Effect of Cognitive Load on Behavioural Patterns of Individuals for User Authentication

Abstract

Passwords and physical biometric approaches, such as fingerprints, are prevailing in the field of user authentication. Due to the security, privacy, and conveniency concerns, alternative techniques arose, including methods based on behavioural patterns of users. Despite mitigating many existing issues, the underline performance of such authentication approaches is not up to par with more traditional methods. One factor that potentially lowers the performance of behaviour-based systems, is the level of cognitive load a person is under. This paper presents a preliminary study on the potential of EEG data for inference of cognitive load and the effect cognitive load has on behavioural patterns of users. We gather and process EEG and behavioural data and apply a Random Forest algorithm to it to predict four different levels of cognitive load. We achieve an 89.79% and 76.00% accuracy from EEG and behavioural patterns data respectively, displaying the connection between the two.