Gustafson, L. A., & Orne, M. T. The effects of perceived role and role success on the detection of deception. Journal of Applied Psychology, 1965, 49, 412-417.

EFFECTS OF PERCEIVED ROLE AND ROLE SUCCESS ON THE DETECTION OF DECEPTION 1

LAWRENCE A. GUSTAFSON 2 AND MARTIN T. ORNE

Institute of the Pennsylvania Hospital and University of Pennsylvania

75 college students participated in a detection of deception experiment designed to investigate conflicting results regarding the effect on the rate of detection of a preinterrogation demonstration of the polygraph's accuracy. It was hypothesized that the differences were due to differential demand characteristics in the 2 experiments. The information S received between Trials I and II and S's perception of his role were the major independent variables. If Ss received information which was consonant with their perceived roles, they were detected significantly less frequently than Ss who received information not consonant with their roles. The findings conform to the "consequences theory of detection" and support the hypothesized explanation of the disparate results

Most field situations involving the detection of deception employ a preinterrogation procedure in which the subject (S) is asked to draw a card at random from a pack and the interrogator then proceeds to inform the S as to the card he has drawn. The theory behind this practice is that, by showing S that he cannot deceive the interrogator, S will be easier to detect in subsequent interrogation. Serious doubt has been cast on this theory by the study reported by Ellson, Davis, Saltzman, and Burke (1952). In their comprehensive report on the detection of deception, they state that Ss who are successfully detected on one trial and are so informed are more difficult to detect on subsequent trials.

The basic problem involved here is the effect which S's knowledge of his performance in one situation will have on his subsequent performance. A S who has been informed that he successfully deceived on one trial will perceive the second trial quite differently from a S who has been told that he was unsuccessful.

A study by Gustafson and Orne (1963) has demonstrated that increasing the Ss' motivation to deceive made them more easily detected. In the same study, three groups were tested with different information of results: one group told that they had been successfully detected on the first trial; the second group told that they had not been detected; and the third group told nothing. The results of this part of the study, while not statistically significant, strongly suggested that the highly motivated Ss who are informed that they have successfully deceived the E will be more difficult to detect on subsequent trials, while those Ss who are told nothing or who are told that they had been detected show no change in detectability.

While the differences between the results of this study and those obtained by Ellson et al. (1952) might be attributed to the high motivation of the Ss in the former (the motivation of the Ss in the Ellson et al. [1952] report is not known), there are other variables to be considered. Specifically, it is necessary to consider not only the degree of motivation displayed by the Ss, but also the direction of motivation -- the goals of the behavior as perceived by the Ss. For example, in the study by the authors which was described above, the Ss were told that individuals with great emotional control and superior intelligence were able to deceive successfully. These instructions proved highly motivating to the college population from which the Ss were drawn. If the S is told


1 This work was supported in part by the United States Army Research and Development Command Contracts DA-49-193-MD-2480 and DA-49-193-MD-2647.

2 The authors wish to express their appreciation to M. J. Moskowitz and J. A. Barmack for their critical comments in the preparation of this manuscript.


412

 

413 LAWRENCE A. GUSTAFSON AND MARTIN T. ORNE

that he successfully deceived E on the first trial, he can assume that he possesses the desired traits. If he is told that he was detected on the first trial, he can interpret this to mean that he is not outstanding on these traits. According to the consequences theory of the detection of deception (Gustafson & Orne, 1963), it would be predicted that the S who has successfully deceived on the first trial would be less concerned about the second trial -- the consequences of deceiving would be less important to him, he would be less motivated and, therefore, more difficult to detect. In the other cases, the already high motivation-to-deceive would not likely change on the second trial.

An alternative situation can easily be imagined. Without specific instructions, the S might well perceive the purpose of the experiment to establish the effectiveness of lie detection. He might further assume that "a normal S" can readily be detected by the equipment used. If the S makes these assumptions and he is told that he was not detected on the first trial, that he successfully deceived the E, then he would of necessity conclude that his responses had been different from that of the "normal S." As a result, he would be more concerned on the second trial. The second trial would be more important to him, and the consequences theory of detection would predict that he would, therefore, be more easily detected on the second trial. Conversely, if the S perceives the experiment in the manner outlined and if the S is told that he has been detected on the first trial, that is, responded in the manner of "normal Ss," the second trial would be of less importance to him and therefore he would be more difficult to detect.

Thus, the manner in which the S perceives the experimental lie detection situation would lead to opposite results. The S's perception of the experimental situation is a variable which is necessarily present in all experimental situations. Orne (1962) elsewhere has termed the sum total of cues which determine the S's perception of the purpose of the experiment as the "demand characteristics of an experimental situation." The type of cues which may be involved include explicit and implicit instructions, the experimental setting, the experimental procedure itself, subtle behaviors of the E (as well as the ancillary personnel), etc., as they are interpreted in the light of S's past knowledge and experience. Any experimental situation will have demand characteristics (i.e., necessarily most Ss form some hypothesis about what they are doing and why) which may be significant factors in the interpretation of the experimental findings.

