A rule-based lens model

Abstract

The purpose of this paper is to introduce a novel approach, called the rule-based lens model (RLM) to model human judgment of a probabilistic criterion. Our method is motivated by a shortcoming of an existing linear-additive model of judgment based on the lens model equation (LME) to adequately represent all rule-based relationships. Through the use of a simple example, we demonstrate the shortcoming of the additive model and set the context for our generalized rule-based formulation of Brunswik's conceptual lens model. To investigate the behavior of our model and the relationship to the traditional lens model based on the LME, we simulate human judgments and criterion values in a “drosophila” domain where the parameters of the problem in terms of the number of cues, organizing principle of the criterion, organizing principle of the judge and the extent of uncertainty within the system can be systematically varied. Our efforts represent a first step toward the formulation of a generalized lens model framework. Relevance to industry The findings of the proposed research would provide theoretical basis toward the design of decision-aiding and decision-training systems that are adapted to human decision strategies.