PSYCHOWIYSIOLOGY

Copyright * 1978 by The Society for Psychophysiological Research. Inc.Vol. 15, No. 2

Printed in U.S.A.

Psychopathy and Detection of Deception In aPrison Population

DAVID C . RASKIN

University of Utah

AND ROBERT D . HARE

University of British Columbia

ABSTRACT

The effectiveness of detection of deception was evaluated with a sample of 48 prisoners, half ofwhom were diagnosed psychopaths. Half of each group were "guilty" of taking $20 in a mockcrime and half were "innocent." An examiner who had no knowledge of the guilt or innocenceof each subject conducted a field-type interview followed hy a control question polygraph examina-tion. Electrodermal, respiration, and cardiovascular activity was recorded, and field (semi-ohjec-tive) and quantitative evaluations of the physiological responses were made. Field evaluations bythe examiner produced 88% correct, 4% wrong, and 8% inconclusives. Excluding inconclusives,there were 96% correct decisions. Using blind quantitative scoring and field evaluations, sig-nificant discrimination between "guilty" and "innocent" subjects was obtained for a variety ofelectrodermal, respiration, and cardiovascular measures. Psychopaths were as easily detected asnonpsychopaths, and psychopaths showed evidence of stronger electrodermal responses and heartrate decelerations. The effectiveness of control question techniques in differentiating truth anddeception was demonstrated in psychopathic and nonpsychopathic criminals in a mock crimesituation, and the generalizability of the results to the field situation is discussed.

DESCRIPTORS: Detection of deception. Heart rate. Polygraph tests. Psychopathy, Respira-tion, Skin conductance. Skin potential, Vasomotor response.

Although many investigators have carried out re-search on detection of deception with criminals andcriminal suspects (e.g., Bersh, 1969; Kugelmass,Lieblich, Ben-Ishai, Opatowski, & Kaplan, 1968;Luria, 1932; Lykken, 1955; Marston, 1921; Ras-kin, Barland, & Podlesny, 1977), there is little

The authors express their appreciation to Wendy Coulthard,Michael Dame, Janice Frazelle, Brent McNeil, and John A, Pod-lesny who aided in the collection and analysis of the data. Theresearch was supported by Grant Number 75-NI-99-0001 fromthe National Institute of Law Enforcement and Criminal Justice,Law Enforcement Assistance Administration, U,S, Departmentof Justice; Biomedical Sciences Support Grant FR-07092 fromthe Research Support Branch, National Institutes of Health; andGrant MA-4511 from the Medical Research Council of Canada,Points of view or opinions stated in this document are those ofthe authors and do not necessarily represent the official positionor policies of the U,S, Department of Justice,

Address requests for reprints to: David C, Raskin, Departmentof Psychology, University of Utah, Salt Lake City, Utah 84112,

evidence on the relationship between psychopathyand the effectiveness of detection of deception tech-niques. However, there is laboratory evidence thatpsychopathic (or sociopathic) criminals are electro-dermally hyporeactive under some experimentalconditions (Hare, 1978). On the basis of such find-ings and the commonly-held belief that psychopathsare very adept at manipulating and deceiving others(e.g., Clancy, Noyes, & Travis, 1974; Cleckley,1964), there are frequent claims that psychopathscan "beat" the polygraph test (Barland & Raskin,1973; Ferguson & Miller, 1974; Floch, 1950; Levitt,1955).

The possibility that psychopaths may be able todefeat the polygraph test is suggested by the factthat they appear to be electrodermally hyporespon-sive to intense, noxious, or painful stimuli or tothe anticipation of such stimuli (Hare, 1978), Ifthe stimuli presented in a polygraph test are of sucha nature and only electrodermal measures are ob-

1260048-5772/78/0300-0126$l. 10/0

1978 The Society for Psychophysiological Research, Inc.

March. 1978 LIE DETECTION IN PRISONERS AND PSYCHOPATHS 127

tained, then it might be the case that deceptivepsychopaths would show relatively weak electro-dermal responses to the questions and be able toescape detection. That outcome might occur inpolygraph tests where lack of responsiveness to thestimuli is taken as an indication of truthfulness,such as with the relevant-irrelevant and guiltyknowledge tests (Podlesny & Raskin, 1977). How-ever, the design of the control question test (Pod-lesny & Raskin, 1977) attempts to protect againstsuch problems of hyporesponsivity by requiringthat reactions occur with greater strength to controlquestions (see Method section) before a truthfuljudgment can be rendered. The major purpose ofthe present study was to assess the extent to whichthe control question test is effective in detectingdeception with psychopaths when standard mea-sures of respiration and cardiovascular activity areincluded along with electrodermal activity.

There are only two published scientific studieswhich have used psychopaths in a detection of de-ception situation. As part of a laboratory studyusing tones and electric shocks, Lykken (1955)required subjects to choose a number between 1 and5 and then conceal the number by answering "no"when asked about the number. Nonpsychopathiccriminals and noncriminal subjects produced largerelectrodermal responses to the chosen number, butpsychopathic criminals showed large reactions toall numbers. Since the Lykken study did not usecurrent polygraph examination techniques and itemployed only electrodermal measures, little canbe concluded with regard to the detectability ofpsychopaths.

Raskin et al. (1977) examined criminal suspectsusing control-question polygraph techniques. Indi-viduals had been referred by prosecutors, law en-forcement agencies, and defense counsel for ex-amination concerning the crime of which they wereaccused. Guilt or innocence was assessed by a panelof experts consisting of two defense attorneys, twoprosecutors, and a judge. They made independentjudgments based on all of the evidence, disregard-ing legal technicalities and without any knowledgeof the polygraph results. Of the 36 subjects con-sidered to be guilty by a majority of the panel,psychopaths and nonpsychopaths were identifiedusing the Pd scale of the Minnesota MultiphasicPersonality Inventory (MMPI). The 14 subjectswith the highest T-scores (mean T=85.4) were des-ignated psychopathic, and the 12 lowest scorers(mean T=50.3) were designated nonpsychopathic.There were no significant differences in overallreactivity for the two groups, and all of those diag-nosed as psychopathic produced deceptive results.

