In this chapter, we focus on several issues associated with deception, including the use of the polygraph and alternatives to the polygraph, the relationship between verbal and nonverbal cues to deception, and methods for detecting the malingering of mental disorders.
THE POLYGRAPH TECHNIQUE
Polygraph: A device for recording individual's autonomic nervous system responses
Physiological measures have long been used in an attempt to detect deception. Polygraphy relies on the same underlying principle: Deception is associated with physiological change. The polygraph (the word is a combination of two Greek words, poly = “many” and grapho = “write”) is a device for recording an individual's autonomic nervous system responses.
- Measurement devices are attached to the upper chest and abdomen to measure breathing.
- The amount of sweat on the skin is measured by attaching electrodes to the fingertips. Sweat changes the conductance of the skin, which is known as the skin conductance response.
- Heart rate is measured by a partially inflated blood pressure cuff attached to an arm.
Each of these measures is amplified and can be printed out on paper or stored in a computer to be analyzed. In a forensic context, a polygraph is used to measure a person's physiological responses to questions asked by an examiner.
Applications of the Polygraph Test
Polygraph tests are used for a range of purposes:
- In Canada, the police often use polygraph tests to help in their criminal investigations. The police may ask a suspect take a polygraph test as a means to resolve the case. If the suspect fails the polygraph test, that person may be pressured to confess, thereby giving the police incriminating evidence.
- Although not common, police may ask alleged victims of crimes to take a polygraph test to help verify whether a crime has occurred.
- Insurance companies may request a polygraph test to verify the claims of the insured.
- More recently, the polygraph has been used in the United States to assess and monitor sexual offenders on probation.
- Used to determine whether the offender is violating the conditions of probation or to test for evidence of risky behaviour, such as sexual fantasies about children.
- For the periodic testing of employees to identify those engaged in theft or using drugs at work and for the screening of prospective employees, to weed out those with criminal tendencies or substance-abuse problems.
Polygraph Disclosure Tests: polygraph tests that are used to uncover information about an offender's past behaviour
Types of Polygraph Tests
The polygraph does not detect lies per se, since the physiological states associated with lying share much in common with many other states, including anxiety, anger, embarrassment, and fear. Instead, polygraph tests rely on measuring physiological responses to different types of questions. Some questions are designed to elicit a larger physiological response in guilty individuals than in those who are innocent. There are two main types polygraph tests.
Comparison Question Test: A type of polygraph test that includes irrelevant questions that are unrelated to the crime, relevant questions concerning the crime being investigated, and comparison questions concerning the person's honesty and past history prior to the event being investigated
The typical CQT includes a pretest interview, followed by a series of questions administered while the suspect's physiological responses are measured (usually three separate question sequences are asked). The polygraph examiner then scores the charts and ends the CQT with a post-test interview in which the test results are discussed.
A critical component of this technique is the pretest interview. During the pretest interview, the polygraph examiner develops the comparison questions, learns about background of the suspect, and attempts to convince the suspect of the accuracy of the polygraph test.
The examiner then asks ten questions to be answered with either “yes” or “no.” Three types of questions are asked:
- Irrelevant questions, referring to the respondent's identity or personal background (e.g., “Is your first name Beatrice?”) are included as a baseline but are not scored.
- Relevant and comparison questions establish guilt or innocence
Relevant questions deal with the crime being investigated (e.g., “On June 12, did you stab your exwife?”). Probable-lie comparison questions (aka. control questions) are designed to be emotionally arousing for all respondents and typically focus on the person's honesty and past history prior to the event being investigated (e.g., “Before the age of 45, did ever try to seriously hurt someone ?”).
Polygraph examiners assume they can detect deception by comparing reactions to the relevant and comparison questions. Guilty suspects are assumed to react more to relevant questions than comparison questions. In contrast, innocent suspects are assumed to react more to comparison questions than relevant questions. The reasoning behind these assumptions is that innocent people know they are telling the truth about the relevant question so they will react more strongly to general questions about their honesty and past history.
