Detecting Deception Through Non-Verbal Behaviour

Mapala, Tessa and Warmelink, Lara (2021) Detecting Deception Through Non-Verbal Behaviour. PhD thesis, Lancaster University.

Text (2021MapalaPhD)
2021MapalaPhD.pdf - Published Version
Available under License Creative Commons Attribution-NonCommercial.
Download (1MB)

Abstract

The security protocols used in airport security checkpoints primarily aim to detect prohibited items, as well as the detection of malicious intent and associated deception to thwart any threats. However, some of the security protocols that are used are not substantiated by scientifically validated cues of deception. Instead, some protocols, such as the Screening of Passengers by Observation Techniques (SPOT) program, have been developed based on anecdotal evidence and invalid cues of deception. As such, the use of these protocols has received a lot of criticism in recent years from government agencies, civil rights organisations and academia. These security protocols rely on security personnel’s ability to infer intent from non-verbal behaviour, yet the literature suggests that the relationship between non-verbal cues and deception is unreliable and that people are poor at detecting deception. To improve upon our understanding of the validity of these protocols, this thesis used virtual reality to replicate a security checkpoint to explore whether there were valid cues of deception, specifically in an airport context. People’s ability to identify whether others were behaving deceptively was assessed, as well as the factors that may be informing decision-making. Chapter Four of this thesis found that the non-verbal cues of interest, which were segment displacement, centre of mass displacement, cadence, step length and speed were not significantly different between honest and deceptive people. A verbal measure, response latency, was found to only distinguish between honest people and those who were deceptive about a future intention, but not those who were deceptive about having a prohibited item. In light of the use of non-verbal measures in practice despite the lack of scientific support, Chapters Five to Seven aimed to gain a greater insight into people’s deception detection capabilities. The findings from Chapters Five to Seven reflected that the ability to detect deception from non-verbal behaviour was no better than guessing. Specifically, Chapter Five found that the accuracy of detecting deception was no different from chance levels. Six themes emerged as the factors that were used to inform decision-making. The themes were physical appearance, disposition, walking behaviour, body positioning, looking behaviour and upper limb movement, though a qualitative analysis revealed that there were subjective interpretations of how the themes mapped onto deception. Chapter Six introduced two techniques of information reduction to assess whether accuracy could be improved above chance levels by lessening the impact of biasing factors. Neither technique resulted in accuracy above chance levels. In Chapter Seven, eye tracking was utilised to assess the gaze patterns associated with the detection of deception. People looked at the legs more than other areas of the body prior to decision-making, though only looking at the left arm and hand were linked with accuracy. Detection accuracy was poor overall, though looking at the left arm was linked with reduced accuracy, whilst looking at the left hand was linked with increased accuracy. Overall, this thesis showed that the non-verbal cues that were assessed could not distinguish between honest and deceptive people. In the absence of valid cues, observers were not able to identify deception at a rate above chance even with the reduction of potentially biasing factors. The results of this thesis reinforce the idea that incorporating nonverbal measures into threat/deception detection protocols may not be warranted because of the dubious nature of their reliability and validity, as well as the poor deception identification capabilities when relying on non-verbal behaviour.