There are two problems in the interpretation ofthe Raskin et al. results. First, the relationship be-

tween MMPI scores or profiles and the clinicalconception of psychopaths has not yet been ade-quately specified (Hare & Cox, 1978), and researchis needed using subjects selected on the basis ofstringent clinical criteria. Second, the study wasbased on the results of field polygraph examinationsin which there was some uncertainty about whetheror not the subject was guilty (ground truth). Al-though all of the psychopathic subjects produceddeceptive outcomes, it is desirable to have the cer-tain knowledge of ground truth that is provided bya laboratory experiment.

The present study attempted to rectify the prob-lems inherent in the Lykken (1955) and Raskin et al.(1977) studies. A control-question test of the typeemployed by many field polygraph examiners (Bar-land & Raskin, 1973) was administered by a trainedpolygraph examiner to subjects in a mock-crimesituation which attempted to simulate the fieldsituation. In addition to electrodermal measures,cardiovascular and respiratory activity were re-corded. The subjects were drawn from a prisonpopulation, and a clinical assessment procedurewas utilized to categorize subjects as psychopathicor nonpsychopathic. On the basis of those pro-cedures, we felt that the outcome would provide areasonable estimate of the extent to which psycho-paths can be detected in deception by means ofphysiological measures and field polygraph tech-niques.

Method

Subjects

Subjects were obtained from the inmate population ofthe Lower Mainland Regional Correctional Centre inBumaby, British Columbia. Inmates volunteered for a"lie-detection" study in which they could win a $20bonus for successful performance. The subjects weredivided into psychopathic and nonpsychopathic groupson the basis of procedures similar to those used in aseries of studies by Hare and his colleagues (Hare &Cox, 1978).

Each potential subject's file was obtained from therecords center at the prison. The information consisted ofreports by correction officers, parole officers, physi-cians, psychiatrists, psychologists, and social workers,as well as background information sheets, criminalrecords, court transcripts, law enforcement and inmateversions of the offenses, parole applications, and dailylog sheets. Using this information, a trained and ex-perienced researcher rated each subject on a 7-point scaleof psychopathy. The ratings were made on the basis ofthedegree to which the inmate's behavior over a long periodof time was consistent with the conception of psycho-pathy outlined by Cleckley (1964). Although based on aglobal assessment of psychopathy, these procedures arereliable when experienced raters are used. In general,inter-rater reliability coefficients in the .75 to .90 rangehave been obtained (Chesno & Kilmann, 1975; Denge-

128 RASKIN AND HARE Vol. 15, No. 2

rink & Bertilson, 1975; Hare & Cox, 1978), In thepresent study, subjects with a rating of 6 or 7 were definedas psychopaths, while those with ratings of 1 or 2 wereconsidered to be nonpsychopaths. Since recent evidence(Hare, 1978) indicates that measures of socialization maybe useful in identifying psychopaths when case historyinfonnation is limited, a measure of socialization wasalso employed.

Four subjects were discarded because of initial prob-lems with the instructions and question structure, onewas actively psychotic, and one failed to conceal the$20 properly. The remaining subjects consisted of 24psychopaths (mean age = 23,1 yrs) and 24 nonpsycho-paths (mean age =26,5 yrs). Half of each group wasrandomly assigned to the "guilty" condition, and halfwas assigned to the "innocent" condition. Within eachcombination of guilt and psychopathy, the 12 subjectswere divided into low and high socialization subgroupsby means of their scores on the Socialization (So) scaleofthe California Psychological Inventory (Gough, 1969),

Procedure

When a subject arrived for the experiment, he was metby an assistant who explained the nature of the experi-ment and administered a number of psychological testsincluding the So scale. Then the assistant flipped a cointo determine whether the subject would be in the "guilty"or "innocent" condition. If the subject was in the"guilty" condition, he was told that there was a $20bill in an envelope in a drawer in a nearby room. Thatroom was normally off-limits to inmates, and a prisonguard was stationed outside the door. The subject wasinstructed to obtain the $20 bill from the envelope whenno one was observing him and to hide it in his pocket.He was informed that he would be taken to another roomand be given a lie detector test concerning the $20, Hewas instructed to deny having taken the $20; and if hesucceeded in being cleared by the polygraph test, hewould win the $20, The "innocent" subjects were in-formed about the "crime" committed by the "guilty"subjects, and they were also instructed to deny havingtaken the money. If they successfully demonstrated theirtruthfulness, they also received a $20 bonus. Thus, allsubjects were instructed to deny having taken the money,and all subjects who produced truthful polygraph chartsreceived a $20 bonus. The $20 represented a considerablesum of money to the subjects, since they earned $,75/dayfor working at a prison job and had no other obvioussources of income.

The subject was then taken to the examination room,where a psychophysiologist (the senior author) trainedin field polygraph techniques administered a control-question polygraph examination. The examiner did notknow whether the subject was "guilty" or "innocent,"or if he was psychopathic. The examination consisted ofa pretest interview followed by physiological recordingsobtained during the test phase. During the pretest inter-view the examiner obtained biographical informationfrom the subject, reviewed the questions with the subject,and explained the physiological measures to be obtainedand the theory underlying the use of the polygraph fordetecting deception.

The polygraph test was a control question test similar

to that employed by Barland and Raskin (1975), It con-sisted of 10 questions including relevant questions atpositions 5,7, and 10 and control questions at positions4, 6, and 9, All of the questions except the control ques-tions were the same for each subject. The control questionswere adjusted to fit each individual, and their wordingwas worked out with the subject prior to administeringthe test so that the subject would answer "no" to allcontrol and relevant questions (see Podlesny & Raskin,1977, for a more complete description of the theoryunderlying control question tests), A typical questionsequence was as follows:

1, Were you bom in Canada?2, Regarding that $20, do you intend to answer truth-

fully each question about that?3, Are you completely convinced I will not ask a

question that hasn't been reviewed?4, Other than what you told me, before you were 18

did you ever steal any money?5, Did you take that $20?6, Did you ever steal anything else from someone

who trusted you?7, Did you take that $20 from the drawer?8, Is your last name ?9, Have you ever taken anything of value from an

employer?10, Do you have that $20 now?Prior to the polygraph test with the above questions,

the subject was attached to the polygraph and a numbertest was conducted in order to adjust the instrument andto demonstrate the effectiveness of the technique to thesubject. He was asked to choose a number between 3 and6, and the number was written on a card and taped onthe wall directly in front of where he was seated. Hewas then told that he would be asked about the numberthat he had chosen and that he was to answer "no" toall of the questions. During the test the subject wasseated facing away from the examiner and the polygraph,and both subject and examiner observed the card with thenumber on it. The examiner then asked a series of ques-tions about the numbers 1 through 7, Following the num-ber test, the subject was informed that everything wasworking well and that the results showed a large reactionto the number he had chosen and very little reaction tothe other numbers. The subject was then told that thetest indicated how his reactions appeared when he waslying and when he was telling the truth, and he was toldthat he had nothing to worry about as long as he truth-fully answered every question on the polygraph test.

