Back to Lie Detection Home Page Psychophysiology, Vol. 25, No.6, pp.683-688 Diff_decep88.doc Differentiation of Deception as a Psychological Process: A Psychophysiological Approach JOHN J. FUREDY, University of Toronto CAROLINE DAVIS, AND MARIA GUREVICH York University, Toronto ABSTRACT If psychophysiology is the study or differentiation of psychological processes by means of physiological measures, then the experimental demonstration of deception as a psychophysiological phenomenon requires a comparison of physiological responses to two conditions (experimental and control) which differ only with respect to deception. To this end, the Differentiation-of-Deception Paradigm controls for differential question significance and frequency of occurrence. Thirty-two subjects were tested in this paradigm, with the skin conductance response as the dependent variable. We examined, within subjects: a) the basic deception comparison which contrasted relatively neutral autobiographical questions answered deceptively with those answered honestly, and b) the mode of answering, which was either an immediate answer to the question (conventional method) or an answer delayed by 10 seconds. The deception phenomenon (greater responding to deceptive relative to honest trials) emerged significantly (and nondifferentially) to both the immediate and delayed questions, but (perhaps because of response interference) not when responding was measured immediately following the delayed answers. Future research should vary other conditions and measure additional dependent variables with the aim of investigating possible psychological and physiological mechanisms, as well as extending the deception phenomenon beyond its present electrodermal form. DESCRIPTORS: Deception differentiation, Skin conductance response, CQT polygraph}, Guilty knowledge technique, Control for significance and frequency, Response interference The approach taken in this research is one that views psychophysiology as the study or differentiation of psychological processes by means of unobtrusively measuring slight changes in physiological functions (for details and criticisms of this perspective, see Furedy, 1983,1984;Obrist, 1976, 1981; Stern, 1964, 1984). The initial step in such study is to demonstrate the phenomenon under consideration, after which issues concerning parametric variations and mechanisms (both psychological and physiological) can be raised. For example, if the This research was supported by an operating grant and a PostDoctoral Fellowship, respectively, to JJF and CD from the National Sciences and Engineering Council of Canada. We are indebted to M. Dawson and an anonymous reviewer for comments on an earlier draft of this paper. Address requests for reprints to: John J. Furedy, Ph.D., Department of Psychology, University of Toronto, Toronto, Ontario, Canada, M5S IA1. psychological process of interest is Pavlovian autonomic (e.g., skin conductance response, SCR) conditioning, then the demonstration of the conditioning phenomenon requires that the experimental and control conditional stimuli be, in principle, identical except for the process being studied: the association between the (experimental) conditional stimulus and the unconditional stimulus. Similarly, if deception is the psychological process of interest, the experimental and control stimuli should differ only with respect to deception. The investigator who wishes to differentiate and study deception faces the same problem as presents itself in the study of processes like Pavlovian conditioning. At the initial phenomenon-demonstration stage there may be differences between the experimental and control conditions that do not involve deception, and yet may be responsible for the observed experimental-control differences in responsiveness. Although there are innumerable potential confounds, two especially important classes 684 Furedy, Davis, and Gurevich of potential confounding variables to control in research on the phenomenon of deception are differential significance and differential familiarity. The confound of differential significance (or emotionality, anxiety, or arousal level) between the two sorts of questions appears to be particularly evident in the polygraphic Control Question Technique (CQT). For example, in a case involving child sex abuse (for details, see Furedy & Liss, 1986, pp. 96-97), one relevant (“experimental”) question was: “Did you lick X’s vagina?” (where X was a four-year-old girl). The so-called control question that was used as a comparison was: “Did you ever do anything you were ashamed of?” Even for an innocent examinee (and perhaps all the more so), it is at least arguable that the significance of the former question would exceed that of the latter question. In the present differentiation-ofdeception paradigm, which was based on a procedure originated by Hemsley (see Hemsley, Heslegrave, 8c Furedy, 1980) and developed further by Heslegrave (1982), the significance confound was controlled by two methods. First, to eliminate gross differences in significance between questions to which subjects were instructed to be deceptive or honest, emotionally neutral questions dealing with biographical issues were employed. Second, to control for slight differences in significance between different questions, the honest and deceptive categories were equated across subjects so that, on the average, each question (of the form, “I was born in ...) received a deceptive and