Machine learning model outperforms human judgement
Our computational trust model is capable of predicting—above human accuracy—the degree of trust a person has toward a stranger by observing the nonverbal behaviors expressed in their social interaction. We used machine learning algorithms, specifically hidden Markov models (HMMs), to model the temporal relationship between specific nonverbal behaviors. By interpreting its resulting learned structure, we discovered that the sequence of low and high trusting behaviors a person emits provides further information of their trust orientation toward their partner. These discoveries shaped the feature engineering process that enabled a support vector machine (SVM) model to achieve a prediction performance more accurate than human judgment.