Prior expectations in cross-modality matching

A mathematical model is proposed to accommodate both the 'regression' effect in cross-modality matching and the central tendency of judgment. The basic idea is that the subject has two sources of data, (a) the stimulus to be judged and (b) previous stimuli in the experiment which afford some idea what magnitude of stimulus to expect. The minimum variance estimate of stimulus magnitude is a weighted average of these two. That weighted average biases judgments towards the prior expected value; at the same time, the variability of the judgments is reduced below the value which would attach to judgments based on datum (a) alone. This basic idea is packaged in a process model which purports to describe the minutiae involved in making a cross-modality match. The process model adds this one substantive assertion to the mathematical idea: that prior expectations cannot be disregarded (notwithstanding that they are irrelevant to the judgment) because the prior experiences from which they are derived are an integral component of the subject making the judgment. (C) 1999 Elsevier Science B.V. All rights reserved.