Eye Gaze-driven Prediction of Cognitive Differences during Graphical Password Composition

Abstract

Evidence suggests that individual cognitive differences affect users' memorability, visual behavior, and graphical passwords' security. Such knowledge denotes the added value of personalizing graphical password schemes towards the unique cognitive characteristics of the users. However, real-time and accurate cognition-based predictive user models are necessary to reach such a breakthrough. In this paper, we present the results of such an attempt, where an in-lab eye-tracking study was conducted with 36 participants who completed a recall-based graphical password composition task. We adopted a credible cognitive style theory, and investigated a variety of eye-tracking metrics to predict participants' cognitive styles. Results' analysis reveals that inferring individual cognitive differences in real-time during graphical password composition is feasible within a few seconds and that specific eye-tracking metrics correlate stronger with certain cognitive style groups. The findings further support the vision of incorporating real-time adaptive mechanisms in graphical password schemes for the benefit of service providers and end-users.