Reprinted with permission from Waid, W.M., Orne, E.C., Cook, M.R., & Orne, M.T. Meprobamate reduces accuracy of physiological detection of deception. Science, 1981, 212, 71-73. Copyright 1981 AAAS (

Meprobamate Reduces Accuracy of Physiological Detection of Deception

William M. Waid, Emily Carota Orne, Mary R. Cook, and Martin T. Orne

Abstract. Normal male subjects attempted to deceive an experimenter recording electrodermal, respiratory, and cardiovascular activity. Those who had ingested a placebo or nothing were detected with statistically significant frequency on the basis of their phasic electrodermal responses, which clearly distinguished them from truthful suspects. That was not the case with deceptive subjects who had ingested 400 milligrams of meprobamate, nor did the examiner detect which subjects had received the drug.

The psychophysiological detection of deception depends upon the subject's having larger physiological reactions to questions associated with deception than to control questions (1). Despite both laboratory support for the basic premises underlying the procedure and its wide­spread use in police investigations and personnel screening, the validity and reliability of the polygraph test have yet to be established and remain a subject of controversy (2, 3).

We have investigated the detection of deception both as a practical problem and as a model for studying social stress (1). One question that is important for both purposes is whether a tranquilizer selectively reduces the physiological response to social stress -- in this instance, the stress of attempting to deceive. Professional polygraphers have assumed that tranquilizers might reduce the physiological response to all test questions as part of a general reduction in tonic arousal levels but that the difference in reactivity to critical and control items would be unaffected (4, 5). Clinical and pharmacological views of tranquilizers, however, suggest otherwise (6); that is, the effect of a tranquilizer might be precisely to reduce the physiological correlates of fear or anxiety concerning the critical questions.

Empirical evidence for either view is sparse (7, 8). Antianxiety drugs have been shown to reduce the electrodermal response (EDR) to some stressful stimuli, such as anticipation of shock, "emotional" words, or riding a Ferris wheel (9), but their effects on the EDR to more common and natural social stressors such as interpersonal conflict have not been investigated.

We report here a double-blind test of the effects of a tranquilizer, meprobamate, on polygraph test results (10). It was hypothesized that the EDR would accurately discriminate between truthful subjects and deceptive subjects who had not ingested a tranquilizer, as in previous studies (1); that deceptive subjects who had taken a placebo would also be accurately discriminated from truthful subjects and would not differ from deceptive subjects who had taken no drug; and that deceptive subjects who had ingested a tranquilizer would not be discriminated from truthful subjects by their EDR's but would be discriminated from no-drug and placebo-treated deceptive subjects. We also tested whether the experimenter could judge which subjects had ingested a tranquilizer, because it has been suggested (4, 5) that a tranquilizer would produce overt effects that would be readily discernible to the experienced examiner.

Each subject (11) was randomly assigned to either guilty (N = 33) or innocent (N = 11) conditions (12). Guilty subjects completed an overlearning task that ensured their sensitization to six common words and were told it would later be their task to convince a polygraph operator that they had not memorized any words. Guilty subjects were randomly assigned on a double-blind basis to one of three groups. Subjects who ingested a pill were told that they were being given a tranquilizer that would




help them avoid detection; 11 of these subjects were given 400 mg of meprobamate orally, the typical minimum clinical dose (6), and the other 11 were given a placebo. The remaining guilty subjects and the innocent subjects were given nothing.

Thirty minutes later (13) experimenter 2, who did not know the subject's assignment, attached him to the polygraph (14) and began the test, using the "guilty knowledge" technique (15). The question list consisted of 24 words, four in each of the six semantic categories, one of the four being a word the guilty subjects had memorized. The subject was asked whether any of the 24 were words he had learned. The interstimulus interval was 10 seconds. The list was prefaced with a dummy word so that the typically large initial response would not be to any of the test items. After the test the examiner completed a rating scale indicating whether the subject appeared to have taken a tranquilizer (16).

