HWNI member Ming Hsu, professor in the Haas School of Business, and collaborators have developed a way to distinguish lying from selfishness in functional magnetic resonance imaging (fMRI) brain scans. In a paper published in the Proceedings of the National Academy of Sciences, the team described how they used machine learning to develop a model that could predict when study participants were lying 79% of the time, from their brain scans alone, but the model also predicted when people were engaged in selfish behavior without lying.
The researchers then refined their algorithm in a way that correctly predicted lying 70% of the time, without predicting selfishness. The study shows that lying can be distinguished from other confounding behaviors in brain scans, and this type of approach could also help researchers who are investigating the neural basis of other complex cognitive processes. Read more from Berkeley Haas News.