The polygraph test concerning the $20 was then ad-ministered. It consisted of a minimum of three chartsobtained while the 10 questions were asked at a rate ofone every 25-35 sec. If the results were not obviousafter the third chart, additional charts were obtained upto a maximum of seven charts. If more than three chartswere obtained, the fourth chart was a silent answer testin which the subject was instructed to answer silentlyto himself (Horvath & Reid, 1972).

Following each chart, the examiner asked the subjectif any questions bothered him and if he would like tochange the wording of any question. The attention of allsubjects was directed toward the control questions, and itwas often necessary to modify the specific wording of one

March, 1978 LIE DETECTION IN PRISONERS AND PSYCHOPATHS 129

or more control questions following admissions or ex-pressions of concern by the subject. The wording of therelevant questions was never modified. Prior to eachchart the subject was cautioned to tell the truth in answer-ing every question. He was told that if he lied to anyquestion, there would be reactions on the chart. Thatprocedure was designed to focus the concern of "inno-cent" subjects on the control questions and to enhancethe concern of "guilty" subjects with regard to the rele-vant questions. After the last chart had been obtained,a field-type numerical evaluation of the charts was per-formed by the examiner (Barland & Raskin, 1975), and adecision was made on the basis of that score as describedbelow.

Apparatus

A Beckman Type R Dynograph was used to record thephysiological activity at a chart speed of 2.5mm/sec. Theamplitude of recordings was adjusted to provide pen ex-cursions of approximately 1.5-3 cm in most cases.Thoracic and abdominal respiration measures were ob-tained from pneumatic tubes positioned around the upperthorax and abdomen. Each pneumatic tube was attachedto a Grass PT5A Volumetric Pressure Transducer, andthe signal from that was fed into a Beckman 9853AVoltage/Pulse/Pressure coupler. Skin conductance (SC)was obtained from electrodes placed on the thenar andhypothenar eminences of the left hand and connected to aBeckman 9842 Galvanic Skin Response Coupler, whichimposed .5V across the electrodes. The recordings weremade using a 22-sec time constant. Skin potential (SP)was obtained from an active electrode placed on thethenar eminence of the right hand and an inactive elec-trode positioned approximately 4 cm below the elbow onthe volar surface of the forearm. The skin beneath theinactive site was vigorously rubbed with an alcohol-dampened, coarse tissue until a local erythema was pro-duced. A time constant of 17 sec was utilized. Cardiacactivity was measured from electrodes attached to theright wrist and left ankle (EKG Lead II) and connectedto a Beckman 9857 Cardiotachometer Coupler. The heartrate (HR) output was sampled on a sec-by-sec basis usinga PDP 8/E online computer. All of the electrodes wereBeckman Biopotential filled with .05M NaCl in a corn-starch paste and attached with Beckman adhesive collars.Prior to electrode application each site was cleaned with95% ethanol. Vasomotor activity was recorded from theleft thumb using a photoelectric plethysmograph con-taining a Fairchild FPA 104 transducer, which consistsof a light-emitting diode and a phototransistor. The trans-ducer was placed in an aluminum cylinder and attachedby means of an adhesive collar and masking tape. Thesignal was fed into a Beckman 9874 Photocell Coupler.Finger pulse amplitude (FPA) was obtained by recordingthe signal with a .1-sec time constant, and finger bloodVolume (FBV) was recorded with a 20-sec time constant.

Quantification of the Data

There were two types of evaluations made on thecharts. The first consisted of on-the-spot, numericalevaluations similar to those reported by Barland & Raskin(1975). The criteria included only those responses whichnave been demonstrated to discriminate between truth

and deception in laboratory studies (Podlesny & Raskin,1977). Specifically, the following criteria were used inscoring responses: Respiration—sustained decreases inamplitude, slowing of rate, increases in baseline, apnea;Skin Conductance—increase in SC, multiple responses,increased duration of response; Cardiovascular—decreasein FBV, decrease in FPA, slowing of HR.

Using the above criteria, the responses to control andrelevant questions were compared for each type of mea-sure (respiration, skin conductance, cardiovascular) oneach chart. The responses to adjacent pairs of controland relevant questions (4-5, 6-7, 9-10) were compared,and a number ranging from +3 to —3 was assigned toeach pair for each type of measure on each chart. Thesign and size of each such score was determined accord-ing to the relative magnitudes of the responses to thecontrol and relevant question in each pair. If the responseto the control question was larger, then a positive scorewas assigned; if the response to the relevant questionwas larger, a negative score was assigned; and a zero wasassigned if there was no diflFerence between the two re-sponses. Whether the response was assigned a value of1, 2, or 3 was determined by criteria used by the UnitedStates Army Military Police School (Note 1). Totalscores for the first three charts and for all charts wereobtained by summing the scores over question pairs,charts, and measures. If the total score for the first threecharts was -1-6 or higher, the subject was judged to betruthful; if it was —6 or lower, the subject was judgeddeceptive; and a score between ±5 was considered in-conclusive. If the score for the first three charts wasinconclusive, the scores from additional charts wereutilized to attempt to arrive at decisions for those sub-jects. That was done for 8 subjects (1 "innocent" psy-chopath, 3 "guilty" nonpsychopaths, and 4 "innocent"nonpsychopaths).

