Heuristics in Multi-attribute Decision Making: Effects of Representation Format

When probabilistic inferences have to be drawn based on cue values stored in long term memory, participants appear to use a number of different heuristics. These heuristics vary in the amount of available cue information that is retrieved and integrated, and in the degree to which cue validities are considered. One factor that is assumed to affect strategy selection is the representational format of cue information: Holistic, image-based representations should induce the parallel retrieval of cue values and thus foster the use of compensatory strategies that take into account multiple cue dimensions. In contrast, verbal representations should favor non-compensatory strategies that draw inferences based on a single cue dimension. We tested this hypothesis in an experiment and explored how the learning procedure modulates the use of cue validities. The results suggest that the acquisition of cue knowledge from holistic images increases the use of a compensatory prototype matching strategy that neglects cue validities.