Deception is defined, for the purposes of this article, as a successful or unsuccessful deliberate attempt to create in another a belief that the sender of the message considers to be untrue. Although it is hard to think of a context in which no deception transpires, the study of deception and how to detect it is especially crucial in the forensic setting. Most law enforcement professionals, who must assess veracity on a daily basis, know that deception is quite frequent in forensic contexts and that making mistakes when assessing veracity can have severe consequences—the innocent may be sentenced to punishment, the guilty may be freed to commit more crimes. To be able to correctly detect deception is therefore of utmost importance. Yet comprehensive study over the past 40 years has shown that the human ability to detect deception is just above the level of chance. The consistency of this finding is striking, although there are factors moderating the rate of correct judgments. For example, accuracy is somewhat higher when listening to rather than watching the liar, when one has access to baseline information about the liar’s behavior, and when detecting unprepared rather than prepared messages.
How to Study Deception Detection
To gain insight into deception, psychologists and other researchers conduct experiments. They instruct some people either to lie or to tell the truth and instruct others to judge the veracity of the resulting statements. Those who lie or tell the truth in these experiments are referred to as senders, the truthful and deceptive statements as messages, and those who judge the messages as receivers. In this research paper, the accuracy of these receivers is at focus, more specifically the accuracy of human judgments made without any specialized tools or aids in detecting deception on the basis of verbal content and the liar’s behavior. The receivers are typically given videotaped or audiotaped statements, and ordinarily, half the messages a receiver encounters are truths and half are lies; hence, the chance level of correct judgments a receiver could expect is 50%. Lie detection ability is most often expressed as percent correct, but other indices of deception detection accuracy, such as standardized differences between truth and lie detection accuracy, are also calculated.
The standard lie detection experiment contains several factors that have been examined through experimental manipulation. For example, the senders of the message can be adults, adolescents, or children, or they can be persons with or without special skills at lying, such as experienced criminals. Furthermore, the content of the lies (and truths) have been varied: People have lied about their personal feelings, about their committing of transgressions such as adultery or sanctioned crimes, or in placing the blame on someone other than the culprit. Lie detection through different media has also been tested: Are people better lie detectors when having access to video or audio or written transcripts? In addition, characteristics of the receivers have been varied: Are certain groups of people, such as police officers, better lie detectors? These are only some of the factors that have been scientifically examined.
Overall accuracy of lie detection has been analyzed in several meta-analyses and reviews. The results are unanimous in terms of the mean percentage of accuracy: In the typical research setting, lies are discriminated from truths at levels that are only slightly better than would be attained by flipping a coin. The mean percentage of accuracy is just under 54. This effect is small, but since it is based on thousands of veracity judgments, it is significantly better than the level of chance. Typically, studies report an accuracy rate between 50% and 60%.
In calculating the just presented overall percentage of accuracy, some exclusion criteria have been applied. Studies in which training to detect deception is provided are not included, nor are studies on adults’ ability to detect children’s deception. Also excluded are studies on implicit lie detection and studies not in the English language.
Because deception judgments can have severe consequences whether or not they are correct, it is important to understand the factors that may bias the judgments in one direction or another. The research literature has evidenced a truth bias—receivers’ tendency to make systematic mistakes in the direction of judging messages as truthful, with a mean percentage of around 56% (which is significantly greater than 50%). One consequence of this truth bias is that people on average correctly identify truthful messages (mean percent correct just above 61%) more often than they correctly identify deceptive messages (mean percent correct just below 48%).
Using percent correct as a measure of accuracy has been criticized, and other measures have been suggested. However, analyses of log-odds ratios or signal detection measures, among others, also indicate an overall accuracy rate of about 54%. The different deception detection measures are highly inter-correlated.
A deception detection accuracy (sometimes referred to as lie/truth discrimination) of 54% is the typical result over a variety of receiver samples, sender samples, deception media, types of lies, and contexts. Conceivably, there might be certain conditions under which judges will show different accuracy rates. To evaluate these possibilities, an inspection of various subsets of the research literature on deception judgments is needed. In the following section, a number of factors that may moderate deception detection accuracy are discussed.
Lies and truths can be evaluated over different media. It is of interest to compare detection rates for lies that can be seen, heard, or read. For example, the video medium might encourage the use of a liar stereotype. Having access to verbal content only may give the receiver the chance to analyze the messages more thoroughly.
Results have shown that lie/truth discrimination accuracy is lower if judgments are made in the video-only medium (rather than audiovisual and audio-only, as well as written transcripts). Further results show that messages are perceived as most truthful if judged from audiovisual or audio presentations, followed by written transcripts and video presentations.
The medium in which deception is attempted thus affects its likelihood of detection—lies being more evident when they can be heard. Given that the stereotype of a liar is largely visual (eye contact, fidgeting, gestures), this stereotype is most strongly brought to mind by the video medium. Those senders who appear nervous, tormented, or distressed are then judged to be lying; but these expressions may be the result of factors other than deceit.
