Researchers at Northeastern University have used Nexi, an "expressive robot" developed at MIT, to conduct an innovative new study of the way humans interact. Seeking to analyze the effect of body language and facial expressions on trust, the scientists asked participants to engage in a conversation with Nexi — when speaking to some people, the robot would remain still, while with others it would mimic what are thought to be tell-tale signs of lying, such as leaning back or touching its face and hands. The volunteers were then asked to play a real-money gambling game with an invisible player who they believed to be Nexi, with those who had experienced the "deceitful" body language playing more cautiously and failing to trust their counterparts.

A variant of the classic prisoner's dilemma, the game involved judging the likelihood of betrayal, with participants able to share a modest prize or stab each other in the back in an attempt to win more. According to lead researcher David DeSteno, the main problem with the experiment was "iden­ti­fying the exact cues that matter" — DeSteno and his team analyzed video footage of conversations between humans, paired with the results of similar games, in order to get Nexi's expressions (demonstrated in the video below) precisely right. The resulting research is due to be published in the journal Psychological Science.