The second type of analysis of the charts consisted ofan objective quantification of responses by a technicianwho had no knowledge of the field evaluations or thetreatments administered to the subjects. Measurementswere made on the responses to each of the three controland three relevant questions on the first three charts ob-tained from each subject. The following scores wereobtained for each of those 18 observations:

1. Thoracic Respiration Amplitude (TRA). The ampli-tude in mm of the last full inspiration preceding thebeginning of the question was subtracted from the ampli-tude of the first complete inspiration following the answerto the question. This provided an index of change in RA,with negative scores indicating decreases in RA and posi-tive scores indicating increases in RA.

2. Abdominal Respiration Amplitude (ARA). Scoresfor this measure were obtained from the abdominaltracing in the same manner as those described for TRA.

3. Respiration Cycle Time (RCT). The distance in mmbetween the points of maximum inspiration for the tworespiration cycles preceding the onset of the question wassubtracted from the distance between the points of maxi-mum inspiration for the two cycles following the answerto the question. This provided an index associated withchange in rate of respiration, with positive scores indi-cating slowing of respiration and negative scores indicat-ing speeding.

130 RASKIN AND HARE Vol. 15. No. 2

4. Thoracic Respiration Baseline (TRB). Deviationsin mm in the baseline were obtained by subtracting thebaseline at the onset of inspiration for the last completeinspiration preceding the beginning of the question fromthe baseline of each of the next five cycles followingquestion onset, excluding the artifacts due to the subject'sanswer. Positive scores indicated rises in respirationbaseline, and negative scores indicated drops in respira-tion baseline.

5. Abdominal Respiration Baseline (ARB). Thesewere obtained from the abdominal tracing in the samemanner as described for TRB.

6. Skin Conductance Response (SCR). The increase inmm was measured from the onset of the first increasewithin 1 sec after the beginning of the question to thehighest point reached within 5 sec after the answer. Thescores were also converted to /Ltmhos. Since the scoresexpressed in mm yielded slightly more reliable results,they are the only ones reported here.

7. Skin Potential Response (SPR). The change in mVwas measured from the onset of the first wave withinthe period beginning 1 sec after question onset and ending5 sec after the answer. If the first wave was a negativewave which was followed by a positive wave, then thechange in mV of the positive wave was measured fromthe highest point reached by the negative wave. Onlywaves which reached their peak within 3 sec followingtheir onset were considered to be positive waves. Thatprocedure was used to eliminate scoring as positivewaves the simple recovery toward baseline following anegative wave (Raskin, Kotses, & Bever, 1969).

8. Finger Blood Volume (FBV). The decrease in mmof the diastolic level of the tracing was measured fromthe highest point within the 4 sec following the onsetof the question to the lowest point within 4-14 sec fol-lowing question onset. Since there was a very large rangeof gain settings across subjects, all scores were correctedto a common gain.

9. Finger Pulse Amplitude (FPA). The decrease inFPA was obtained by calculating the difference in mmof the sum of the amplitudes of the two largest adjacentpulses within 4 sec after the onset of the question andthe sum of the amplitude of the two smallest adjacentpulses within 4-14 sec after the onset of the question.These values were also corrected to a common gain.

10. Heart Rate (HR). Changes in HR were measuredin bpm by subtracting the mean HR for the 3 sec preced-ing question onset from the HR for each of the 20 secfollowing question onset.

Results^

Numerical Evaluations

Accuracy of Decisions. Using an inconclusiveregion of ±5 inclusive, the outcomes were assessedon the basis ofthe total score for the first three chartsand for all charts. Table 1 shows the results for"guilty" and "innocent" subjects and for psycho-paths and nonpsychopaths. Using only three charts,categorizations were 81% correct, 2% wrong, and

statistical tests employed a .05 rejection region.

TABLE 1

Examiner decisions based on total numerical scores for the first3 charts and for all charts

Charts

First 3

All

Groups

"Guilty""Innocent"

"Guilty""Innocent"

First 3

All

PsychopathsNonpsychopathsCombined

PsychopathsNonpsychopathsCombined

Examiner Decisions

Decfp- Truth- Incon-tive ftil elusive

21 0 31 18 5

21 0 32 21 1

Correct Wrong Incon-clusive

22 1 117 0 739 1 8

23 1 019 1 442 2 4

17% inconclusive. Excluding inconclusives, the ac-curacy rate was 98%. When additional charts wereused, 88% were correctly categorized, 4% werewrong, and 8% were inconclusive. Excluding in-conclusives, the accuracy rate was 96%.

Although there were only 2 errors, both of themwere false positives, i.e., "innocent" subjectswhose polygraph charts indicated deception. Withregard to psychopathy, one of the errors occurredwith a psychopathic subject and one with a non-psychopath. It should be noted that no "guilty"subject was able to produce a truthful result. Therewere no significant differences in accuracy rates forpsychopaths and nonpsychopaths.

Using the total score for all charts, a post factomanipulation of the cut-offs of the inconclusiveregion was made for boundaries ranging from zeroto ±12, and the results are shown in Fig. 1 interms of percent accuracy of decisions and percentinconclusives for the "guilty" and "innocent"groups. When the inconclusive region was limitedto scores of zero, 96% of the "guilty" subjectsand 88% ofthe "innocent" subjects were correctlycategorized, and there were no inconclusives. Withboundaries of ±2 there was maximal accuracy ofdecisions combined with a relatively low rate ofinconclusives. As the boundaries of the inconclu-sive region were widened, there was no improve-ment in accuracy of decisions and the percent ofinconclusive cases eventually increased dramati-cally. It appears that the optimal boundaries aresomewhere in the region of ±2 to ±4. At any

March, 1978 LIE DETECTION IN PRISONERS AND PSYCHOPATHS 131

100- • GUILTY-D£]CEPTIVE

INNOCENT-TRUTHFUL

INNOCENT-INCONCLUSIVE

GUILTY-WCONCLUSIVE

BOUNDARIES OF INCONCLUSIVE REGION

Fig. 1. Accuracy of decisions and percent inconclusives fordifferent inconclusive regions.

point in that region there were no false negatives,4.2% false positives, and 6.2% inconclusives.

