Assessing Evidential Support in Uncertain Environments

In this chapter, we explore a classic problem in psychology: How do indi- viduals draw on their previous experience in an uncertain environment to make a prediction or diagnosis on the basis of a set of informational cues? Intuitive predictions based on multiple cues are often highly sensitive to environmental contingencies yet at the same time often exhibit pronounced and consistent biases. Our research has focused on judgments of the probability of an out- come based on binary cues (e.g., present/absent), in which the diagnostic value of those cues has been learned from direct experience in an uncertain environment. For example, in medical diagnosis, we might predict which disease a patient has based on a symptom that a patient does or does not exhibit. We have developed a model that captures both the strengths and weaknesses of intuitive judgments of this kind (the Evidential Support Accumulation Model, ESAM; Koehler, White, & Grondin, 2003). This chapter is organized as follows: After describing ESAM and show- ing how certain cues are selected for sampling, we review recent experimental results that are consistent with the basic predictions of the model. Following this, we explore how ESAM's performance varies (relative to that of a Bayesian model) based on the number of cues that are sampled in different types of multiple-cue learning environments. We then briefly review the work of other researchers in this area and offer some general conclusions.