Amplitude of the EDR, smallest inspiration, and change in relative blood pressure following each stimulus were scored in millimeters, by research assistants who had no opportunity to observe the subject, and number of lies detected was scored separately for each channel (17). There are several ways to evaluate such data statistically. We report the analysis that provided the most accurate discrimination between truthful and deceptive subjects. Each guilty subject knew six critical items of information, one from each of six semantic categories, and the list was presented twice, so that each guilty subject lied 12 times. The probability of a subject's reacting more to a critical item than to any of the three control items purely by chance is .25. Thus, even an innocent subject might be expected to give as many as three responses indicating deception purely by chance. For a final classification of "deception indicated" (Table 1), we adopted a conservative criterion of five items with responses indicative of deception.

The EDR identified most guilty subjects in the no-pill and placebo groups; the differences between those groups and the innocent group are statistically significant, P < .005 and P < .005, respectively, by Fisher's exact probability test (18). Most meprobamate subjects, however, were mistakenly classified as truthful; and the discrimination between meprobamate and innocent groups was not statistically significant. The meprobamate group differed significantly from both the no-pill and the placebo groups, P < .01 and P < .04, respectively, by Fisher's exact probability test.

These results were not due to lack of electrodermal responsiveness among drug subjects; there were no differences among the four groups in the mean number of critical words (X = 10 for all subjects) that evoked a measurable EDR. Thus, although all subjects responded electrodermally throughout the test, the drug subjects did not respond more strongly when lying than when telling the truth, whereas the guilty no­pill and placebo subjects did (19). The groups did not differ in the mean amplitude of the EDR to all questions, but there was a tendency for the meprobamate subjects to give smaller EDR's as the test progressed; comparison of the mean EDR to the first presentation of the word list with the mean EDR to the second showed the effect to be most striking with the drug subjects (t = 2.37, d.f. = 10, P < .05). For the other groups the decline in amplitude was not statistically significant (t = 0.27, d.f. = 10 for the innocent subjects; t = 0.34, d.f. = 10 for no-pill subjects; and t = 0.27, d.f. = 10 for placebo subjects).

The respiratory and cardiovascular measures did not discriminate between guilty and innocent subjects. The superior detection with the EDR is consistent with other studies in which detection was scored blindly and separately for each channel (20, 21).

The examiner's judgments of whether a subject had received a tranquilizer did not approach significant accuracy. In clinical studies ratings of patients do discriminate between placebo and tranquilizer conditions (22), but patients in such studies have typically been observed over a considerable period of time under both drug and placebo conditions. The field polygraph examiner has, as in the present study, little or no previous experience with the subject to provide a baseline for the judgment. Other studies suggest that clinical doses of minor tranquilizers do not cause overt impairment of behavior or performance (23).

It is possible that meprobamate is effective in the experimental laboratory but would be ineffective in the field lie­detection laboratory, where fear of detection is presumably greater. It should be noted, however, that 400-mg doses of meprobamate are effective in reducing the anxiety of psychiatric patients (24). It is also possible that substantially higher doses than we used here could be used unobtrusively to defeat the field test; several studies (23) show no evidence of observable behavioral impairment with single doses of 800 mg. Finally, it may be that meprobamate would be ineffective with the "control question" test more commonly used in the field. Since for guilty subjects there is little difference in principle between the two types of test, the differences arising mainly for innocent subjects (15), the present results may well generalize to the control question test. On the other hand, the more arousing circumstances surrounding a field test and the more intrusive questioning practiced by field examiners might overcome any effects of a tranquilizer.

Nonetheless, the results reported here are consistent with the hypothesis, based on clinical observations, that minor tranquilizers such as meprobamate selectively reduce the phasic physiological response to disturbing social stimuli rather than simply lower tonic levels of arousal.





Institute of Pennsylvania Hospital and University of Pennsylvania, 111 North 49 Street, Philadelphia 19139

References and Notes

1. M. T. Orne, R. I. Thackray, D. A. Paskewitz, in Handbook of Psychophysiology, N. S. Greenfield and R. A. Sternback, Eds. (Holt, Rinehart & Winston, New York, 1972); W. M. Waid, M. T. Orne, S. K. Wilson, Psychophysiology 16,15 (1979); S. K. Wilson, W. M. Waid, M. T. Orne, Psychosom. Med. 40, 91 (1978); W. M. Waid, E. C. Orne, M. R. Cook, M. T. Orne, J. Appl. Psychol. 63, 728 (1978).

