the polygraph’s validity. Polygraph theories have been largely silent about these possibilities, and empirical polygraph research has made little effort to assess their influence on polygraph readings or interpretation.

Most alternative technologies for the psychophysiological detection of deception that are being pursued (see U.S. Department of Defense, 2000; U.S. General Accounting Office, 2001) rest on similar theoretical foundations and are subject to the same theoretical limitations. This statement holds both for measures of brain function and for peripheral measures of autonomic activity. The underlying assumption remains that someone who is trying to hide something will respond differently (i.e., show “leakage,” physiological arousal, or orienting responses to specific questions) than someone who is not trying to hide something. The objective of the new approaches, therefore, continues to be to measure a naturally occurring physiological response or profile of responses that not only differentiates known deceptive from truthful answers but also allows accurate classification of answers as deceptive or truthful. Improvements have been and continue to be made in the design of transducers, amplifiers, data recording, and display techniques, and in the standardization of procedures and data reduction. Data interpretation, however, still depends on the validity of the assumption that relevant, in contrast to comparison, questions are more evocative to those giving deceptive answers and equally or less evocative to those giving true answers.

Screening uses of polygraph testing raise particular theoretical issues because when the examiner does not have a specific event to ask about, the relevant questions must be generic. If a comparison question testing format can meet the challenge of calibrating questions to elicit the desired level of response in a specific-incident test, it does not follow that the same format will meet the challenge in a screening application because the relevant questions do not refer to a specific event. It is reasonable to hypothesize that autonomic reactions are more intense, at least for guilty individuals, when a target event is described concretely than when it is merely implied by mention of a generic category of events. Nothing in current knowledge of psychophysiology gives confidence that a test format will work at the same level of accuracy in a screening setting that requires generic questioning as it does in a specific-incident application.

The theory of comparison question polygraph techniques as currently used for screening can be summarized as follows:

• An examinee will respond differently when trying to hide something (i.e., show leakage or greater physiological arousal or orienting responses to relevant questions) than when not trying to hide something.⁴

• Those who have nothing to hide will be less reactive to key (rel-