This paper addresses the evaluation of the performance of the decision support system that utilizes face and facial expression biometrics. The evaluation criteria include risk of error and related reliability of decision, as well as their contribution to the changes in the perceived operator's trust in the decision. The relevant applications include human behavior monitoring and stress detection in individuals and teams, and in situational awareness system. Using an available database of cross-spectral videos of faces and facial expressions, we conducted a series of experiments that demonstrate the phenomenon of biases in biometrics that affect the evaluated measures of the performance in human-machine systems.