Exploring a Strategic Metacognitive Approach to Detecting Deception

Abstract

This thesis explores detecting guilty knowledge, starting with an introduction to established approaches: arousal-based, behavioural control and cognitive load. It then introduces and assesses the potential of a novel strategic metacognitive approach, which introduces cognitive tasks specifically designed to invite those attempting to conceal guilty knowledge, to use response strategies which diverge from those without guilty knowledge, thus resulting in strategic cues which can identify those with guilty knowledge. Chapters 2, 3, and 4 investigate the serial position effect (SPE) as a means of detecting guilty knowledge. Chapter 2's experiment with a neutral video found that uninformed and informed liars (liars told how truthful people would likely respond) deviated from typical truthful SPE patterns when concealing knowledge, achieving 87.3% combined detection accuracy. Chapter 3 replicated this using a crime-related video, again finding uninformed liars showing deviations in response patterns, though informed liars were better at mimicking the response patterns of truth-tellers. This experiment achieved a combined detection accuracy of 86.3%. Chapter 4 extended this to genuinely experienced events, finding partial support for uninformed liars deviating from SPE, but not informed liars, with an overall combined detection accuracy of 78.75%. Chapter 5 examines the Stroop task for detecting guilty knowledge. It found that uninformed liars had significantly longer response times for guilty words compared to non-guilty words when asked to conceal information. However, informed liars were able to successfully mimic truthful participants' response times, showing no significant difference. The concluding chapter (6) evaluates the overall findings, considering the practical implications for law enforcement and security services. It acknowledges that while the empirical evidence for the new approaches is promising, significant further research is needed before practical application is possible. The discussion also highlights the limitation that informed liars were able to successfully use countermeasures to appear truthful. However, it is suggested that the use of Artificial Intelligence and Machine Learning could examine for more subtle strategic differences which are more robust against countermeasures.

Date of Award	7 Dec 2025
Original language	English
Awarding Institution	University of Winchester
Supervisor	Gary Lancaster (Supervisor) & Rhiannon Jones (Supervisor)