an honest answer 50% of the time. The differential familiarity or frequency of occurrence confound is the primary reason why the Guilty Knowledge Technique (GKT) introduced by Lykken (1959) does not necessarily assess deception, although it is a more scientifically valid method of assessing guilt than the CQT (for a more detailed comparison of the CQT and GKT from a psychophysiological perspective, see Furedy, 1986, and Furedy & Heslegrave, 1988a, 1988b). According to the GKT’s rationale, the guilty will produce bigger responses to the relevant question (the one containing the information that only the guilty know) than to the control questions (which do not contain that information). However, control questions occur more frequently than do relevant questions, so that, for the guilty (who, by assumption, are the only examinees who can make the relevant:control distinction), the rate of response habituation will be greater for the control than for the relevant questions. Accordingly, any difference between relevant and control questions may be due not to deception but to differential habituation rates. This, of course, does not raise any concerns of confounding, if the aim Vol. 25, No. 6 of the procedure is to detect guilt. Response habituation has been proposed by some (e.g., Ben-Shakhar, 1977) as an alternative mechanism to that of stimulus significance in accounting for the demonstrated GKT phenomenon of increased responding to relevant questions by the guilty. However, if the purpose is to compare physiological responses to the relevant and control questions in order to differentiate the processes of deception and honesty, then the frequency-of-occurrence difference between the two conditions constitutes a serious confound particularly for the skin conductance response, which clearly habituates as a function of frequency of presentation (e.g., Furedy, 1968). In the present experiment, frequency was controlled by having subjects answer 50% of the questions deceptively (experimental) and 50% honestly (control), so that, in all important respects, the only difference between the Deceptive and Honest conditions would be the presence of deception. In this first full-length report of the differentiation-ofdeception paradigm (for previous abstract reports, see Hemsley et al., 1980; Heslegrave, 1982), skin conductance response (SCR) was measured as the dependent variable. This choice was based on the SCR’s relatively high sensitivity to psychological processes like attention (orienting) and classical conditioning, and the fact that it is one of the least esoteric (and hence most easily replicable) of psychophysiological measures. In addition to the basic Deceptive:Honest comparison we also varied, within subjects, the verbal response latency—whether the answer to the question was given immediately or after a 10-s delay. This factor was varied by Dawson (1980), who, however, employed a laboratory version of the polygraphic CQT. The purpose for introducing the delayed-answer condition was that this condition could possibly separate two hypothetical processes involved in the deception phenomenon, namely the intention to deceive (as indexed by SCRs to the question) and the act of deception (as indexed by SCRs to the 10-s delayed answer). It is possible that such a separation may produce clearer Deceptive:Honest differentiation, as well as provide clues to understanding the mechanisms of deception. Method Subjects Thirty-two volunteer subjects (16 males and 16 females) aged 19-40 yrs were recruited through advertisements posted at the University of Toronto. Each subject received $10 for participation in the study. Apparatus and Questionnaire Materials Continuous skin conductance was recorded, through a Coulbourn Instruments preamplifier (S71 -22; this unit November, 1988 Electrodermal Differentiation of Deception is based on specifications recommended by Lykken & Venables, 1971) onto a Narco (E & M Instruments) Physiograph (Model FOUR-A) running at a chart speed of 1 cm/s. Two Grass Ag/AgCl electrodes (1 cm2 dia.) in conjunction with a Beckman NaCl electrode paste were attached by adhesive tape to the volar surfaces of the subject’s distal phalanges of the first and second fingers of the left hand (cleaned with soap and water). The questionnaire materials comprised two lists of 10 questions (List 1 & List 2) such as “What is your mother’s age,” and “How many brothers do you have?”. Each pair of questions in the two lists was matched for topic (e.g., the matched question for the first example would deal with the father’s age), so that the biographical content of the two lists was roughly equal. Procedure Two experimenters (CD and MG) conducted the study. On arrival at the laboratory, each subject was given a brief written description of the study by CD. after which MG attached the electrodes to the subject but did not connect the leads to the polygraph. In general, it can be argued that truthful answers are easier to retrieve than deceptive answers. However, it is possible to minimize the influence of this ease-ofretrieval factor by procedural modifications that: a) render the retrieval component as easy as possible, and b) present the task to the subject as a deception rather than a memory test. Accordingly, the subject was then taken by CD to a small room separate from the testing area, and was helped to prepare honest and deceptive answers, with CD indicating which 10 of the 20 questions should be answered