Deception + Intent

Deception and its detection have been a source of fascination for centuries, with literally thousands of publications on the topic testifying to its importance in the conduct of human affairs. Researchers and practitioners have pursued a host of detection techniques, from divining trustworthiness from the shape of one's head and ears, to the use of physiologically-based instruments such as the polygraph and vocal stress analyzer, to reliance on behavioral cues as potentially telltale cues to deception. However, no single physiological or behavioral cue has proven to be accurate 100% of the time.

All these efforts notwithstanding, deception detection accuracy has typically hovered around 50-50, or chance, even among trained professionals. The most promising avenues for distinguishing deceit from truth are in tools and techniques that utilize configurations of cues. Moreover, because deception indicators are subtle, dynamic, and transitory, they often elude humans' conscious awareness. If computer-assisted detection tools can be developed to augment human detection capabilities by discerning features of language and nonverbal behavior and by tracking these highly elusive and fleeting cues over a course of time, accuracy in discerning both truthful and deceptive information and communications should be vastly improved. Coupled with improved human interviewing and interrogation techniques that are aided by multimedia training and real-time feedback, these tools can greatly enhance deception detection.

AutoID - Automated Intent Determination

This project delves into the automated identification of non-verbal gestures to determine a subject's intent. This research grant is funded by the U.S. Department of Homeland Security.


Speech Act Profiling

Speech act profiling is a method of analyzing and visualizing conversations according to how participants go about conversing rather than what it is they talk about. It is useful in detecting deception as deceivers often adopt telltale communication patterns.


C-BAS (C# Behavioral Annotation System)

C-BAS is a template driven application written to assist researchers with behavioral coding.