New technology for lie detection, based on facial analysis, is soon to be tested at a British airport. Scientists at Bradford and Aberystwyth universities unveiled their “real-time dynamic passive profiling” system, which will analyse expressions, eye movements, blood flow and temperature changes at the British Science Festival in Bradford on Monday.
It is being developed in collaboration with the Home Office and HM Revenue and Customs, and with funding from a £500,000 Engineering and Physical Sciences Research Council grant. Qinetiq, the defence security company is also involved.
Hassan Ugail, professor of visual computing at Bradford University, said it was “a complete step change” from the traditional polygraph lie detector, which requires the subject to be wired up to a range of physiological detectors.
“The new non-invasive format means that those being questioned won’t have to be attached to a machine – and therefore could also provide an opportunity to find out if someone is lying without them actually knowing they are being assessed,” said Prof Ugail.
The researchers first discovered from lab experiments with 40 volunteers which facial cues are common when people lie. They include lip pressing, nose wrinkling and subtle changes in breathing. At the same time there are small changes in the temperature of the face, especially around the eyes, which can be measured by thermal imaging cameras.
The system is currently around 70 per cent accurate and the researchers said they expected soon to reach the 90 per cent accuracy level achieved by the polygraph under ideal conditions.
But the technology can never be made 100 per cent accurate – and therefore will never be appropriate to prove that someone is lying, in the absence of other evidence.
The Home Office is to carry out a field test at a UK airport, Prof Ugail said, “but we are still working with them on the details of the trial. We do not yet know the set-up, the date or the place.”
The system works best if it can first establish a facial baseline for each subject when they are feeling relaxed and have no reason to lie – and then look for changes that might accompany lying.
Prof Ugail gave a hypothetical example of an airline passenger being asked routine, unthreatening questions while checking in for a flight to the US. This baseline profile could then be transmitted to the American authorities for use on his or her arrival in the US.
The researchers said they knew of no similar system of facial analysis for lie detection in development in the US or anywhere else. Beyond security applications, they anticipate possible medical applications for detecting early stages of diseases that affect facial expression, such as Parkinson’s and Alzheimer’s. It could also be used by marketing companies to assess consumers’ reactions to new products.