deceptively. For deceptive answers, the subject was asked to provide plausible ones (e.g., if the question asked about father’s age, which was 60, then a deceptive answer should not be, say. 35). The questions were grouped as in the two original lists (List 1 and List 2), with half the subjects being required to be honest to List 1 questions, and the remainder being asked to be honest to List 2 questions. After agreeing on the answers, the subject and CD rehearsed the list of questions and answers until the subject felt confident with the prepared answers. No subject required more than one run through the list of questions before indicating confidence with the answers, a fact that indicates the relative ease of the retrieval task. The subject was then told that, during the interrogation (test) process. MG would ask the same 20 questions, but in a different order from that used during the rehearsal process. The subject was also reminded that MG did not know which answers were honest or deceptive, and that an attempt should be made to appear honest at all times. Finally, it was made clear that CD would not know what answers were given to MG. As a result, the procedure was unlikely to be construed as a memory test by the subject. The subject was then taken to the interrogation room (immediately adjacent to the room containing the physiograph and stimulus-control equipment), and the electrode leads were attached to the physiograph via 685 cables extending through a hole in the wall adjoining the two rooms. In the interrogation room, the subject and MG were seated, facing each other, across a narrow (approx. 50 cm) table. The questionnaire material was administered according to the following arrangement. The questions of List 1 and List 2 were randomly allocated to a third, 20question list, List 3. The first and second 10 questions in List 3 were presented under the Immediate and Delayed conditions, respectively, for half the subjects, and the reverse for the remaining subjects. In the Immediate condition, the subject was instructed to answer as soon as MG had completed the question (question duration was approx. 2-3 s). In the Delayed condition, the subject was asked to wait 10 s. A small light, mounted on a board immediately in front of the subject, came on as soon as MG had completed the question, and signalled the subject to answer the question when it offset 10 s later. Accordingly, with respect to the main within-subject comparisons of interest (i.e., the Honest/Deceptive and Immediate/Delayed differences), there was counterbalancing control for question content, frequency of occurrence, and serial position in the list (List 3). The between-subject factor was Sex, which was varied orthogonally with the Truthfulness and Measurement Method (MM). It will be noted that the MM factor can be considered to comprise three levels: skin conductance responses to the Immediate, Delayed-Question (Q), and Delayed-Answer (A) conditions. The timing of the questions was controlled by MG, who had a small light (visible only to her) mounted on a board in front of her. At the onset of each question, the light was switched on, and remained on for either 15 s (Immediate condition) or 25 s (Delayed condition). The next question in the series was presented by MG following light offset, so that the interval between the answer to the last question and the next question was approximately 10 s throughout the experiment. The interrogation did not begin until CD had completed calibrating the physiograph. and was ready to begin recording (indicated by closing the door between the interrogation and recording rooms). Before beginning the questions in List 3. CD asked the subject’s name, but the response to this question was not scored. The physiograph record was marked by MG to indicate the onset of each question, and, in the Delayed condition, the answer onset. Between the first and second set of 10 questions, there was a break of about 1 min. during which time MG reminded the subject about the change in the method of answering. At the completion of the questions in List 3. the electrodes were removed, and the subject was thanked and paid. Results The skin conductance response (SCR) for the Immediate and Delayed(Q) conditions was defined as any response that was initiated (showing an inflection point) within 1-5 s following question onset. Magnitude (in mm, and then convened to 686 Furedy, Davis, and Gurevich microSiemens, S) was expressed as the difference between response onset (inflection point) and the highest point following response onset until the end of the 1-5 s latency window. Non-occurrences of SCRs were changes less than .5 mm change on the chart, and were scored as zero responses. This chart-length criterion varied among subjects (because of wide individual differences in skin conductance level), but was constant for all within-subject variables. The scoring was the same for the Delayed(A) condition, except that the latency window was 1-5 s following (delayed) answer onset. Initial ANOVAs including Sex, Truthfulness, and MM as factors indicated that Sex was not significant either as a main effect or in interaction with the other factors. Accordingly, subsequent analyses collapsed the data across Sex groups and tested only within-subject effects. A two-factor repeated-measures ANOVA with two levels of Truthfulness (Honest and Deceptive) and three levels of MM (Immediate and the two Delayed conditions) was conducted, with