Effectiveness of Numerical Scores. Since all sub-jects had at least three charts, the total numericalscores for the first three charts were compared for"guilty" and "innocent" subjects using the fourcombinations of psychopathy and socialization.The mean scores for those groups are shown inTable 2. Analysis of variance indicated that themean scores of -11.1 for all "guilty" subjectsand +9.4 for all "innocent" subjects producedreliable discrimination between those conditions(f(l/40)=83.21, M5e=60.85). There was no re-liable difference in the magnitude of scores forpsychopaths and nonpsychopaths (F<1), nor wasthere any difference in the magnitude of scores for"guilty" and "innocent" subjects (F<1). How-ever, low socialization subjects showed signifi-cantly higher magnitude of scores than high so-cialization subjects (7^(1/40)=6.96, MS^=52.97).Thus, psychopaths and nonpsychopaths showed

equally appropriate responses, the numerical sys-tem identified "guilty" and "innocent" subjectswith equal power, but differentiation between"guilty" and "innocent" subjects was better forlow socialization as compared to high socializationsubjects.

Physiological Measures. The scores for each ofthe three types of physiological measures weresummed over the first three charts, and the meantotal scores for "guilty" and "innocent" subjectswithin the four combinations of psychopathy andsocialization are shown for each measure in Table2. The mean total scores for respiration showedsignificant discrimination between "guilty" and"innocent" subjects (F(1/40)=63.34, M5e = 8.42),as did the mean total scores for skin conductance(F(l/40)=41.83, MSe=25.m and cardiovascular(F(l/40)=25.37, MSe = 8.88). There was a signifi-cant interaction between "Guilty-Innocent" andphysiological measures (F(2/80)=7.08, MSe —11.35), and a Newman-Keuls test showed that theskin conductance measure produced better dis-crimination between "guilty" and "innocent" sub-jects than did the respiration and cardiovascularmeasures, and the respiration measure producedbetter discrimination than the cardiovascular mea-sure. Furthermore, the respiration measure iden-tified "innocent" subjects significantly better than"guilty" subjects (F(1/40)=5.70). The only sig-nificant effects related to psychopathy or socializa-tion were found with the cardiovascular measure.A significant Socialization X "Guilty-Innocent"interaction (F(1/40)=5.86) indicated that high so-cialization-" innocent" subjects showed no differ-entiation between control and relevant questionson the cardiovascular measure.

TABLE 2

Measures

Respiration

Skin Conductance

Cardiovascular

Total Score

Mean scores for

Groups

"Guilty""Innocent"

"Guilty""Innocent"

"Guilty""Innocent"

"Guilty""Innocent"

each measure and total scores for

Psychopaths

Low So

-2 .5-f-6.7

-7 .2-1-3.5

-5 .0-1-1.5

-14.7+ 11.7

High So

-2.5-1-2.3

-6 .2+ 5.8

-2 .8-0 .5

-11.5+7.7

the first 3 charts

Mean Scores

Nonpsychopaths

Low So

-2 .5+4.7

-6 .7+5.5

-2 .3+4.0

-11.5+ 14.2

High So

-1.8+ 3.7

-3 .0+0.2

-1.8+0.3

-6 .7+4.2

Combined

-2 .3+4.3

-5.8+3.8

-3 .0+ 1.3

-11.1+9.4

132 RASKIN AND HARE Vol. 15. No. 2

Quantitative Analyses

The first three polygraph charts for each subjectwere analyzed using the procedures previouslydescribed, and each of the dependent variableswas subjected to analysis of variance. Using thecontrol question technique, discrimination between"guilty" and "innocent" subjects would be indi-cated by a significant interaction between "Guilty-Innocent" and Question Type. That prediction isbased on the principle that deceptive subjects shouldshow larger responses to the relevant questions andtruthful subjects should respond more strongly tothe control questions.

The mean responses to the control and relevantquestions for "guilty" and "innocent" subjectsare shown for several dependent variables in Table3. The predicted interactions were obtained for alldependent variables except respiration cycle time.The results of the F tests on the interactions arepresented in Table 3, and details of the results forthe various physiological measures are describedbelow. Except as noted below, there were no differ-ential effects related to psychopathy or socializa-tion.

Respiration Responses. For thoracic and ab-dominal respiration the "guilty" subjects showedrelatively more reduction in amplitude to the rele-vant questions, and the "innocent" subjects showedrelatively more reduction in amplitude to controlquestions. The interaction for respiration cycle timewas not significant. Further analyses indicated thatresponses of "innocent" subjects were in oppositedirections to control and relevant questions in mea-sures of thoracic respiration amplitude, r(44)=2.18, and respiration cycle time, /(44)=2.78. The"innocent" subjects showed a decrease in ampli-tude and a slowing of rate in response to controlquestions and an increase in amplitude and a speed-ing of respiration rate in response to relevant ques-tions. Similar results were obtained with abdominal

respiration baseline, and they are shown in Fig. 2There was a significant "Guilty-Innocent" xQuestion Type X Cycles interaction (F(4/128)=3.49, M5p = 1.36). An inspection of Fig. 2 revealsthat the "guilty" subjects showed a larger rise inabdominal respiration baseline in response to rele-vant as compared to control questions, whereasthe "innocent" subjects showed greater reaction tothe control questions. In addition, the "innocent"subjects showed a drop in baseline in response torelevant questions.

Electrodermal Responses. For SCR, the "guilty"subjects produced relatively larger reactions to rele-vant questions, and the "innocent" subjects pro-duced relatively larger reactions to the controlquestions. Overall SCR was greater to relevant thanto control questions (F(l/44)=9.40). With mea-sures of positive and negative skin potential, the"guilty" subjects showed differentially larger reac-tions to relevant questions, but the "innocent"

GUILTY-RELEVANT

GUILTY-CONTROL

INNOCENT-CONTROLINNOCENT-RELEVANT

PRE

Fig. 2. Changes in abdominal respiration baseline producedby control and relevant questions in "guilty" and "innocent"subjects.

