A Method for Applying Statistical Decision Theory to System Task Analysis 1

Abstract

For system tasks involving signal detection, a method is presented by which statistical decision theory may be used to derive limiting conditions for adequate operator performance from results obtained in an abstracted laboratory task. The limiting conditions may then serve as a basis for making design decisions regarding functions allocation and for specifying operating rules. The method is illustrated with a collision-prediction task in which previous research suggests that performance depends on the ability to detect a nonzero rate of change in the relative bearing between the two objects moving on converging paths.