2. D. T. Lykken, Am. Psychol. 29, 725 (1974); Contemp. Psychol. 23, 81 (1978); Psychol. Bull. 86, 47 (1979); ibid., in press; J. A. Podlesny and D. C. Raskin, ibid. 84, 782 (1977); ibid. 86, 54 (1979).

3. Polygraph Control and Civil Liberties Protec-



tion Act -- Hearings on S. 1845 before the Subcommittee on the Constitution of the Senate Committee on the Judiciary, 95th Congress, first and second sessions (1977 and 1978); R. Phillion v. Her Majesty the Queen (Supreme Court of Canada, October 1977).

4. G. H. Barland and D. C. Raskin, in Electrodermal Activity in Psychological Research, W. F. Prokasy and D. C. Raskin, Eds. (Academic Press, New York, 1973).

5. J. E. Reid and F. E. Inbau, Truth and Deception: The Polygraph ("Lie-Detector") Technique (Williams & Wilkins, Baltimore, ed. 2, 1977).

6. "The anti-anxiety agents share a similar central depressant action: the ability to produce mild sedation in doses that are generally unlikely to cause soporific effects or to adversely affect the quality of consciousness and the quality of psychomotor performance" [American Medical Association Department of Drugs, American Medical Association Drug Evaluations (Publishing Sciences Group, Acton, Mass., 1973), p. 317].

7. D. M. McNair, Arch. Gen. Psychiatry 29, 611 (1973).

8. Conflicting reports of the effects of tranquilizers on detection have been made, but these have been based on rather informal observations: M. A. Berman, J. Am. Polygraph Assoc. 4, 329 (1975); C. Hess, ibid., p. 314; A. B. Kristofferson and R. H. Cormack, in Drugs and Behavior, L. Uhr and J. G. Miller, Eds. (Wiley, New York, 1960).

9. E. L. Kelly et al., Arch. Neural. Psychiatry 80, 247 (1958); V. G. Laties, J. Abnor. Soc. Psychol. 59, 156 (1959); G. A. Leinert and W. Traxel, J. Psychol. 48, 329 (1959).

10. In field polygraph testing the actual guilt or innocence of subjects usually does not become established for a long time, if at all. More serious for the purpose of scientific investigation is that the ultimate disposition of the case cannot be considered to be independent of the polygraph test outcome: suspects who appear truthful may not be examined as thoroughly as those who appear deceptive. Finally, it is not typically feasible to introduce experimental treatments, such as tranquilizers, into field polygraph tests. Consequently, we must depend upon laboratory experiments. The experimental subject cannot be expected o suffer the same degree of apprehension as the field subject. Although this must be kept in mind, it would be an obstacle to experimental research only if the detection of deception failed under such conditions. Many studies document the effectiveness of the psychophysiological detection of deception in the laboratory, particularly with the guilty knowledge test.

11. The subjects were male college students, 18 to 24 years of age, recruited through advertisements and paid $2 an hour.

12. "Guilty" subjects were told that the polygrapher would do his best to obtain a confession but that it was possible to "beat the polygraph" by controlling one's emotions. "Innocent" subjects were told that the polygraph examiner would suspect them and that it was often very difficult to prove one's innocence in a lie-detection test. As a test of their "ability to perform under stress," the innocent subjects completed the same timed, interpolated tasks as did the guilty ones, but without learning words.

13. Meprobamate taken orally reaches its peak plasma concentration in 1 to 2 hours [B. J. Ludwig and J. R. Potterfield, in Advances in Pharmacology and Chemotherapy, S. Garattini, A. Goldin, F. Hawking, I, J, Kopin, Eds. (Academic Press, New York, 1971), vol. 9].

14. Experimenter 2 was a male medical student. Sensors were attached as in field exams, a blood-pressure cuff on the left arm, EDR electrodes on the first phalange of the second and third fingers of the right hand, and a respiration belt around the chest; recordings were made on a Stoelting polygraph.