Subtle differences in demand characteristics could have existed between the experiment of Ellson et al. (1952) and our original pilot studies. In these pilot studies, Ss were motivated to deceive, whereas in the Ellson et al. studies no specific preexperimental instructions were given. However, many Ss come to the experiment with the conviction that normal individuals cannot fool the lie detector and that only individuals who lack strong conscience and who have some kind of criminal tendencies are able to lie successfully. It seems plausible that these preconceptions were present in the population from which the Ellson et al. sample was drawn. Unless specific instructions are given to the contrary, this type of preconception might well cause the S to perceive the experiment in the alternative way discussed above. It should be clear that problems of experimenter bias are not being considered here, rather the possibility that Ss in the two experiments perceived the purposes in two alternative fashions. In the light of the consequences theory of deception, the interaction between these different sets of Ss' perceptions (demand characteristics) with the experimental variable would lead to opposite results.

The present experiment was designed to test whether differences in the S's perceptions explicitly created by instructions similar to those which could have been introduced by implicit differences in the experimental situation would yield opposite findings even though otherwise the identical set of experimental variables were employed. The major independent variables were: (a) the S's perception of his role in the experiment, and (b) the information he received as to his success on the first trial. The primary dependent variable was his detectability in the second trial of the experiment. In keeping

 

414 DETECTION OF DECEPTION

with a widely used method for categorizing the direction of human motivation, one group of Ss was designated as "need deceive" (n Deceive) and the second group as "need detected" (n Detected)."

It was predicted that:

1. Ss who perceive their task to be the successful deception of E (n Deceive) and are told that they have indeed been successful in deception will be detected less frequently on the next trial than Ss with the same perception who are told that they have been detected.

2. Ss who perceive their task to be that of being detected by E (n Detected) and are told that they were indeed detected on the first trial will be detected less frequently on the next trial than Ss with the same perception who are told that they had not been detected.

3. Ss in the n Deceive and the n Detected groups who are given the same information as to whether they were detected or not will be detected on the next trial at significantly different rates.

METHOD

Subjects. Seventy-five male undergraduates were recruited from local school employment offices and paid for their participation. Many Ss had participated in psychological experiments, but none had previously taken part in detection of deception experiments.

Procedures. The Ss were randomly divided into two groups: n Deceive and n Detected. Each major group was divided further into two subgroups according to the information given the S between Trial I and Trial II. In subgroup one, the S was informed that he had successfully deceived E, while in the second subgroup the S was told that he had been detected.

A tape recording was played to each S at the beginning of the experiment. The tape included information designed either to motivate S to deceive or to motivate him to be detected, according to which of the two major groups S belonged, and procedural instructions which were to be followed "for the duration of the experiment." These instructions were identical for both the n Deceive and n Detected groups and were the same as those used in an earlier study (Gustafson & Orne, in press), Those Ss in the two n Detected groups heard the following:

Thank you for participating in our research. Just a few words about the experiment in which you are about to take part. This study is designed to see whether or not you can withhold information from equipment which is based on a principle similar to that of the lie detector. Of course, the equipment which we are using is a good deal more sophisticated than that which is usually used as a lie detector device. I am sure that you know that these devices have not yet been fully recognized by the courts, primarily because the circumstances under which they work and the mechanisms which make them work are not yet fully understood. However, certain facts are known based in good part on our own research as well as that of others. We know that individuals who are normal and well-adjusted find it extremely difficult and even impossible to prevent themselves from giving certain physiological reactions when they lie. This is largely due, or so it seems, to certain childhood experiences which seem to have caused a kind of conditioned involuntary autonomic response associated with lying. The machine does nothing more than record this response making it possible to recognize a lie. Of course this only works with normal individuals; individuals who have so-called psychopathic tendencies, who are able to lie without any feeling of guilt, or who are mentally disturbed, do not appear to show these kinds of changes associated with lying.

In this study we are interested in your attempting to prevent yourself from showing the response to lying. We realize that this is extremely difficult for normal individuals, however, we want you to try. Presently the rules of the experiment will be explained to you. You are to follow them as carefully as you can. Your job is to try to show no response whenever you can.

Good luck!