an alpha level of .05 set for statistical significance. Univariate tests for within-subject effects were adjusted for violation of sphericity using the Greenhouse-Geisser epsilon correction (epsilon=.9066). The only significant effect was a Truthfulness X MM interaction, F(2/62)=4.16, p=.0239, MS=0.215. As suggested by inspection of Figure 1, this interaction was due to the Deceptive>Honest difference being eliminated (but not reversed) under the Delayed(A) condition. This account of the interaction was confirmed by the following analyses: a) a 2 X 2, Truthfulness X MM ANOVA using the Immediate and Delayed(Q) conditions as the only (two) MM levels, yielded a significant main effect only for Truthfulness, F(1 /31) = 10.76, p=.0026, MS=0.370, but no Truthfulness X MM interaction; and b) a t-test of the Deceptive < Honest difference in the Immediate(A) condition failed to reach significance, t(30)=1.71, p<.l. To test for response and deception-differentiation habituation to the Immediate and Delayed(Q) conditions, a 2 X 5 repeated-measures ANOVA was conducted with Truthfulness and Trial Blocks as the two factors. For each subject there were (interspersed) 10 trials with honest questions and 10 trials with deceptive questions. The five levels of the trial block factor were obtained by comparing skin conductance responses (SCRs) across pairs of questions 1-2, 3-4, 56, 7-8, and 9-10. No evidence for habituation was obtained, since the ANOVA yielded no significant Trial Block effect, F<1, with mean SCR values of 0.63, 0.56, 0.53, 0.49, and 0.53 S for the first to fifth Trial Block levels. Only the Truthfulness effect was significant, F(1/31)=6.63, Vol. 25, No. 6 Figure 1. Mean skin conductance responses to honest (H) and deceptive (D) questions following the question in Immediate and Delayed conditions, and following the answer in the Delayed condition. p=.0150, MS=0.071, with responding to the Deceptive condition exceeding that to the Honest condition. The Truthfulness X Trial Block interaction was also not significant, F<1, so there was no evidence for differentiation-of-deception (as represented by the Deceptive >Honest effect) habituation. Discussion The major finding is that skin conductance responses (SCRs) in the interval immediately following the question in the differentiation-of-deception paradigm were clearly and significantly greater when the question was answered deceptively than when it was answered honestly. This constitutes a relatively unequivocal demonstration of the psycho-physiological deception phenomenon, because it is at least plausible to suggest that the only relevant difference between the experimental (deceptive) and control (honest) conditions was the presence of deception. It is also encouraging that although mean SCR magnitude to the questions was relatively small (see Figure 1), and no greater than that typically obtained in orientingreaction (OR) studies with tones and lights as stimuli (e.g., Furedy, 1968), nevertheless, in contrast to those OR studies, the SCR in this differentiation-of-deception paradigm did not November, 1988 Electrodermal Differentiation of Deception habituate. Nor did the differentiation-of-deception phenomenon itself diminish over trials. The demonstration of a phenomenon is the initial step in a process that requires the experimental manipulation of a variety of additional factors in order to uncover underlying mechanisms. It will be recalled that we varied the measurement method (MM) of deception differentiation, with the idea of separating the intention to deceive from the act of deception, and with the hope of producing clearer Deceptive:Honest differentiation. This manipulation did not yield the expected results. In the first instance, the Immediate versus Delayed variable did not affect the extent of SCR differentiation between deceptive and honest questions. Secondly, and surprisingly, in the Delayed(A) condition the Deceptive>Honest difference was eliminated, with a nonsignificant tendency (p<.1) to show a reversal, i.e., Deceptive<Honest (see Figure 1). In terms of the hypothetical intention/act distinction, this pattern of outcomes suggests that it is only the intention to deceive rather than the act of deception that is responsible for the differentiation-of-deception phenomenon. This account is consistent with the nodifference-in-differentiation finding between the Immediate and Delayed(Q) conditions (i.e., the absence of the act from the latter condition not affecting differentiation), and the lack of differentiation in the Delayed( A) condition (which has no intention component). An alternative account, however, is suggested by a factor for which there is more direct psychophysiological evidence than for the hypothetical intention/act distinction. This is the factor of response interference, which is known to be influential in skin conductance measurement (see, e.g., Schell & Grings, 1971, where skin conductance response magnitude differentiated between subjective stimulus intensities). Response-interference is a relativerefractory-period-like effect, whereby the response to a stimulus is reduced as a function of a previously occurring response. With a slow-onset-latency response like the skin conductance response, the effect is evident up to intervals of about 30 s, and would be markedly influential with the present 10-s interval between the question- and answer-elicited skin conductance responses in the two 687 Delayed conditions. The