15. D. T. Lykken, J. Appl. Psychol. 43, 385 (1959); ibid. 44, 258 (1960); Am. Psychol. 29, 725 (1974). Lykken has outlined the logic underlying the technique of detecting guilty knowledge, in contrast to the controversial lie test or control­question test more widely used in the field. Field polygraphers assume that the conditions for the guilty knowledge test, the existence of information known only to the guilty person, can rarely be met, but Reid and Inbau (5) report many ingenious uses in the field of a peak-of-tension test, a field variant of guilty-knowledge detection. The guilty-knowledge technique may come to be used more widely in light of the relatively low accuracy of the physiological data generated by the lie test as widely used in the field [F. S. Horvath, J. Appl. Psychol. 62, 127 (1977)].

16. Finally, experimenter 3 conducted a postexperimental interview and debriefing and answered the subject's questions about the experiment.

17. Amplitude of the EDR, suppression of breath amplitude, and amplitude of the cardiovascular response for each stimulus were measured as in Thackray and Orne (20). Detections were scored separately for each channel. A detection was counted if the critical stimulus in a set of four words evoked a larger physiological response than any of the three other items. For innocent subjects one word in each category was randomly designated to be the critical item for purposes of analysis.

18. S. Siegel, Nonparametric Statistics for the Behavioral Sciences (McGraw-Hill, New York, 1956). Since each of the comparisons reported were planned comparisons, significance was evaluated in terms of the per-comparison error rate as described in G. Keppel, Design and Analysis: A Researcher's Handbook (Prentice­Hall, Englewood Cliffs, N.J., 1973). The use of Fisher's exact test on an ordinal variable that is dichotomized at some arbitrary point introduces the possibility of misclassification. This does not affect the validity of the significance test but reduces its power [I. Bross, Biometrics 10, 478 (1954)].

19. The three meprobamate subjects who were classified as deceptive, however, were detected as frequently as the most detectable placebo or no­pill subjects. Thus, although there were statistically significant differences between the groups in the number of deceivers detected, there was no such difference in the mean number of lies detected.

20. R. I. Thackray and M. T. Orne, Psychophysiology 4, 329 (1968).

21. R. J. Cutrow, A. Parks, N. Lucas, K. Thomas, ibid. 9, 578 (1972). Significant effects on cardio­vascular and respiratory measures have been reported [G. A. Barland and D. C. Raskin, ibid. 12, 231 (1975)], with analysis of variance of group data. However, no study has reported significant discrimination of individual subjects as deceptive or truthful on the basis of respiratory or relative blood pressure measures alone.

22. K. Rickels, R. B. Cattell, C. Weise, B. Gray, R. Yee, A. Mallin, H. G. Aaronson, Psychopharmacology 9, 288 (1966).

23. N. H. Pronko and G. Y. Kenyon, Psychol. Rep. 5, 217 (1959); L. Uhr and J. G. Miller, Am. J. Med. Sci. 240, 204 (1960).

24. D. F. Klein and J. M. Davis, Diagnosis and Drug Treatment of Psychiatric Disorders (Williams & Wilkins, Baltimore, 1969).

25. Supported in part, by, a grant from the Institute for Experimental Psychiatry. We thank David F. Dinges, Kevin M. McConkey, Frederick J. Evans, and William H. Putnam for their substantive comments during the preparation of the manuscript.

17 June 1980; revised 7 October 1980

The preceding paper is a reproduction of the following article (Waid, W.M., Orne, E.C., Cook, M.R., & Orne, M.T. Meprobamate reduces accuracy of physiological detection of deception. Science, 1981, 212, 71-73.). It is reproduced here with the kind permission of the American Association for the Advancement of Science, ©1981. Readers may view, browse, and/or download material for temporary copying purposes only, provided these uses are for noncommercial personal purposes. Except as provided by law, this material may not be further reproduced, distributed, transmitted, modified, adapted, performed, displayed, published, or sold in whole or in part, without prior written permission from the publisher.