We use the term Interrogative Polygraphy to refer to a body of techniques that is popularly known as “lie detection.” and commonly referred to by its practitioners as “polygraphy.” The popular term is inaccurate and misleading, because none of the techniques actually detect lies (see Ekman, 1985). The term polygraphy is too broad if left unqualified. All psychophysiologists use polygraphs. This paper is concerned with those psychophysiologists who are using polygraphs as aids in interrogations, and that is why the term interrogative polygraphy is deemed appropriate.
The Truth Will Out: Interrogative Polygraphy (“Lie Detection”) With Event-Related Brain Potentials
Version of Record online: 30 JAN 2007
Volume 28, Issue 5, pages 531–547, September 1991
How to Cite
Farwell, L. A. and Donchin, E. (1991), The Truth Will Out: Interrogative Polygraphy (“Lie Detection”) With Event-Related Brain Potentials. Psychophysiology, 28: 531–547. doi: 10.1111/j.1469-8986.1991.tb01990.x
- Issue online: 30 JAN 2007
- Version of Record online: 30 JAN 2007
- (Manuscript received May 24, 1990; accepted for publication October 20, 1990)
- Event-related potentials;
- Lie detection;
- Guilty Knowledge Test;
The feasibility of using Event Related Brain Potentials (ERPs) in Interrogative Polygraphy (“Lie Detection”) was tested by examining the effectiveness of the Guilty Knowledge Test designed by Farwell and Donchin (1986, 1988). The subject is assigned an arbitrary task requiring discrimination between experimenter-designated targets and other, irrelevant stimuli. A group of diagnostic items (“probes”), which to the unwitting are indistinguishable from the irrelevant items, are embedded among the irrelevant. For subjects who possess “guilty knowledge” these probes are distinct from the irrelevants and are likely to elicit a P300, thus revealing their possessing the special knowledge that allows them to differentiate the probes from the irrelevants. We report two experiments in which this paradigm was tested. In Experiment 1, 20 subjects participated in one of two mock espionage scenarios and were tested for their knowledge of both scenarios. All stimuli consisted of short phrases presented for 300 ms each at an interstimulus interval of 1550 ms. A set of items were designated as “targets” and appeared on 17% of the trials. Probes related to the scenarios also appeared on 17% of the trials. The rest of the items were irrelevants. Subjects responded by pressing one switch following targets, and the other following irrelevants (and, of course, probes). ERPs were recorded from Fz, Cz, and Pz. As predicted, targets elicited large P300s in all subjects. Probes associated with a given scenario elicited a P300 in subjects who participated in that scenario. A bootstrapping method was used to assess the quality of the decision for each subject. The algorithm declared the decision indeterminate in 12.5% of the cases. In all other cases a decision was made. There were no false positives and no false negatives: whenever a determination was made it was accurate. The second experiment was virtually identical to the first, with identical results, except that this time 4 subjects were tested, each of which had a minor brush with the law. Subjects were tested to determine whether they possessed information on their own “crimes.” The results were as expected; the Guilty Knowledge Test determined correctly which subject possessed which information. The implications of these data both for the practice of Interrogative Polygraphy and the interpretation of the P300 are discussed.