Nothing but lies: improving the validity of neural predictors of deception

Abstract

Deception is a universal human behavior. Yet longstanding skepticism about the validity of measures used to understand the biological mechanisms underlying deceptive behavior has relegated such studies to the scientific periphery. Here we address these fundamental questions by applying novel machine learning methods and functional neuroimaging to signaling games capturing motivated deception in human participants. First, we develop an approach to test for the presence of confounding processes and thereby validate past skepticism by showing that much of the predictive power of neural predictors trained on deception data comes from confounding processes. Second, we show that the presence of confounding signals need not be fatal, and we improve the validity of our neural predictor via a novel machine learning procedure that identifies and removes these confounding signals. Together, these findings point to a scientific approach for studying a neglected class of behavior, with important methodological and societal implications.