The two n Deceive groups listened to a tape which was similar to the one used in our preliminary study cited in the introduction (Gustafson & Orne, 1963):

We would like to thank you for participating in our research. Just a few words about the experiment in which you are about to take part. This study is designed to see whether or not you can withhold information from equipment which is based on a principle similar to the lie detector. Of course the equipment we are using is a good deal more sophisticated than the usual kind of lie detection device. I am sure that you know that these devices have not yet been fully recognized by the courts largely because there is some question about their validity. These tests are designed to see whether or not an individual who really wants to is able to withhold information from the machine; in other words, when certain significant things are said whether you can suppress your autonomic, your involuntary bodily reaction to them. Some few people are able to do this. However, I should tell you that it is rather difficult. We have found that these are individuals


3 These do not refer to components of Murray's theoretical system.

 

415 LAWRENCE A. GUSTAFSON AND MARTIN T. ORNE

who have more than the usual amount of control, who are quite superior in intelligence.

I should tell you that in this experiment you should try your level best to withhold information from the machine. As I say, while it is very difficult to do this, it is possible. Now the rules of the experiment will be explained to you.

Good luck!

Transducers for recording skin resistance, heart rate, and respiration were then attached. The method for recording skin resistance was that used in previously reported research (Gustafson & Orne, 1963). (The other measures were included only for exploratory purposes and will not be discussed further.)

Each S drew a card from a seven-card deck which consisted of two blank cards and five cards, each of which had a different two-digit number printed on it. The decks had been arranged in such a manner that E could identify the card which S selected by its position in the deck. Although this procedure is not usual in detection of deception studies, it was followed because it was necessary for E to be able to present S with the information appropriate to his subgroup without regard to his performance on Trial I.

Because only Ss who drew number cards on both trials could be used in testing the hypotheses, the blank cards were arranged so that they were always in either the first or second position at the front and the back of the deck. Observations from earlier studies indicate that Ss draw cards from near the middle of the deck much more frequently than they do from the extremes. Thus, only a few Ss were lost because they drew blank cards.

Each S was instructed to write the number on another card 10 times while E was out of the room and to write zeros if he drew a blank card. This was done to insure that S had actually looked at the card. Both cards were turned over by the S before the E returned. Prior to each trial, the E reminded S to respond verbally with "no" to each number presented to him during the interrogation.

The deception task was based upon what the present authors have designated as the "guilty-person paradigm" (Gustafson & Orne, in press). In this design, S's task is to appear as though he has drawn a blank card and, therefore, has no "special" information concerning any of the numbers. This model is contrasted to the "guilty-information paradigm" in which S may deceive E by forcing a response to a noncritical item (Gustafson & Orne, in press). The task of all Ss in this experiment, therefore, was to suppress responses to all numbers and thus try to appear innocent. Of course, in the case where S had drawn a blank card, there was no deception involved.

The numbers were presented in a manner similar to the relevant-irrelevant method used in commercial lie detection (Inbau & Reid, 1953; Lee, 1953). The five numbers were presented in random order to S by a tape-recorded voice, one number every 15 seconds. The first number on the tape was a dummy so that the inordinately large GSR response which usually appears on the first stimulus presentation could be discarded. Recordings were made on an Offner Type R Dynograph at a paper speed of 2.5 millimeters per second. After a stable resting GSR had been obtained, the tape-recorded interrogation was played to S.

At the conclusion of Trial I, E re-entered the subject room and presented the S with the information appropriate to his subgroup. One-half of the Ss were told that they must have drawn blank cards (though they had not) and thus were made to feel that they had successfully deceived E, while the other half were informed of the number and were asked to verify that E had, in fact, correctly detected them.

The S was then asked to draw a card from a second deck. The procedure followed during Trial II was identical to Trial I except the interrogation tape was made to correspond to the numbers in the second deck. These numbers did not duplicate any of the numbers in the first deck.

Analysis of the data. The analysis was performed by individuals who did not know what the critical numbers were and who were unaware of the authors' hypotheses concerning the experiment. Of the 75 Ss who took part in the experiment, 11 were discarded because, by chance, they happened to have chosen a blank card on either Trial I or Trial II, or because they did not follow instructions.

The average GSR response to each of the five numbers was determined. The largest mean response was ranked as one, the second largest as two, etc. If the number which S drew was given a rank of one, S was considered to have been detected. If the rank assigned was not one, S was considered to have deceived E. Chi-square tests were used to compare the number of successful and unsuccessful detections between different conditions. Trial I and Trial II were treated individually in the analysis.

RESULTS

While the outcome of Trial I was not a principal concern of this study, it was important to determine whether there were any systematic differences between the subgroups prior to the introduction of the treatments presented at the conclusion of Trial I. The number of successful and unsuccessful detections on Trial I for the n Deceive and n Detected groups are presented in Table 1 according to the kind of information S received between Trials I and II. There were no significant differences.

Prediction 1. The Ss of the n Deceive group who were told that they had successfully deceived E (on Trial I) were detected significantly less often on Trial II than n Deceive Ss who were told they had been detected.

 

416 DETECTION OF DECEPTION

Prediction 2. The Ss of the n Detected group who were told that they were unsuccessful in deceiving E were detected significantly less often on Trial II than n Detected Ss who were told they had not been detected (see Table 2).

