Episode Blending as Result of Analogical Problem Solving

We know that misinformation presented in interrogating questions or in advertising produces blendings, that even imagining a possible episode might produce blending as well, however, we do not know whether reasoning and problem solving can produce the same effect. On the other hand, models of analogy-making assume “perfect memory” for old episodes. The AMBR model of analogical problem solving has mechanisms for interaction between memory and reasoning which explain partial memory and memory distortions and has predicted blending effects which are due to the reasoning process. Such predictions have no parallel in any other model we know of. There has been no experimental support for these predictions so far. The current paper describes an experiment explicitly designed to test these predictions. It consists of three sessions: 1) solving three problems, 2) solving two more target problems by analogy with some of the problems in the first session, and 3) reproduction of the three problems in the first session. The results demonstrate that the degree of blending in the recalled stories depends on the target problem solved in the second