The claim that people tend to look in one direction - up and to the left - when they are being honest, and a different direction - up and to the right - when they are lying, turns out not to be true.
The suggestion that gaze-direction is a dishonesty giveaway is one of the tenets of "neuro-linguistic programming" (NLP), a collection of psychological techniques that, among other things, aim to enhance an individual's communication skills.
Practitioners are taught that, when right-handed people are visualising made-up events in their mind's eye, they tend to look upwards and right. Recalling a real memory, on the other hand, is claimed to be signalled by a look to the upper left.
These claims are based on work in the 1960s by the psychologist Paul Bakan, who suggested that activity one part of the brain could sometimes spill-over into an adjacent region called the lateral eye field, which coordinates eye movements, provoking a look in one direction or the other.
But subsequent studies on this work painted a shaky picture, so University of Hertfordshire researcher Richard Wiseman and his colleagues have carried out a series of 3 "blinded" experiments to comprehensively test whether individuals really do "look shifty", or at least away from the left, when they lie.
In the first experiment, the team asked student volunteers to take a mobile phone and either conceal it about their person, or put it in a drawer and then convince an interviewer, regardless of what they'd done, that the phone was in the drawer.
As they related either the lie or truth, they were filmed. Their eye-movements were then logged, which revealed no bias in gaze direction, regardless of whether they were telling the truth or not.
In a second experiment, to determine whether training in NLP techniques somehow makes individuals more sensitive to picking up porkies, the team asked a second group of volunteers who had been previously primed about what to look out for, to review the films from the first experiment.
Although the raters who received the NLP training reported greater confidence in making judgements about lies or truth-telling, they were actually no more accurate than a group of untrained controls.
Finally, to test real-life examples of high-stakes lying, to see whether this might make a difference, the researchers analysed 52 video sequences of relatives making public pleas for the safe return of a missing relative. But in half of these cases, the person was subsequently proved to have been lying. These individuals showed no biases in their gaze directions either.
According to Wiseman and his colleagues, "These results provide considerable grounds to be skeptical of the notion that proposed patterns of eye-movements provide a reliable indicator of lying. As such, it would seem irresponsible for such practitioners to continue to encourge people to make important decisions on the basis of such claims."