effect is also a direct function of the magnitude of the previously-occurring response, which, because of the Deceptive:Honest differentiation in the Delayed(Q) condition, would have been greater for the Deceptive than the Honest conditions in the interval immediately following the question. Accordingly, the response-interference factor could have eliminated the Deceptive:Honest differentiation to the answer in the Delayed(A) condition, and could even have reversed it (although this reversal did not reach significance in the present study). One way to distinguish between the psychological intention/act and physiological response-interference accounts would be to add a dependent physiological index where, with a 10-s interval, the response-interference problem is not as severe as with the skin conductance response. The faster-acting event-related potential appears to be an obvious and promising candidate for this purpose. Another issue regarding causal mechanisms involved in the deception phenomenon concerns the relative importance of differential mental load (inasmuch as in the Deceptive condition, the subject may have to think of both the true and the false answer) and differential conflict (potentially greater in the Deceptive condition, where there is a conflict between the true and false answers). This psychological issue can be investigated by adding to the Deceptive and Honest conditions (Heslegrave, 1982). Finally, a potentially informative physiological extension involves adding cardiac performance variables, in order to more clearly separate the roles of sympathetic and parasympathetic influences in the deception phenomenon. Specifically, with the joint measurement of heart rate and one of the candidate sympathetic indices (e.g., pulse transit time, T-wave amplitude, etc.), it is possible to identify both sympathetic (change in the sympathetic index) and parasympathetic (change in heart rate without change in the sympathetic index) effects (Furedy, Heslegrave, & Scher, 1984). In addition, the more general benefit of providing these other dependent variables is that this could serve to extend the differentiation of deception beyond its present status as only an electrodermal phenomenon. REFERENCES Ben-Shakhar, G. (1977). A further study of the dichotomization theory in detection of information. Psychophysiology, 14, 408-413. Dawson, ME. (1980). Physiological detection of deception: Measurement of responses to questions and answers during countermeasure maneuvers. Psychophysiology, 17, 8-17. Furedy, JJ. (1968). Human orienting reaction as a function of electrodermal versus plethysmographic response modes and single versus alternating stimulus series. Journal of Experimental Psychology. 77, 70-78. Furedy, J.J. (1983). Operational, analogical, and genuine definitions of psychophysiology. International Journal of Psychophysiology, 1. 13-19. 688 Furedy, Davis, and Gurevich Furedy, J.J. (1984). Generalities and specifics in defining psychophysiology: Reply to Stern (1964) and Stern (1984). International Journal of Psychophysiology, 2, 24. Furedy, JJ. (1986). Lie detection as psychophysiological differentiation: Some fine lines. In M. Coles, E. Don-chin, & E. Porges (Eds.). Psychophysiology: Systems, processes, and applications—A handbook (pp. 683-700). New York: Guilford Press. Furedy, J.J., & Heslegrave, R.J. (1988a, in press). The forensic use of the polygraph: A psychophysiological analysis of current trends and future prospects. In J. Jennings, M. Coles, & P. Ackles (Eds.), Advances in psychophysiology (Vol. 4). Greenwich, CT: JAI Press. Furedy, J.J., & Heslegrave, R.J. (1988b, in press). A reply to commentators: Some elaborations on the specific-effects orientation’s application to the North American CQT polygraph. In J. Jennings, M. Coles, & P. Ackles (Eds.), Advances in psychophysiology (Vol. 4). Greenwich, CT: JAI Press. Furedy, J.J., Heslegrave, R.J., & Scher, H. (1984). Psychophysiological and physiological aspects of T-wave amplitude in the objective study of behavior. Pavlovian Journal of Biological Science, 19, 182-94. Furedy, J.J., & Liss, J. (1986). Countering confessions induced by the polygraph: Of confessionals and psycho- Vol. 25, No. 6 logical rubber hoses. The Criminal Law Quarterly, 29, 92114. Hemsley, G.D., Heslegrave, R.J., & Furedy, J.J. (1980). Can deception be detected when stimulus familiarity is controlled? [Abstract]. Psychophysiology, 17, 286-287. Heslegrave, R.J. (1982). An examination of the psychological mechanisms underlying deception [Abstract]. Psychophysiology, 19, 298. Lykken, D.T. (1959). The GSR in the detection of guilt. Journal of Applied Psychology, 43, 385-388. Lykken, D.T & Venables, P.H. (1971). Direct measurement of skin conductance: A proposal for standardization. Psychophysiology, 8, 656-672. Obrist, P.A. (1976). The cardiovascular-behavioral interaction— As it appears today. Psychophysiology, 13,95-107. Obrist, P.A. (1981). Cardiovascular psychophysiology: A perspective. New York: Plenum Press. Schell, A.M., & Grings, W.W. (1971). Judgments of UCS intensity and diminution of the unconditioned GSR. Psychophysiology, 8, 427-432. Stern, J.A. (1964). Toward a definition of psychophysiology, Psychophysiology, 7, 90-91. Stern, J.A. (1984). On ‘Unobtrusive Psychophysiology’— Fact or fiction? A rejoinder to Furedy (1983). International Journal of Psychophysiology, 2, 1-2. (Manuscript received October 21, 1987; accepted for publication April 23, 1988)