Redefining heuristics in multi-attribute decisions: A probabilistic framework

AbstractIn this paper, we highlight the shortfall of conventionally described heuristics in multi-attribute decision theory, and propose recasting these heuristics within a novel probabilistic framework. This framework is based on defining a psychological feature space, with rule-based heuristics represented as prototypical representations within this space. We provide various examples of meaningful heuristics that can be constructed under this representation, including recasting probabilistic versions of popular heuristics such as take-the-best. Next, we propose an evaluation framework to measure the effectiveness of a consideration set of heuristics. This framework measures whether the set of heuristics are sufficient to describe, predict and infer strategy selection and learning behavior. We propose that this is a step towards a robust framework within which models of strategy selection and learning should be evaluated. The framework aspires to develop a consideration set of heuristics that can be represented as a mathematically well-posed inference problem. We show that the heuristics redefined under our probabilistic framework generally perform better than conventional heuristics under this evaluation. We conclude with a discussion on the possible applications of this framework.

Return to previous page