TABLE 3

Groups

•"Guilty"

""inrKK'ent"

F(l/44)

Mean Square

Mean respiration, electrodermal, and cardiovascular responses to control andfor "Guilty" and "Innocent" subjects

Questions

ControlRelevant

ControlRelevant

Error

Thoracic-RespirationAmplitude

r

- 1 . 3

.6

9.77

5.09

AbdominalRespirationAmplitude

-1.3-1 .9

-1 .2- .1

8.47

6.54

RespirationCycleTime

0t

.2- .1

2.69

.41

Means

SkinConductance

10.716.7

12.710.2

34.52

36.23

NegativeSkin

Potential

.4

.6

.6

.6

12.01

.07

relevant

PositiveSkin

Potential

.5

.8

.6

.7

7.56

.13

questions

FingerBlood

Volume

5.66.7

6.45.8

4.15

.13

FingerPulse

Amplitude

4.96,2

3.83.9

7.53

3.70

March, 1978 LIE DETECTION IN PRISONERS AND PSYCHOPATHS 133

subjects did not respond diflferentially to the twotypes of questions. However, the psychopaths pro-duced disproportionately larger positive skin po-tential responses to the relevant questions (F(l/44)=:5.98), and both negative and iK)sitive SPR wasgreater to relevant as compared to control questions(f (l/44)=8.04 and 22.44, respectively).

Since hyporesponsivity of psychopaths has oftenbeen obtained with SCR, an additional analysis wasperformed by utilizing extreme subgroups selectedby a combination ofthe ratings of psychopathy andsocialization scores. An analysis which includedonly the low socialization-psychopaths and the highsocialization-nonpsychopaths also produced a sig-nificant interaction of "Guilty-Innocent" withQuestion Type (F(1/20)=12.49, MSe= 35.16). ANewman-Keuls test indicated that the low socializa-tion-psychopaths in the guilty condition showedsignificantly stronger differentiation between con-trol and relevant questions (Af=7.4) than did theother three groups (M<3.1). Thus, there was someindication that the most extreme group on thepsychopathy dimension showed stronger electro-dermal diflFerentiation when they were deceptive tothe relevant questions.

Cardiovascular Responses. For finger blood vol-ume and pulse amplitude, the "guilty" subjectsshowed relatively larger decreases to relevant ascompared to control questions. However, the "in-nocent' ' subjects produced diflferentially larger de-creases to control questions only in the bloodvolume measure, and decreases in pulse amplitudewere greater to relevant as compared to controlquestions (F(1/44)=9.64).

Due to equipment problems, HR data were avail-able for only 10 psychopaths and 7 nonpsychopaths

in both the "guilty" and "innocent" conditions.The sec-by-sec HR for psychopaths and nonpsycho-paths are presented in Fig. 3. The overall form ofthe HR response consisted of an initial accelerationwhich reached a peak about the time of the sub-ject's answer, and HR then returned to prestimuluslevel or below (F(19/570)=25.58, MSe = l6.4S).There was a significant "Guilty-Innocent" xQuestion Type X Seconds interaction (F(19/570)=4.27, Af5e=9.84). The "guilty" and "innocent"subjects showed an initial acceleration followingthe onset of both types of questions. However,the HR of "innocent" subjects then returned toprestimulus level whereas the "guilty" subjectsshowed a mean deceleration of 3.7 bpm followingthe answer to relevant questions. Furthermore, the"guilty" psychopaths showed slightly less ac-celeration (3.4 bpm) than the "guilty" nonpsycho-paths (5.5 bpm) to both types of questions, andthey showed stronger deceleration (—4.2 bpm) thanthe "guilty" nonpsychopaths (—3.2 bpm) to therelevant questions (F(19/570)=2.33).

Discussion

The results of this study clearly demonstrate theeflfectiveness of the control-question technique indetecting deception employing a mock-cdme para-digm with a population of prison inmates. Usingthe numerical -scoring system the accuracy of de-cisions was 95.5%, and only 8.3% of the testsyielded inconclusive outcomes. Those results pro-vide support for the usefulness of the numericalscoring system developed and employed in fieldsituations.

Since the examiner who numerically scored thecharts had direct contact with the subjects and his

UI

GUILTY-PSYCHOPATH

O Control

INNOCENT-PSYCHOPATH

PRE 2 4 6 8 10 12 14 16 le 20

UJXzUJ

o- 4 -

GUILTY-NONPSVCHOPlWrH

JMbQ OControl

• I I > I I I I I 1 — I — I

PRE 2 4 6 8 10 12 14 16 » 20

6

4

2

0

-2

- 4 -

O Control

6

4

2

0

-2

-4

PRE 2 4 6 8 1 0 1 2 1 4 1 6 1 8 2 0

INNOCENT-NOI«>SYCHOPATH

O ControlRiJovont

J _LPRE 2 4 6 8 K) 12 14 16 e 20

SECONDS^ig. 3. Heart rate changes prcxluced by control and relevant questions in "guilty" and "innocent" psychopaths and non-

Psy ehopaths.

134 RASKIN AND HARE VoL 15, No. 2

scoring may have been influenced by the demeanorof the subjects, a question arises about the extentto which the accuracy of the results was basedsolely on the physiological recordings. Althoughthe present study does not resolve that question,previous research in our laboratory has shown thenumerical scoring technique to be highly reliableand minimally affected by contact with the subject(Raskin et al., 1977). That paper includes threestudies in which the fK)lygraph charts were numeri-cally scored by one or more examiners who had nocontact with the subjects and were blind with regardto their guilt or innocence. Using polygraph ex-aminations from 102 criminal suspects, a correla-tion of .91 was obtained between the numericalscores assigned by the original examiner and theblind scorer, and there was 1(X)% agreement on de-cisions. Similar procedures in a laboratory ex-periment obtained 1(X)% agreement between thedecisions of the original examiner and those of theblind scorer. The third study obtained 99% accu-racy from seven examiners who employed numeri-cal scoring in the blind interpretation of 16 setsof polygraph charts from confirmed criminal cases.On the basis of those studies it appears that thenumerical scoring technique is a highly reliable andaccurate method for determining the outcome ofthe polygraph test and is relatively unaffected bycontact with the subject. Of course, adequate train-ing and experience are necessary to employ thattechnique successfully.

