PSYCHOLOGI ICAL SCIENCE Researcl h Report Paul Ekman,' Maureen O'S: "Department of Psychiatry, University of California. San ran ‘and ‘Department of Communicarion, Ru A FEW CAN CATCH A LIAR ullivan,? and Mark G. Frank? isco: Department of Psychology. University of San Francisco: ingers-The State University of New Jersey Abstruct—Research suggests shat most people caret tell from ‘demeanor when others are lying. Suck poor performance i typical not ‘nny of laypeople but also of most professionals concerned with Ising In thie sues, vee professional groups with special interest or skill in deception, ro law-enforcement groups and a select group of clinical | psveliologists, obtained high accuracy in judging videotapes of people ‘who were lying or telling the truth about thelr opinions. These findings strengthen earlier evidence that some professional lie catchers are highly accurate, and that behavioral clues to lying are detectable in real time, This study also provides the frst evidence that some psy chologsts can achieve high accuracy in catching les ‘Most research on how well people can identify lies from demeanor has focused on college students; a few studies have examined a single law-enforcement group. Almost all have found that accuracy is close to chance (see DePaulo, 1994, 1998; DePaulo, Stone. & Lassiter. 98S; and Zuckerman & Driver, 1985, for reviews). Ekman (1992) suggested two reasons for these results. First the stakes for success oF failure in lying were quite low in most ofthese studies, and therefore ‘emotions that could beray a lie—Fear, guilt, oF excitement—were not likely to be strongly aroused. There may have been no signs ofthese emotions about lying that could have contradicted the liars” verbal claims. Without these emotional resctions interfering with thought ‘processes it is easier forthe Har to assemble words nto a credible fab- Fication, Consistent with this reasoning, DePaulo and Kirkendol (1989) found more motivated Hiars were more easily detected “The second explanation of why high accuracy was not found isthe possibility that the liars and truth tellers may not have behaved very tiferently. Most studies of observers’ accuracy in detecting deceit have not included behavioral analyses of how many of the subjects shown in the videotapes actually provided clues to deceit in their face, body, voice, of speech when they led In a previous study (Ekman & O'Sullivan, 1991), we remedied these problems by showing videotapes known to contain behavioral clues that could provide the basis for accurate deception judgments (Ekman, O'Sullivan Friesen, & Scherer, 1991). We showed the video | tapes to more law-enforcement personnel (a total of 330), and! more different law-enforcement groups (seven), than had previously been ‘examined in any one study. Yet we Found that only one group, the US, ‘Secret Service, was quite accurate. A skeptic could argue tha this was ‘random occurrence ‘The primary purpose ofthe current study was to determine whether ‘other occupational groups could also achieve high levels of accuracy in detecting deceit from demeanor. We examined not only law- enforcement groups who differed in their imerest in detecting decep- | Son and in their seuation for doing so, bu aso groups of ‘Address correspondence 10 Paul Ekman, Depanment of Psychiatry. Ui ‘ers of California, Sn Francisco, 401 Paras Ave, San Frain, CA, 914.0984, VOL. 10,NO.3, MAY 1999, | psychologists who might differ in thei interest in and knowledge about this topic. ‘The second purpose of our study was to generalize our previous Finding dot very accurate judgments are possible by using ale differ cent from the one we studied before, In the previous work, we studied a lie about emotions; subjects claimed to have positive feelings when in fact they were experiencing very strong negative emotions. The cur- rent study wsed a deception scenari in which subjects lied or told the teuth about strongly held opinions. ‘THE DECEPTION JUDGMENT TASK ‘We used an adaptation (Frank & Ekman, 1997) of Mehrabian's (1971) false-opinion paradigin. Twenty males (ages 18-28) were asked the strength of ther opinions on anumber of current controversial socal issues. The opinion about which each subject felt most strongly was the cone he or she was then asked to discuss with an interrogator. Some sub- jets were told to describe thie opinions tuthully others were told to falsely claim to hold the position opposite their true opinion. Truth | tellers who were believed by the interrogator received a $10 bonus: irs ‘who were believed received a $50 bonus: las or ruth tellers who were isbelieved received no money, and half of them faced an additional punishment. (See Frank & Ekman. 