Examining the Dual Process Model of Decision Making in Visual Threat Detection and Human-Automation Interaction

This study examined the degree to which emotions and/or rational thinking should be emphasized to improve performance in target detection in complex tasks. Specifically, we evaluated the dual-process model of decision making to determine whether emotion or cognition (or a combination of the two) is more powerful in influencing decisions. Additionally, we considered how decisions are influenced with the addition of an automated agent to assist human decision making. Participants (n = 120) were randomly assigned to one of three framing manipulations, each with positive and negative incentives: the analytical frame emphasized numerical gains or losses; the affective frame presented gains and losses as "lives saved" and "lives lost"; the comparative frame emphasized better or worse performance relative to one's peers. Performance was assessed both with and without the assistance of an automated decision aid in a simulated airline luggage screening task. Performance was measured by hit rates and false alarm rates; human-automation interaction was assessed via compliance, reliance and system trust. Results revealed that participants in the analytical frame were able to balance their hits and false alarms whereas the affective and comparative frames were unable to achieve this balance. Although the affective frame generated several hits, they also generated a large proportion of false alarms; they were strongly influenced by the negative incentive of "lives lost" and complied frequently with automation, even in instances when the automation was possibly incorrect. Results imply that in contexts such as security screening, emphasizing analytical thinking enables operators to perform effectively and utilize automation appropriately. However, emphasis on affective thought hinders the ability to weigh outcome probabilities and leads to inappropriate automation dependence. The findings have implications for both operator training and automation design in complex tasks.