Artificial interrogators will work if we believe there’s a human behind the scenes
Fri 22 Apr 2016

A study from the University of Twente in the Netherlands shows that using avatars in interrogation situations can be effective, but only if the subjects believe that the avatar is controlled by a human.
A reliable, efficient system for lie detection under interrogation would be an important asset for government bodies and investigation authorities. Avatars could be used to promote the efficiency of interrogation, removing the human element to increase standardization of questioning. But is using avatars effective in creating the human lie response in the subjects?
A group of psychologists from the Institute for Innovation and Governance Studies at the University of Twente created a study in which groups of subjects were interrogated by an avatar. Some subjects were instructed to lie, others to tell the truth.
The researchers used skin conductance to measure perspiration, considered a fairly reliable index of whether someone is lying. Of the subjects who lied under interrogation, researchers found that if they believed they were lying to an avatar that was controlled by a computer, there was no measurable difference in skin conductance.
But if the subjects believed they were lying to an avatar controlled by a human, there was a clear, measurable difference in skin conductance. In an interrogation situation, an avatar that subjects believe is human-controlled can be effective at exposing falsehoods.
The researchers used a study group of 79 subjects who believed they were part of a recruitment and selection process. In this study, the avatar was controlled by a human, but this information was not given to the subjects – they were left to assume either that it was controlled by a human or by a computer. One of the lead researchers, Elze Ufkes, said that they are planning a follow-up study to further explore the subject’s perception of the avatar. “During the next step in this study we will manipulate our subjects by convincing some of them that they are communicating with a computer or, in other cases, with a human-controlled system.”