the density of the option set were varied. Better memory for minor attributes of relatively desirable alternatives was observed for those conditions requiring the more difficult decisions, namely the Choose 1 task with a dense set of good alternatives. A third experiment replicated these results and demonstrated that the minor attributes did influence choice. The results of all three experiments are generally consistent with winnowing models. However, detailed analysis of the memory data revealed that only attributes of the chosen alternative were remembered better; this is not entirely consistent with those models. Implications of these results for models of memory and choice are discussed.