1997, for more details.) ‘To verify that the subjects did manifest different behaviors when Iying versus telling the tuth, we analyzed facial muscular movements. using the Facial Action Coding System (Ekman & Friesen, 1978). This ‘analysis, whieh codes all discernible facial movements, verified that there were significant behavioral differences beween the subjects who lied and those who told the trath (Frank & Ekman, 1997). ‘In our previous study (Ekman & O'Sullivan, 1991), we found that a test comprising I-min samples ofeach of 10 different people discrimi nated among various law-enforcement groups. We constructed a similar task agin showing 10 differen people, half of whom were truthful and half of whom were lying about their opinion, The 10 subjects selected for inclusion on the videotape were chosen so that both pro and con positions on each opinion were represented, independent of whether the subject was lying oF telling the truth. The videotape showed face-and= shoulder close-ups with full audio. The interrogator could be heard but ‘not seen. (See Frank & Ekman, 1997, for more procedural details.) ‘Observers were given about as much information about the decep- ‘ion scenario as has been provided bere. and were told that between ‘one fourth and three fourths of the men were lying. Observers made their judgments during a 10-s interval after viewing each man, LAW-ENFORCEMENT GROUPS Four law-enforcement groups completed the deception judgment task asthe first part of a workshop on deception. Two of these groups ha special interest or experise in the area of deception. The federal officers group included 23 officers who had been chosen by their agency | co participate in adaylong workshop (taught by Ekman) because oftheir Copyright © 1989 American Pychological Society 263PSYCHOLOGICAL SCIENCE = A Few Can Catch a Liar special interest and experience in deception and demeanor, Most of these officers were ftom the Central Intelligence Agency (CIA): the remainder ceme from other federal law-enforcement agencies. The sheriffs group included 43 Los Angeles County sherifl who had been Identified by their department as oustanding interrogators. The 10 ‘other law-enforcement groupe ad less interest or experience in detect- ing deception, The mixed la-enforcement officers group included 36 ‘municipal, state, and federal lav-enforcement personnel who had not been chosen because oftheir reputation as interogators. The last group. {federal judges, inchided 84 judges who attended a 90-min presentation ‘on deception part of « 3-day federal judiciary program that did not focus on deception, except for this one presentation, Table I presents demographic information about these four groups "The mean accuracy scores and standard deviations for each ofthe four law-enforcement groups are given on the left side of Table 2. ‘Additionally, the percentage of exch group who obtained fow, average, ‘or high scores is given in the middle section of Table 2. A one-way fanalysis of variance (ANOVA) on the accuracy scores forthe four groups was computed, There was a significant between-groups eect FFG, 181) = 11.4, p< 001. A Scheffé procedure showed tat the fed cral officers were significantly more accurate than the Federal judges ‘andthe mixed law-enforcement group, but not significantly more accu | tate than the sheriffs The sherifs and the federal judges both were sig nificantly more accurate than the mixed law-enforcement group, PSYCHOLOGIST GROUPS As with law-enforcement groups, we sought to study psychologists who varied in ther intrest in detecting deception. The frst group was 107 fll-time, practicing clinical psychologists who showed an unusu- a interest in deception, by virtue of their deciding to atend a 2-day workshop on “Lying, Deception and Malingering.” held all day on. both a Friday and a Saturday. We presume this group had a special Jnetest in deception because they’ elected to invest 2 days on this spe cifie