Sometimes people have anticipated that they have to lie and are therefore prepared in their attempted deceit. On other occasions, the lie is told in response to an unanticipated need, and people are then unprepared for the task of lying. Being prepared or not should, in principle, affect the sender’s believability. The available research suggests that receivers achieve higher deception detection accuracy when judging unprepared than prepared messages. It has been found that it is easier to discriminate between unprepared lies and truths than between prepared lies and truths. Furthermore, prepared messages appear more truthful than messages that were unprepared.
However, differences in experimental design have been shown to lead to differences in accuracy rates. Studies in which the senders produced both prepared and unprepared messages yielded the result just described. Studies in which the preparation factor was examined by having messages from unprepared participants compared with those from prepared participants did not show any reliable difference in receivers’ ability to detect deception and truth. Here, the unprepared messages were more often judged as truthful. Further research on this issue is certainly needed.
Common sense would predict that a receiver should more correctly pinpoint the lies of a sender he or she has some familiarity with (“baseline exposure”). If one has more knowledge of someone’s behavior than one gets from just watching a few minutes on a videotape, one should be able to detect deviations from that behavior if telling a lie causes deviations in behavior.
Results indicate that baseline exposure does indeed improve lie/truth discrimination: Receivers achieve a higher accuracy when given a baseline exposure. However, one should be aware that senders who are familiar to the receiver are more likely to be judged as truthful. People seem unwilling to infer that someone familiar to them is lying.
Sometimes deception studies are criticized because the research participants do not have any incentive to be believed, and this lack of motivation in the task could influence participants’ believability. Deception research has, however, addressed this issue and investigated the effects of different levels of sender motivation. Furthermore, the influential deception scholar Bella DePaulo has hypothesized that senders are undermined by their efforts to get away with lying. According to her motivational impairment hypothesis, the truths and lies of highly motivated senders will be more easily discriminated than those of unmotivated senders. Experimental studies show that lies are easier to discriminate from truths if they are told by motivated rather than unmotivated senders, in accordance with the hypothesis.
However, this result has been found for within-study comparisons and has generally not been found when comparing between studies. Here, the reliable difference found is that motivated participants appear less truthful than those with no motivation to succeed. It matters little if a highly motivated speaker is lying or not; what matters is the fear of not being believed. Research further indicates that motivation in itself affects how the sender is perceived differently for different media: Motivation reduces senders’ video and audiovisual appearance of truthfulness but has no effect on how truthful a sender sounds. Is seems as if motivation makes people resemble a visible stereotype of a liar. If so, motivational effects on credibility might be most apparent in video-based judgments.
In conclusion, the accumulated evidence suggests that people who are motivated to be believed appear deceptive, whether or not they are lying.
In some studies, the deceptive and truthful senders are alone, talking to a camera. In other studies, an experimenter, blind to the veracity of the person in front, asks a standardized set of questions. Sometimes, the interaction partner is attempting to judge the veracity (such as in a mock police interview or interrogation); on other occasions, an observer may be making this judgment. The latter occurs, for example, when the interaction partner is the experimenter and the observer is the receiver assessing veracity on the basis of the video-taped interaction. In principle, social interaction might influence the receiver’s veracity judgments and/or the receiver’s success at detecting deception.
The literature produces clear evidence that receivers are inclined to judge their interaction partners as truthful much more often than observers do. The overall pattern in the literature further suggests that observers are better than interaction partners at discriminating lies from truths. It seems as if people do not want to believe that someone has just lied to them without their spotting it. Alternatively, the reluctance to attribute deception to interaction partners could be the result of not wanting to insinuate that the partner is a liar.
In summary, research suggests that lies told in social interactions are better detected by observers than interaction partners.
Usually, those making veracity judgments in deception research are college students. They have no special training and may have no interest in or reason for succeeding in the task. Reasonably, people with more experience would be better at judging deceit, and to assess this possibility, researchers have also tested presumed deception detection experts. These are individuals whose occupations expose them to lies, and they include law enforcement personnel, judges, psychiatrists, job interviewers, and customs officials.
The results are clear-cut. The “experts” are not experts at lie detection—there is no reliable difference in deception detection accuracy compared with novices. The accumulated research further suggests that experts are more skeptical than nonexperts, meaning that they are less inclined to believe that people are truthful. Having been targets of deceit in their professional roles, these experts may have overcome the usual unwillingness to infer that certain people are liars. However, it should be noted that the experimental setting that the experts have been tested in may not make possible a fair representation of their expertise. For example, police officers very rarely assess veracity on the basis of one, short videotaped interview and without having access to evidence. Therefore, the conclusion that experts are not better than laypeople at detecting deception may be premature. Future research is needed to shed light on experts using their expertise in a more ecologically valid setting.
