Decision making during paced arrival of probabilistic information.

Abstract Four exploratory experiments are described in which the basic assumptions of the Edwards-Wald model for decision making in probabilistic sequential tasks are tested. The assumptions are (1) continuous revision of the likelihood ratio on the basis of incoming data and (2) a fixed decision criterion on the basis of costs and pay-offs. The results suggest that the decision criterion shifts from rather strict to quite risky as clear evidence is postponed, so that the criterion is certainly not fixed. The findings were not contrary to the idea of revision of the likelihood ratio.