Three models of sequential effects in psychophysica l tasks are defined and experimental results described. These appear most consistent with a model in which the momentary value of the criterion is modified by memory traces, an independent trace being retained for each relevant past event. On this basis, a theory of criterion setting is developed: A long-term process determines an initial value for the criterion, and two short-term processes adjust the criterion (a) to match current changes in the expected probabilities of signals and (b) to maximize the information transmitted by the subject's responses. The theory is applied to results in the literature. Although signal detection theory (SDT; Green & Swets, 1966), which is now widely applied, includes the hypothesis of a decision criterion and puts forward normative prescriptions for its value, the problem of how a subject maintains his criterion at the appropriate value and adjusts that value to take account of relevant intercurrent events has largely been ignored. It is regrettable that criterion setting should generally be treated as a given but unexamined fact, because its proper understanding may clarify important psychological aspects of decision in all areas and may help to explain some still unresolved problems such as the causation of sequential effects. The object of the present article is to present a theory of criterion setting developed by the first author. As this theory arose from a consideration of some features of sequential dependencies, we begin with a brief discussion of the latter. It is a familiar but still unexplained observation that the outcomes of successive psychophysical decisions are related (Verplanck,
No takes yet. Share an insight, caveat, or question.
Treisman et al. (1984) studied this question.
Synapse has enriched one closely related paper. Consider it for comparative context: