In a 2004 study out of the University of Portsmouth by Samantha Mann, Aldert Vrij and Ray Bull it was found that police officers were sixty-five percent accurate in detecting lies when they watched the proceedings of an interrogation. This success rate is significantly higher than that which could arise by chance alone and also shows that familiarity with the subjects can have a role in increasing accuracy. Most research thus far has used college students, but this shows that police who frequently deal with suspects might have an advantage reading them over reading others. By a similarly notion, this advantage would theoretically be non-existent for police officers in a business meeting or with regard to a salesperson on a car lot, unless they had particular experience with such matters. This study does tell us that familiarity with the subject and the context can help us in detecting lies.

Police manuals give the impression that officers who interview suspects often, are good lie detectors, despite of course the vast research that says otherwise. When the researchers qualified their observations however, they found some surprising findings. Officers who named visual cues such as those mentioned in Inbau’s research, mentioned previously, which forms part of the manual on lie detection for police, such as gaze aversion, unnatural changes in posture, self touching, mouth and eye covering were less likely to be accurate in reading others. In fact, these cues proved counterproductive. Specifically, female participants who claimed to use Inbau’s cues most often where poorer at detecting truths, than the males who did not. In particular, gaze aversion was unhelpful and in fact distracting when analyzing for truth. So despite the moderate success of officers at detecting lies, there still remains severe shortcomings because it was not necessarily due to observations of body language or other anything else that could be described, catalogued, and hence put to use. If an inherent skill amounts to a sixty-five percent success, but one can’t describe that skill in a way that makes it useful to other people, then it simply appears like a hunch. Hunches are not reliable, nor do they meet the scientific principle of reproducibility or have predictive (useful) qualities.