Beliefs about Deception
The most often stated reason for the low accuracy rates found in deception research is that there is a disparity between what actually is indicative of deception and what people believe to be indicative of deception. As hinted at earlier, there is a stereotypical belief concerning a liar’s behavior. A belief is a feeling that something is true or real; it can be strong or weak, correct or incorrect. The beliefs that a person holds are often reflected in his or her behavioral dispositions; that is, beliefs guide action. Hence, if one wants to learn about deception detection, it is important to study people’s beliefs about deception.
Two different methods have been used to investigate people’s beliefs about cues to deception: surveys and laboratory-based studies. In the surveys, participants have typically been asked to work through a list of prespecified verbal and nonverbal behaviors and for each particular behavior (e.g., gaze aversion and head movements), rate the extent to which they believe that this behavior is indicative of deception. The second method is used in studies where participants first watch videotapes of deceptive and truth-telling senders and then judge these in terms of veracity. Most studies on beliefs about deception have employed college students as participants, but there is also research on experts’ beliefs about deception (e.g., police officers, customs officers, prison guards, prosecutors, and judges).
The available research shows that the beliefs held by experts and lay people are very similar. In terms of nonverbal cues, the evidence suggests that both experts and lay people consider nervous behaviors to indicate deception. For example, both experts and lay people believe that eye contact decreases during lying, but research on objective cues to deception has shown that this particular cue is not a reliable predictor of deception. Furthermore, both experts and lay people have indicated a strong relationship between deceptive behavior and an increase in bodily movements, which is also incorrect. In terms of verbal indicators of deception, both experts and lay people believe that truthful messages are more detailed than deceptive ones, and to some extent, research on objective cues to deception supports this belief. Researchers have in addition studied cross-cultural beliefs about deception and found that people around the world believe that deception can be spotted in the eye behaviors of the sender, such as gaze aversion. As regards accuracy in cross-cultural deception judgments, the available research shows that, as expected, deception is even harder to detect when the sender and receiver are not from the same cultural or ethnic group.
In sum, research on beliefs about deception has shown that the beliefs are similar for experts and lay people and that these beliefs to a rather large extent are misconceptions about how liars actually behave.
Training to Detect Deception
In a number of published studies, researchers and scholars have tried to train people in detecting deceit. The training programs have differed markedly in content and duration, but information about the mismatch between beliefs about deception and actual indicators of deception seems commonplace. Often, feedback on the veracity judgments made has been provided as well. In general, training has been shown to significantly increase the accuracy of lie detection—a small but detectable increase is most often found. However, even if an increase is found, it usually is from, say, 55% to 60%, which is still of limited practical value. Furthermore, the long-term effect(s) of training is not as yet known. Unfortunately, the one group of participants that has been the hardest to train to become better in the deception detection task is police officers.
Limitations and Future Challenges
When deception detection research has been criticized, it is often the type of lies studied that has come under attack. For example, most of the lies studied have not been about transgressions, so some critics have argued that the lies told are not high-stake lies; others argue that the social aspects of lying and lie detecting are too constrained in experimental settings; and legal scholars point out aspects of the forensic world that have not been examined in research contexts. Deception researchers have tried to answer the critics by, for example, studying murderers’ and other criminals’ lies in police interviews, lies that could harm children, and lies to lovers. Researchers have also begun to study naturalistic deceptive interactions, jurors’ credibility judgments, and police officers’ assessments of veracity after conducting the interviews themselves.
In experiments, the receivers come across one brief message and must judge the veracity of that message on the spot, with no time or opportunity to collect additional information. Outside the laboratory, however, additional information is important. When asked to describe their discovery of a lie, people rarely state that the discovery was prompted by behaviors dis-played at the time of the attempted deception. Rather, they say that lie detection took days, weeks, or even months and involved physical evidence or third parties. In police interviews, for example, the evidence in the case may be used as a tool to detect deception. Future studies will be needed to examine the impact on lie detection of these and other forms of extra-behavioral information. At present, across hundreds of experiments, the typical rate of deception detection in adults remains just above the level of chance.
- Bond, C. F., Jr., & DePaulo, B. M. (2006). Accuracy of deception judgments. Personality andSocial Psychology Review, 10, 214-234.
- DePaulo, B. M., Zuckerman, M., & Rosenthal, R. (1980). Humans as lie detectors. Journal of Communication, 30, 129-139.
- Granhag, P. A., & Stromwall, L. A. (Eds.). (2004). The detection of deception in forensic contexts. Cambridge, UK: Cambridge University Press.
- Granhag, P. A., & Vrij, A. (2005). Deception detection. In N. Brewer & K. D. Williams (Eds.),Psychology and law. An empirical perspective (pp. 43-92). New York: Guilford Press.
- Kraut, R. (1980). Humans as lie detectors: Some second thoughts. Journal of Communication, 30, 209-216.
- Vrij, A. (2000). Detecting lies and deceit: The psychology of lying and the implications for professional practice. Chichester, UK: Wiley.
Return to Police Psychology overview.