Investigating the truthfulness of autobiographical events through mouse dynamics

This study examines the kinematic characteristics of mouse movements as a means to assess the truthfulness of reported autobiographical memories. Ninety participants answered double-choice questions about an autobiographical event using a computer mouse. To induce cognitive load in deceptive responses, complex questions were introduced. Participants were divided into three groups: truthful responders (n = 30), those instructed to fabricate an entirely fictional holiday (n = 30), and those asked to falsify specific details of a real holiday (n = 30). Temporal and spatial features of mouse trajectories were recorded and analysed. Findings indicate that deceptive responses were associated with slower and more erratic mouse movements compared to truthful ones. Furthermore, machine learning models classified deceptive versus truthful responses with an average accuracy of 75% (for liars reporting completely faked holidays) and 80% (for liars providing true holidays with false details). Notably, participants who fabricated an entire event exhibited different movement patterns than those who altered specific details, suggesting that fabricating an entirely false memory may be cognitively less demanding than modifying real details. These findings provide novel insights into cognitive processes underlying deception and highlight the potential of kinematic analysis in lie detection.