Perceptual Orientation and Legitimacy of Interpersonal Cues to Endangerment

Abstract

The thesis developed a novel, reliable, and useful assessment tool to evaluate the perception of interpersonal threat in different populations, including those with professional training (e.g. the police), personal training (e.g. martial artists), and those without. Across seven studies, participants perception of threat was assessed using naturalistic video stimuli, categorised by their depiction of interpersonal interactions that did or did not result in an aggression. All videos categorised as violent were cropped before any violence occurred to allow participants to predict the outcome. To refine the tool, the number of videos were altered, maintaining a 1:1 ratio of videos categorised as violent or non-violent. The dependent variables across studies included a prediction of the likelihood of violence occurring (Threat Perception Index, TPI), the presence of visual cues such as invading personal space, or the aggressor clenching their hands, and the level of concern these cues elicited (Objective Outcome Discrimination, OOD).
Studies one and two explored how threat perception develops over time in decontextualised video segments, and if the frequency of behavioural cues also changed over time. This provided insight into a more refined video selection for future studies for the assessment tool. Threat perception was idiosyncratic and non-linear, as were behavioural cues. However, perception did not differ based on demographics.
Studies 3 and 4 explored the reliability and validity of the measures. Both the Threat Perception Index and Objective Outcome Discrimination indicated high reliability and construct validity. This was echoed in later studies within the thesis. Several cues were indicated to occur more frequently in violent scenarios. These were the aggressor tensing up their whole body, clenching their hands, tensing their jaw muscle, stretching the arm or shoulder, rolling their head, staring towards the target, forming an angry expression, exaggerated hand gestures, pacing, assuming a boxing stance, and invading personal space. Invading personal space, assuming a boxing stance, forming an angry expression and staring towards were found be the most concerning cues; while the aggressor having their hands in their pockets, blinking rapidly, staring away, and having their hands on their hips were found to be least concerning. Study 3 included forming a discrimination measure to assess video selection criteria.
Study 5 applied the tool to a real word setting exploring differences between those with professional, personal or no training. Results demonstrated that the police sample perceived more threat only in the violent videos compared to non-trained individuals but were not more accurate.
Study 6 replicated findings from previous studies, providing additional evidence. All cues were indicated to be more present in violent videos. Again, the aggressor invading personal space, assuming a boxer’s stance, forming an angry expression, and staring towards the target were the most concerning cues. Exaggerated hand gestures also became a cue of significant concern.
2
Study 7 compared threat perception between the Spanish military and an untrained UK sample. The Spanish military perceived more threat in non-violent videos only, compared to the untrained UK sample.
The results indicate the subjective nature of threat and how certain behavioural cues to be of particular importance, with invasion of personal space found to be the most concerning cue across studies. There are significant implications of different types of training, which has clear ramifications for our understanding of threat in various settings. This novel assessment tool can provide fundamental guidance regarding the training of individuals to accurately assess potentially threatening situations and optimise decision-making with regards to the use of force, to aid public safety.

Date of Award	1 Apr 2026
Original language	English
Awarding Institution	Bangor University
Supervisor	Lara Maister (Supervisor) & Paloma Marí-Beffa (Supervisor)