Subscribe now

Humans

We’re discovering new ways to detect if someone is lying

Brain Scanner is Simon Oxenham's weekly column that sifts the pseudoscience from the neuroscience

By Simon Oxenham

25 July 2016

Man points at white board while looking questioningly at other man

The truth is in his eyes

Thomas Barwick/Getty

Can you tell if someone is lying? Our ability to spot a lie is only just better than guessing with the flip of a coin. But, surprisingly, it’s easier to tell whether a person is fibbing if they are wearing a veil, suggests a fresh study.

The experiment was devised by researchers at the University of Ontario Institute of Technology in Canada, and the University of Amsterdam in the Netherlands. They filmed two videos of a woman watching a stranger’s bag, one of which showed the woman stealing items from it. They then played one or the other video separately to female volunteers designated as “witnesses”.

After this, the witnesses were asked to “testify” to a video camera that they had not seen the woman steal anything – meaning that half ended up telling the truth and half had to lie. They were each motivated by a prize of $50 if they could convince people that they were being truthful.

But there was a twist. A third of the volunteers were asked to wear a black niqab – which hides the whole face apart from the eyes – and a third a black hijab, which surrounds the face without covering any of it. The other third did not wear a veil.

Veil of truth

When the researchers then played these video testimonies to other volunteers, they found that they were much better at telling whether a woman was lying or not if she was wearing a hijab or niqab.

This could be because in those cases, the viewer was less likely to rely on appearance. In the case of the niqab, the veil may have focused attention on the women’s eyes, helping to block out distracting information. Liars tended to make less eye contact with the camera than those who told the truth, so focusing on the eyes of those wearing niqabs could help people notice this.

It also appears that niqabs caused the volunteers wearing them to reveal more verbal information than those who wore hijabs or no veil. This might suggest that the niqabs resulted in more information for those watching to base their judgements on, shifting the focus away from appearance and body language.

This finding is interesting in light of a 2013 UK court case in which a judge instructed a defendant charged with intimidating a witness to remove her veil. The judge argued that it was necessary to see her face to assess how truthful she was being – but the new findings suggest that this may not have been the case.

However, even though the lie-spotters guessed more accurately when a woman was wearing a veil, they still showed a slight bias. This might be linked to them being keen to show they weren’t prejudiced: if a woman had a hijab on, the volunteers were more likely to guess that she was telling the truth.

Look to language

The role of eyes in lie detection is complicated. There’s a popular theory that “liars look up and to the right”, but this has been debunked – there is no correlation between lying and which particular direction your eyes move. Blink frequency doesn’t seem to be related to telling a lie either. Instead, eye contact and speech patterns seem to be more promising ways to detect a fibber through observation alone.

Our lie-detection odds of just over 50:50 remain about the same when we’re reading lies on paper too. But a team led by Stephan Ludwig at the University of Westminster, UK, recently tried using automated text analysis on computers, with astounding results.

The researchers designed a model that looks for linguistic cues that could indicate a lie – for example, not using personal pronouns like “I” or “you”, which might imply someone is trying to distance themselves from what they are saying. They also made the model measure factors such as self-deprecating or flattering language.

To test it, they worked with a technology firm that often needs to detect deception in business communications. In the trial, the model analysed more than 8000 emails bidding for awards based on a company’s performance. Using text-mining software, the model analysed these bids for fibs, with the results then compared against those of an independent investigation by the company’s account managers.

The investigation verified that the model correctly classified 70 per cent of reward requests as either truthful or deceitful. A person reading those emails would have done far worse.

Machine prosecutors

It’s important to remember that we’ve attempted to use technology to identify lies before, and have been wrong. Techniques such as the polygraph test are now widely discredited and seen as pseudoscience, but this has not stopped their continued use by police forces in the UK and around the world.

However, when research, verification and validation are used to test and improve lie-detection technology, such techniques may pave the way to a world in which lies are not detected by people, but by machines. The Westminster researchers suggest that their technique may be able to detect deception in everything from visa applications to dating profiles.

Perhaps in the distant future we’ll be tried by computers, before being sentenced by judges and juries and women will be allowed to wear veils in court.

Read more: Lie-detecting algorithm spots fibbing faces better than humans

Topics:

Sign up to our weekly newsletter

Receive a weekly dose of discovery in your inbox! We'll also keep you up to date with New Scientist events and special offers.

Sign up