An investigation into changes in decision criteria and other details of a decision-making task

Abstract

A method is described for constructing a decision-making task which satisfies the formal assumptions of signal-detection theory. Two experiments using this task are described. It is shown that Ss did not maintain a constant likelihood-ratio criterion and, hence, that the subjective d’ was always less than the true value, even though Ss knew what that value was. Introducing confidence judgments did not significantly affect d’. It is suggested that the method described is a useful one for examining decision responses.