The Recognition of Acted Interpersonal Stance in Police Interrogations

This research aims at finding how suspects in police interrogations express their interpersonal stance -in terms of T.Leary's interpersonal circumplex- through body postures and facial expressions and how this can be simulated by virtual humans. Therefore, four types of stances were acted by eight actors. To see if the resulted postures are valid, short recordings were shown online in a survey to subjects who were asked to describe them by a selection of a number of adjectives. Results of this annotation task show that some stance types are better recognized than others. Validity (recognizing the intended stance) and inter-rater agreement do not always go hand in hand. The body postures and facial expressions of the best recognized fragments are annotated so they can be implemented in the artificial agent. The results of this study are used in a serious game for police interrogation training where the role of the suspect is played by an artificial embodied conversational agent.