Prediction 3. The Ss in the n Deceive and n Detected groups who were given the same information with regard to their success or failure at deceiving on Trial I were detected at significantly different rates on Trial II (see Table 2).

On Trial II, it is interesting to note that the Ss in the two groups with low-detection rates showed markedly different overall response patterns. Often a very flat GSR record was obtained. Frequently Ss responded only to the first number (which was a dummy and not included in the analysis). This led to a number of ties when the average responses were ranked. The GSR records were similar to those of the unmotivated group in a study reported earlier (Gustafson & Orne, 1963).

DISCUSSION

The results support the hypothesis that the demand characteristics (S's perception of the purpose of the experiment) significantly affect the rate of detection. The principal

 

417 LAWRENCE A. GUSTAFSON AND MARTIN T. ORNE

finding is that subtle motivational variables affect what has often been assumed to be a relatively mechanical procedure. When S perceives his role to be one in which he will be detected and yet is told that he has successfully deceived E, he becomes relatively easy to detect. Similarly, if S is motivated to deceive and yet is told that he has been detected, he also becomes relatively easy to detect. More generally, if S is given information indicating to him that he is not performing in a manner consonant with his role, he is more easily detected than if he is given information indicating that he is behaving in accordance with his role.

The results indicate that the differences between detection rates on Trial II are not due to differences in Ss prior to the experimental manipulation which occurred between Trials I and II. It should be remembered, however, that E knew the composition of the groups beforehand and could conceivably have communicated cues to S during or after the experimental treatment. Indeed, one may be legitimately concerned with the problems of bias in these results if, as has been demonstrated, subtle factors affect the rate of detection.

Previous research (Rosenthal, 1963) has indicated that E is by no means neutral to the outcome of his study. Only one E (LAG) took part in the running of Ss. Because of his theoretical frame of reference, he had a considerable investment in the attainment of the expected outcome of the study. It is possible, therefore, that in his interaction with the S, E subtly communicated his expectations. In the light of the consequences theory of deception, however, any S who became aware of E's wish that he not be detected should, in effect, be detected more easily. In other words, any subtle communication of E's expectations to S might well have militated against the hypothesis. Furthermore, in a previous experiment the same E had confidently expected to replicate the findings of Ellson et al. (1952) and was surprised when contrary results were obtained. The findings may thus be ascribed to the differential affect in motivation introduced by the tape recordings rather than to artifact or bias introduced by E. Nonetheless it would seem desirable to replicate these findings with E unaware as to which group S belongs.

These results support the view that the detection of deception is a subtle phenomenon in which psychological variables play a crucial role. The study appears particularly relevant since the conclusion which might have been drawn on the basis of the work of Ellson et al. (1952) would be that the standard technique as used in the field should be revised, that is, that the proof given to S of the polygraph's effectiveness would decrease rather than increase its subsequent utility. This conclusion is not justified, however, since the psychological parameters of the field situation correspond to the n Deceive group. In this situation, proof of successful detection maximizes subsequent detection.

It is assumed that the study conducted by Ellson et al., where motivation is not specified, corresponds to the n Detected group, suggesting that the psychological setting in which an experiment is performed may interact in a crucial fashion with experimental variables.

REFERENCES

ELLSON, D. G., DAMS, R. C., SALTZMAN, I. J., & BURKE, C. J. A report on research on detection of deception. DDC Technical Report ATI-168902, 1952, Indiana University, Contract No. Nonr-60nr-18011, Office of Naval Research.

GUSTAFSON, L. A., & ORNE, M. T. The effects of heightened motivation on the detection of deception. Journal of Applied Psychology, 1963, 47, 408-411.

GUSTAFSON, L. A., & ORNE, M. T. The effects of task and method of stimulus presentation on the detection of deception. Journal of Applied Psychology, in press.

INBAU, F. E., & REID, J. E. Lie detection and criminal interrogation. (3rd ed.) Baltimore: Williams & Wilkins, 1953.

LEE, C. D. The instrumental detection of deception. Springfield, Ill.: Charles C Thomas, 1953.

ORNE, M. T. On the social psychology of the psychological experiment: With particular reference to demand characteristics and their implications.American Psychologist, 1962, 17, 776-783.

ROSENTHAL, R. On the social psychology of the psychological experiment. American Scientist, 1963, 51, 268-283.

SUTCLIFFE, J. P. A general method of analysis of frequency data for multiple classification designs. Psychological Bulletin, 1957, 54, 134-137.

(Received August 7, 1964)


The preceding paper is a reproduction of the following article (Gustafson, L. A., & Orne, M. T. The effects of perceived role and role success on the detection of deception. Journal of Applied Psychology, 1965, 49, 412-417.). It is reproduced here with the kind permission of the American Psychological Association ©1965. No further reproduction or distribution of this article is permitted without written permission of the publisher.