Most examiners now numerically score the charts to ensure that decisions are based solely on the physiological responses. A polygraph test has three possible outcomes:
During the post-test interview, the examiner tells the suspect the outcome, and if the outcome is deceptive the examiner attempts to obtain a confession.
Concealed Information Test: A type of polygraph test designed to determine if the person knows details about a crime
The CIT does not assess deception but instead seeks to determine whether the suspect knows details about a crime that only the person who committed the crime would know. The general form of the CIT is a series of questions in multiple-choice format. Each question has one correct option (often called the critical option) and four options that are foils-alternatives that could fit the crime but that are incorrect.
A CIT question in the context of a homicide might take the following form: “Did you kill the person with (a) a knife, (b) an axe, (c) a handgun, (d) a crowbar, (e) a rifle?”
The guilty suspect is assumed to display a larger physiological response to the correct option than to the incorrect options. An innocent person, conversely who does not know the details of the crime, will show the same physiological response to all options.
Underlying the CIT is the principle that people will react more strongly to information they recognize as distinctive or important than to unimportant information. Suspects who consistently respond to critical items are assumed to have knowledge of the crime. The likelihood that an innocent person with no knowledge of the crime would react most strongly to the critical alternative is one in five for each question. If ten questions an asked, the odds that an innocent person will consistently react to the critical alternative are exceedingly small (less than 1 in 10 000 000).
Iacono and Patrick (1999) suggested two reasons for the lack of widespread acceptance of the CIT:
- First, since polygraph examiners believe in the accuracy of the CQT, they are not motivated to use the more difficult-to-construct CIT.
- Second, for law enforcement to use the CIT, salient features of the crime must be known only to the perpetrator. If details of a crime appear in the media, the crime-related details given cannot be used to construct a CIT.
Validity of Polygraph Techniques
How is the accuracy of polygraph tests assessed? Accuracy is determined under ideal circumstances by presenting information known to be true and false to individuals and measuring their corresponding physiological responses. Studies of the validity of polygraph techniques can be classified into two types: laboratory and field studies.
In laboratory studies, volunteers (often university students) simulate criminal behaviour by committing a mock crime. Volunteers come to a laboratory and are randomly assigned to one of two conditions: committing a mock crime or not committing a mock crime. The main advantage of these studies is that the experimenter knows ground truth.
Ground Truth: As applied to polygraph research, the knowledge of whether the person is actually guilty or innocent
However, because of the large motivational and emotional differences between volunteers in laboratory studies and actual suspects in real-life situations, the results of laboratory studies may have limited application to real life. In laboratory studies, guilty participants cannot ethically be given strong incentives to “beat” the polygraph, and both guilty and innocent participants have little to fear if they “fail” the polygraph exam.
Field studies involve real-life situations and actual criminal suspects, together with actual polygraph examinations. Field studies often compare the accuracy of “original” examiners to “blind” evaluators. Original examiners conduct the actual evaluation of the suspect. Blind evaluators are provided with only the original examiner's charts and are given no information about the suspect or the case. Original examiners are exposed to extra polygraph cues—information about the case in addition to that obtained via the polygraph, such as the case facts and the behaviour of the suspect during the examination.
The largest problem with field studies is establishing ground truth. Indicators of guilt, such as physical evidence, eyewitness testimony, or DNA evidence, are often not available. In such situations, truth is more difficult to establish. To deal with this problem, two additional ways of establishing ground truth have been developed: judicial outcomes and confessions.
- Judicial outcomes are problematic because some people are falsely convicted and some guilty people are not convicted.
- Confessions are also problematic. Although rare, some people may falsely confess. More significant, however, is the problem that confessions are often not independent from the polygraph examiner's decisions. Concessions are often elicited because a person fails a polygraph exam.
Polygraph Tests: Accurate or Not?
The accuracy of the polygraph for detecting lies is debatable. There are problems when relying on typical mock-crime scenarios to estimate real-life accuracy. As a consequence, only field studies of the CQT will be described. Although the CQT has been investigated for more than 30 years, its ability to accurately measure deception remains controversial. Most of the studies have used confessions to classify suspects as guilty or innocent, and as noted above, there are problems with using this as the criterion.
