Effects of feedback lag variability on the choice of an automated diagnostic aid: A preliminary predictive model

This paper outlines the Signal Detection Model of Choice (SDMC) that predicts operators' use of automated diagnostic aids based on system feedback lag characteristics. Specifically, the SDMC predicts that users will prefer an aid having variable rather than fixed feedback lag characteristics when the value of correct decisions (hits and correct rejections) are large relative to the negative consequences of incorrect decisions (misses and false alarms). A preference reversal is predicted by the model when the positive consequences of correct decisions are small relative to the negative consequences of wrong decisions. Furthermore, this pattern of choice is predicted even when aids are not equally reliable. Theoretical explanations of the possible cognitive processes that underlie the model, including temporal and reliability distortion theories are presented. Results from recent laboratory experiments that provide preliminary support for the model's predictions and theoretical assumptions are summarized. The bridge that this new model provides between traditional behavioural models of choice and cognitive models of decision-making are also discussed. The hope is that the SDMC will provide fruitful avenues of inquiry into the relatively unexplored effects that automation feedback lag has on choice and decision making processes.