topic, losing income in addition to the costs of attending the meeting. (Later we were told that many of these psychologists do at feast some forensic work.) We refer to this group as deceprion- interested clinical psychologists. | The second group of psychologists atended a I-hr plenary session on deception, pat of a 4-day national conference on brief prycho- | therapy, We excluded from the data analyses those who didnot havea | Ph. lial psychology and those who did not do flsime ei | sw eving san of 208 rear nel pct, Aa 8 froup, they had no special intrest in deception: they tended this Eonference to learn about other matters, not about deception, As ful time, practicing clinical psychologists, however. they should have teen interested and experienced in interpreting discrepancies i inter- view bebavir “Atha group consisted of academe psychologists who bad tend ca Tehrivited address deception as pat of the 1996 Western Psychological Association Convention. Excluding those who sai they {id any clinical work left 125 academic psychologists. We expected this group to-do less well than the other two psychologist groups ‘ecause most of the psyebologats in tis group didnot do interview ing asa major part of their work, ‘Tuble | presents demographic information about the psychologist ‘10ups, and te bottom half of Table 2 gives the mean accuracy scores, Standard deviations, and perentage distributions for these groups. We computed s one-way ANOVA on the total accuracy sere forthe three groups and found a significant beween-groups effect, F2, 439) 21231. p< 0001. A Schellé procedure showed thatthe decepion- imerested clinical psychologists diflered from both the eguae clinical ychelogist andthe academic psyhologiss The regula clinics psy Chologiss were also more acura than the academic psychologists LYING VERSUS TRUTHFULNESS, “Table 2 also shows that forthe most aecurate groups, performance | wae ar te seen inthe judgments of the less accurate groups. A mukivarte tnalyss of vaniance in which the dependent variables were accuracy for lies and accuracy for traths was significant, F6, 620) = 15.67 p< (00. A Seheffé analysis suggested tha there were no diferenees mong the groups in accurecy for truths, but that ferences among the group in acuray for lies were the same a the differences found for | oul accuracy Te federal ofcers were more accurte onthe lies than Table 1. Demographic characteristics of the seven professional groups Experience Percentage A or kil Group ” ‘female Men 3D Mea SD Federal officers 2 3 48 79 3649 Sheritis 8 ° 407 61 3-66 Federal judges & 18 24 75 Soo Mined law-enforcement officers 36 n us 78 8306 Deception-interested ‘clinical psychologists 107 3 496 93 280s. Regular clinical psychologists 209 48 491 101 270s Academic psychologists 1s a Is = = 392 ‘hil and experince (U= Binning, 2= ited, 3 = advanced) 264 “Forte ln-enfocement groups, the measure was years of ob experience. For psychologists, the measure was selatings of therapeutic VOL, 10, NO.3, MAY 199PSYCHOLOGICAL SCIENCE, Paul Ekman, Maureen O'Sullivan, and Mark G, Frank ‘Table 2. Mean accuracy scores and the percentage ofeach group achieving them Percentage achieving ‘Accuracy by | accuracy level type of item Accuracy “O30 40-60% TOTOOR «= Lie i Group Mean SD” accuracy accuracy accuracy Mean SD. Wen SD Federal officers Bo B60 26 4 300 159 66.1 164 Sherif co ry a 56 TnI 176 338° 212 Federal judges 20 49 6 30 “4 609 178 1 182 Mixed law-enforcement officers 08 178 19 58 n 478 276 539 196 Deception-interested clinical psychologists os 134 1 2 56 70 181 9 149 Regular clinical psychologists. 621.1524 2 44 643 198 598 18.1 Academic psychologists S17 1507 4 2 370 1841 584197 the truths (121] = 3.81, p <_001) as were the sheriffs (1142) = 6.12, P< 0001), the deception-inerested clinical psychologists (1[106} {63,p-<.0001), andthe regular clinical psychologists (208) = 2.86 <.008), DISCUSSION ‘We have shoven that itis possible for some people to make highly ccurate judgments about lying and truthfulness without any special Aids such as slowed mation, repeated viewing, and the scoring of sub- ile changes by either trained coders or computer-based measurements. Such fine-grained behavioral measurements (Ekman eta. 1991) had previously revealed differences between people who lied and those | ‘ono told the truth, but we