The results provided substantial confirmation ofpredictions generated from the theory of control-question polygraph tests (Podlesny & Raskin,1977). Although Lykken (1974) argued that thecontrol question cannot function as proposed, "in-nocent' ' subjects showed substantially larger reac-tions to control questions in a variety of respiration,electrodermal, and cardiovascular measures usingfield numerical scoring and blind quantitative anal-yses. On the basis of numerical evaluation, "in-nocent" subjects produced scores which were asstrongly indicative of truthfulness as the scores of"guilty" subjects were indicative of deception. Thefailure to support Lykken's (1974) analysis is notsurprising since he erroneously stated that controlquestions are designed to be answered truthfully(p. 729) and he erroneously assumed that a lack ofdifference in responses to control and relevant ques-tions yields a truthful outcome.

The effectiveness of the control question for aninnocent subject is crucial to the accuracy of acontrol-question polygraph test. Since the wordingof each control question must be selected by theexaminer and individually adjusted according to thesubject's answer to the question, the training andskill of the examiner play an important role in de-

termining the accuracy of the outcome. Further-more, the manner in which the examination is con-ducted and the way in which the examiner interactswith the subject can also influence the outcome.In the present study, procedures were utilized toenhance the effectiveness of control questions byconstantly drawing attention to them between suc-cessive polygraph charts. That appears to be aneffective method which has been used successfullyin a subsequent experiment (Raskin et al., 1977).

The findings of the present study do not supportthe belief that detection of deception techniquesare ineffective with psychopaths. Using a controlquestion test, not a single ' 'guilty" subject was ableto produce a truthful result. In a variety of physio-logical measures psychopaths were as responsive asnonpsychopaths, and they showed evidence ofstronger reactivity in measures of HR deceleration,positive skin potential, and skin conductance. Theuse of the socialization measure to attempt to selecta more extreme group of psychopaths did not alterthe basic findings and provided additional evidenceof stronger differentiation among low socializationsubjects.

In some respects, these findings appear to be inconflict with previous research in which psycho-pathic inmates were found to be electrodermallyhyporesponsive to a variety of physical stimuli(Hare, 1978). However, there is evidence that suf-ficiently aroused or motivated psychopaths are nothyporesponsive. For example. Hare (1968) re-ported that skin conductance fluctuations of psy-chopaths engaged in solving arithmetic problemsunder time pressure were similar to those of non-psychopaths. Schmauk (1970) found that prison in-mates with high MMPI Pd scores and low anxietyscores gave relatively small anticipatory electro-dermal responses in an avoidance task involvingelectric shock or social disapproval, but theirelectrodermal responses equalled those of the non-psychopaths when the consequence was loss ofmoney, as in the present experiment.

The results of this study appear to indicate thatcontrol question tests may be used effectively withpsychopaths in the field situation. Since such testsemploy cardiovascular and respiration measures onwhich psychopaths have not been shown to behyporesponsive and differential responsivity tocontrol or relevant questions is required to arrive ata decision, the risks in the field situation seem tobe minimized by the control question test. How-ever, the present study was conducted in the con-text of a mock crime, and the extent to which thatsituation generalizes to the real-life situation is opento some question. If the field situation is motiva-tionally more similar to aversive physical stimula-tion than to the loss of money utilized in this

March, 1978 LIE DETECTION IN PRISONERS AND PSYCHOPATHS 135

Study, then psychopaths might be electrodermally dermal measures and finger pulse amplitude showedhyporesponsive and more difficult to detect in de- reliably larger responses to relevant questions. Theception. If the loss of money is motivationally simi- findings raise some cautions for interpreting thoselar to the potential loss of freedom inherent in the measures since they were more effective withfield situation, then there should be no particular "guilty" than with "innocent" subjects. The twoproblem of generalizability. It is also possible that "innocent" subjects whose polygraph charts erron-psychopaths are especially challenged and threat- eously indicated deception showed almost all ofened by the lie detection procedures. Since they their inappropriate reactions in the electrodermalhave a life style which is characterized by insin- and cardiovascular measures, and one of themcerity and manipulation, the prospect of being showed respiration changes associated with truth-evaluated on the basis of their involuntary bodily fulness. It appears that the small risk of errors inreactions measured by an inanimate instrument may control question tests is associated with inappro-be a unique and threatening experience which pro- priate electrodermal and cardiovascular responsesduces a high degree of reactivity. to relevant questions in "innocent" subjects, and

The only data available on the effectiveness of the presence of respiration responses in the truthfulcontrol-question polygraph examinations with psy- direction should create some suspicion of a possiblechopathic criminal suspects in the field situation false positive.are those reported by Raskin et al. (1977). Their Finally, the finding of a pronounced accelerationresults also seem to indicate that psychopaths pose of heart rate preceding the answer by "guilty" andno special problem for detection with control- "innocent" subjects to both types of questions fol-question polygraph examinations. However, the lowed by a strong deceleration to relevant questionslimitations of that study prevent it from providing in "guilty" subjects raises some interesting ques-a definitive answer to the question. tions about the type of responses being manifested.

The data from the present study provided addi- On the basis of Obrist's concept of cardiac-somatictional information on the utility of various physio- coupling (Obrist, Howard, Lawler, Galosy, Meyers,logical measures in detection ofdeception as well as & Gaebelein, 1974), it seems reasonable to con-some specific response characteristics associated elude that the initial acceleration is a result of thewith truth and deception. Although all measures subject's preparation to answer and his act of an-significantly discriminated between "guilty" and swering. However, the subsequent deceleration ob-"innocent" subjects, numerical evaluation showed served in "guilty" subjects' responses to relevantthat skin conductance discriminated better than res- questions appears to be more of an attentional re-piration and cardiovascular measures. However, sponse which may be associated with an attem'pt torespiration scores provided the strongest indicator obtain information from the environment (Laceyfor "innocent" subjects. Further analyses of the & Lacey, 1974; Porges & Raskin, 1969). This latterrespiration measures provided evidence that there finding has been replicated in our laboratory (Ras-are changes which go in opposite directions for kin et al., 1977) and is consistent with that re-tmth as compared to deception. When "guilty" ported by Ellson, Davis, Saltzman, and Burkesubjects were deceptive to the relevant questions, (Note 2). It may be produced by the subject's at-they showed decreases in thoracic respiration am- tempts to pick up cues regarding the extent to whichplitude and increases in abdominal respiration base- the examiner has observed a reaction produced byline. However, when "innocent" subjects were deception to that question. If that decelerationtruthful to the relevant questions, they showed in- represents an attentional response by the subject,creases in thoracic respiration amplitude, decreases the extent to which the subject is concemed aboutin abdominal respiration baseline, and speeding of the responses which his body is displaying may playrespiration rate. No other physiological measures an important role in producing high rates of ac-have shown characteristics which change in oppo- curacy in detecting deception. Leading the subjectsite directions depending on whether the subject is to believe that any attempt to deceive will be mani-being deceptive or truthful (Podlesny & Raskin, fested in the polygraph recordings is a typical field1977). procedure (Barland & Raskin, 1973), and the heart

All of the other measures showed reliable effects rate deceleration observed in this study lends somein the predicted directions. However, the electro- support to its usefulness.

