Memory and Representativeness

We explore the idea that judgment by representativeness reflects the workings of memory. In our model, the probability of a hypothesis conditional on data increases in the ease with which instances of that hypothesis are retrieved when cued with the data. Retrieval is driven by a measure of similarity which exhibits contextual interference: a data/cue is less likely to retrieve instances of a hypothesis that occurs frequently in other data. As a result, probability assessments are context dependent. In a new laboratory experiment, participants are shown two groups of images with different distributions of colors and other features. In line with the model's predictions, we find that (a) decreasing the frequency of a given color in one group significantly increases the recalled frequency of that color in the other group; and (b) cueing different features for the same set of images entails different probabilistic assessments, even if the features are normatively irrelevant. A calibration of the model yields a good quantitative fit with the data, highlighting the central role of contextual interference