did not know whether these differences ‘would be noticeable when the videotapes were seen once in realtime, much as real-life judgments are made. In our previous study (Ekman ‘& O'Sullivan, 1991), only one law-enforcement group—e small sam- ple of U.S. Secret Service agents—made accurate judgment. To that study, we speculated thatthe Secret Service agents” superi- ‘or performance was due, in part, to elements of their job taining, experience, and interest. Other occupational groups might also have either interest or expertise in detecting deception. When we selected law-enforcement groups that varied with respect their special inter- cst in deception, or their reputed expertise at doing so, and contrasted them with less interested of less exper groups, we found support for this reasoning. We now have firm evidence that accurate judgments about truthfulness and lying ean be made when a videotape is viewed ‘once, in rel time. Not everyone, however, ean make such accurate judgments; most ofthe unselected groups we studied did poorly The fact thatthe most accurate groups did especially well in judg- ‘ng the ies compared with the truths cannot be attributed simply to & bias to identify more ofthe subjects as liars, Because all groups were {old that Between one fourth and three fourths of the subjects were lying. Instead, there may be more identifiable signs of lying than of ihfulness. This possibility i consistent with what we have found in -asuring the facial, bodily, and vocal differences between lying and Fruthfuless. ‘Our study has, fr the frst time, shown that accurate judgments ae ‘ot confined to selected law-enforcement groups. Psychologists with VOL. 10, NO, 3, MAY 1999 «special imerestin deception aso showed more accurate performance than other groups of psychologists. ‘We have also provided the first evidence that accuracy is possible for judging more than one kind of lie. The subjects who lied in our pre vious study were in an unusual situation—being interviewed while they watched a film designed to evoke emotions. Inthe present study. the subjects who led and told the tuth were ina situation that more closely resembled an initial interview. ‘As has been reported before, neither the age nor the sex of the ‘observer was related to accuracy. “Although there were significant group differences, even the more sccurate groups showed a range of scores, with some observers doing, ‘much better than others. In every group there was a substantial num ber who performed ator below chance. It is unlikely that judging eception from demeanor will ever be sufficiently accurate to be fadmisible in the courtroom. Nevertheless, judgments based on ‘demeanor ean be quite useful in pointing tothe need to develop more information. Our findings suggest that judgments that someone may be lying will have value only if they are made by certain profession- als, and even then no ll ofthese judgments will be aceurate, Most of lus would do wel to entertain some skepticism about our ability to detect deception from demeanor. ‘Acknowiedpments—This research was supponed in pan by a Senior Research Scientist Avard from the National Istite of Mental Health (KS-MH06092) Paul Ekman, REFERENCES DePalo, BM. (199. Spting et: Can aris lear 0 do Beer? Curent Directions DePaul, (108 May Deceiving ond dct ek: nih and oes foe ica several unded ss ase resented te aun Meg ot tte Ameen PacolnilSoc, Washington DC erwin Ken (1989) The moeaton! impale ett inte com oventon of deception Ia. Yale (Ed). Cre cases (pp. 81-70 Done The Neterans Kluwer Depule BM Sosy & Lar, D195). Oeseng an detecting dec BR ‘Schlenker (E1) Teel od sia ep 28-30), New Ys: Mec Tinded) Naw Yrk WW Noon, nan se Pesce, W (178) The Facial Acton Codie Sytem Plo Allo, CA: Cone alg ayo Pree, 265PSYCHOLOGICAL SCIENCE ee ‘A Few Can Catch a Liar "Biar,P & O'Sullhae, M (1991). Who can ach he? American Pychooge, 4, mn, O'all, M, Fin, W. Schr K (1991) Fae ie an Body “icing dct ura of Nomera ebvir 15,125-15, Frank BUG, & aan. (1997). Te sity desc generals cos ie ‘ges of ghee Nex. Journal of Prabal ed Sonal Poche 7 Ns. Metbia, A. (199), Nona bey of felng. Jounal of EsperimentalReerch it | asker, Mu Ore RE. (1988). Teli br: Veal an nomial cals of | deci Ih WA Sigman 5 Felsen (Es) Malan gration of (RscerveD 29/98; Revision accerreD 9729198) VOL. 10,NO. 3, MAY 1999