REFERENCES

Barland, G, H,, & Raskin, D. C. Detection of deception. In Barland, G. H., «& Raskin, D. C. An evaluation of field tech-W- F. Prokasy & D, C, Raskin (Eds.), Electrodermal ac- niques in detection of deception. Psychophysiology, 1975.^i'^ity in psychological research. New York: Academic Press, 12, 321-330.1973. Pp. 417-477. Bersh, P. J. A validation study of polygraph examiner judg-

136 RASKIN AND HARE Vol. 15, No, 2

ments. Journal of Applied Psychology, 1%9, 53, 399-403.Chesno, F. A., & Kilmann, P. R. EflFects of stimulKtion inten-

sity on sociopathic avoidance learning. Journal of AbnormalPsychology, 1915,84, 144-150.

Clancy, J., Noyes, R., & Travis, T. A. The psychopathic per-sonality. Postgraduate Medicine, 1974,55(4), 175-180.

Cleckley, H. The mask of sanity. St. Louis, Mo.: Mosby, 1964.Dengerink, H. A., & Bertilson, H. S. Psychopathy and physio-

logical arousal in an aggressive task. Psychophysiology,1975,12, 682-684.

Ferguson, R. J., & Miller, A. L. Polygraph for the defense.Springfield, Illinois: Charles C Thomas, 1974.

Floch, M. Limitations of the lie detector. Journal of CriminalLaw and Criminology, 1950, 40, 651-652.

Gough, H. G. Manual for the California Psychological Inven-tory, Palo Alto, Califomia: Consulting Psychologists Press,1969.

Hare, R. D. Psychopathy, autonomic functioning and the orient-ing response. Journal of Abnormal Psychology, Monograph,1968, 73(1),

Hare, R. D. Electrodermal and cardiovascular correlates of psy-chopathy. In R. D. Hare & D. Schalling (Eds.), Psychopathicbehavior: Approaches to research. Chichester: Wiley, 1978.Pp. 107-144.

Hare, R. D., «& Cox, D. N. Clinical and empirical conceptionsof psychopathy and the selection of subjects for research.In R. D. Hare & D. Schalling (Eds.), Psychopathic behavior:Approaches to research. Chichester: Wiley, 1978. Pp. 1-22.

Horvath, F. S., & Reid, J. E. The polygraph silent answertest. Journal of Criminal Law, Criminology and PoliceScience, 1972. 63, 285-293.

Kugelmass, S., Lieblich, I., Ben-Ishai, A., Opatowski, A., &Kaplan, M. Experimental evaluation of galvanic skin responseand blood pressure change indices during criminal investiga-tion. Journal of Criminal Law, Criminology and PoliceScience, 1968,59, 632-635.

Lacey, B. C , & Lacey, J. 1. Studies of heart rate and other

bodily processes in sensorimotor behavior. In P. A. ObristA. H. Black, J. Brener, & L. V. DiCara (Eds.), Cardio-vascular psychophysiology—Current issues in responsemechanisms, biofeedback, and methodology. Chicago: Al-dine, 1974. Pp. 538-564.

Levitt, E. E. Scientific evaluation of the "lie detector." IowaLaw Review, 1955,40, 440-458.

Luria, A. R. The nature of human conflicts. New York: Live-right, 1932.

Lykken, D. T. ^ study of anxiety in the sociopathic personality(Doctoral dissertation, University of Minnesota). Ann Arbor,Mich.: University Microfilms, 1955, No. 55-944.

Lykken, D. T. Psychology and the lie detector industry. Ameri-can Psychologist, 191 A, 29, 724-739.

Marston, W. M. Psychological possibilities in the deceptiontests. Journal of Criminal Law and Criminology, 1921, II,551-570.

Obrist, P. A., Howard, J. L., Lawler, J. E., Galosy, R. A.,Meyers, K. A., & Gaebelein, C. J. The cardiac-somaticinteraction. In P. A. Obrist, A. H. Black, J. Brener, &L. V. DiCara (Eds.), Cardiovascular psychophysiology—Current issues in response mechanisms, biofeedback, andmethodology. Chicago: Aldine, 1974. Pp. 136-162.

Podlesny, J. A., & Raskin, D. C. Psychological measures andthe detection of deception. Psychological Bulletin, 1977,84,782-799.

Porges, S. W., & Raskin, D. C. Respiratory and heart ratecomponents of attention. Journal of Experimental Psychol-ogy, 1969, 81, 497-503.

Raskin, D. C , Barland, G. H., & Podlesny, J. A. Validityand reliability of detection of deception. Polygraph, 1977,6, 1-39.

Raskin, D. C , Kotses, H., & Bever, J. Autonomic indicatorsof orienting and defensive reflexes. Journal of ExperimentalPsychology, 1969, 80, 423-433.

Schmauk, F. J. Punishment, arousal, and avoidance learningin sociopaths. Journal of Abnormal Psychology, 1970, 76,325-335.

REFERENCE NOTES

1. United States Army Military Police School, PolygraphBranch. Chart interpretation (CI5516). Fon Gordon, Geor-gia: USAMPS, 1970.

2. Ellson, D. G., Davis, R. C , Saltzman, I. J., & Burke,

C, J, A report of research on detection of deception (Tech.Rep. Contract N6onr-18011). Bloomington, Indiana: IndianaUniversity, Department of Psychology, 1952.

(Manuscript received April 27, 1976; revision received August 11, 1977; accepted for publication October 25, 1977)