Parallel belief updating in sequential diagnostic reasoning

In sequential diagnostic reasoning the goal is to determine the most likely cause for a number of sequentially observed effects. Potential hypotheses are narrowed down by integrating the cumulating observed evidence leading to the selection of one among several hypotheses. In the reported diagnostic reasoning experiment, thirty-eight participants were tested with quasi-medical problems consisting of four sequentially presented symptoms with four candidate diagnostic hypotheses. We used ambiguous sequences that could be equally caused by two chemicals to investigate possible order effects and explicitly highlighted alternative hypotheses by using a stepwise rating procedure that also enabled us to compare participants' ratings with belief updating in a Bayes net. Even though alternatives were explicitly highlighted, participants were biased towards the initial hypothesis in a pair of equally supported hypotheses. We conclude that ambiguous symptom sets and non-diagnostic symptoms invite biased symptom processing and can produce primacy effects even in a step-by-step procedure.

(PDF) Parallel Belief Updating in Sequential Diagnostic Reasoning. Available from: https://www.researchgate.net/publication/265301970_Parallel_Belief_Updating_in_Sequential_Diagnostic_Reasoning [accessed Sep 24 2018].