User behavior is a critical element in the success or failure of computer security protections. The field of Human Security Informatics (HSI) combines security informatics and human-computer interaction design to learn how the design of a human-computer interface can affect the security of a computer system. This research project is contributing to the scientific foundations of HSI by modeling how multitasking users behave when making security-critical decisions. In particular, the researchers are modeling user behavior when the users are engaged in typical PC-based mobile messaging with security concerns such as phishing or spam. The project is evaluating how well the models capture the impact of incentives and interventions on user security behaviors. This project extends the cognitive modeling (CogM) architecture to characterize and improve user security decision-making and behaviors. Focusing on cognitive constructs in the ACT-R and Soar architectures, it models the multi-tasking application and security activities with varying cognitive traits and security constraints, through representations of productions and information chunks, as well as their utility and activation calculations. An analytic user model not only describes a problem in making a security decision, but also can explain why and how it happens for incentive and intervention selection. Moreover, CogM models and empirical user testing comparatively study common and advanced users in typical messaging applications, regarding security mistakes and efficiency in task completion. This project is focused on establishing the principles for analytically modeling user cyber behaviors and bridging the gap from understanding security behaviors to effectively improving security performance.
EAGER: Collaborative: Computational Cognitive Modeling of User Security and Incentive Behaviors