Most guilty suspects (84% to 92%) are correctly classified as guilty however, the picture for innocent suspects is less optimistic, with accuracy rates ranging from 55% to 78%. Many of the innocent suspects were classified as inconclusive. Between 9% and 24% of innocent suspects were falsely identified as guilty. Such a high false positive rate indicates that innocent people respond more to relevant than comparison questions, suggesting that the premise underlying the CQT does not apply to all suspects.
The CIT appears to be vulnerable to false negative errors (falsely classifying guilty suspects as innocent), whereas the CQT is vulnerable to false positive errors (falsely classifying innocent suspects as guilty).
Can the Guilty Learn to Beat the Polygraph?
Is it possible to use countermeasures to beat the polygraph?
Countermeasures: As applied to polygraph research, techniques used to try to conceal guilt
30 minutes of instruction on the rationale underlying the CQT was sufficient for community volunteers to learn how to escape detection in a mock-crime study. Participants were told to use either physical countermeasures (eg., biting their tongue or pressing their toes on the floor) or mental countermeasures (e.g., counting backward by 7 from a number greater than 200) when asked a comparison question during the polygraph exam.
Both countermeasures worked, with 50% of the guilty suspects beating the polygraph test. In addition, the polygraph examiners were not able to accurately detect which participants had used the countermeasures.
Anti-anxiety medications also do not have an effect on the accuracy of the CIT.
Scientific Opinion: What Do the Experts say?
Most knowledgeable scientists are skeptical about the rationale underlying the CQT and its accuracy. But what does the public believe? There were two surveys: one done with 195 members of the Society for Psychophysiological Research and another one done with 411 people from the general public. Both samples are not very supportive of the polygraph and using the results in courts. However, the public does have more positive beliefs concerning the accuracy of the polygraph.
The United States National Research Council (NRC) established a panel of 14 scientists and 4 staff to review the validity of the polygraph (NRC, 2003). In a comprehensive report, the committee concluded the following:
- The theoretical rationale for the polygraph is quite weak, especially in terms of differential fear, arousal, or other emotional states that are triggered in response to relevant and comparison questions
- The existing validation studies have serious limitations. Laboratory test findings on polygraph validity are not a good guide to accuracy in field settings. They are likely to overestimate accuracy in field practice, but by an unknown amount
- In summary, we were unable to find any field experiments, field quasi-experiments, or prospective research-oriented data collection specifically designed to assess polygraph validity and satisfying minimal standards of research quality
- What is remarkable, given the large body of relevant research, is that claims about the accuracy of the polygraph made today parallel those made throughout the history of the polygraph: practitioners have always claimed extremely high levels of accuracy and these claims have rarely been reflected in empirical research
Despite scientists' negative view of it, the CQT is still used by law enforcement as an investigative tool. To understand why, we have to know only that whatever its actual validity, the polygraph will cause many suspects to confess, thereby providing resolution of the criminal investigation
Admissibility of Polygraph Evidence
Polygraph results were first submitted as evidence in court in the United States in Frye v. United States (1923). James Frye was denied the opportunity to have the results of a polygraph test conducted by William Marston admitted as evidence. This ruling led to the requirement that a technique must obtain “general acceptance” by the relevant scientific community before it can be admitted as evidence.
*Polygraph evidence is not admissible in Canadian criminal courts of law*
In R. v. Beland (1957), the Supreme Court of Canada ruled that polygraph evidence should not be admitted to help to determine whether a person is telling the truth. The court referred to the polygraph as being falsely imbued with the “mystique of science,” thus causing jurors to give polygraph evidence more weight than it deserves when determining the verdict.
BRAIN-BASED DECEPTION RESEARCH
In the past decade, researchers have attempted to use brain-based responses to detect deception.
Event-Related Brain Potentials: Brain activity measured by placing electrodes on the scalp and by recording electrical patterns related to presentation of a stimulus
ERPs reflect underlying electrical activity in the cerebral cortex. One type of ERP that has shown promise is known as the P300. This ERP occurs in response to significant stimuli that occur infrequently. When using CIT procedures, guilty suspects should respond to such crime-relevant events with a large P300 response, compared with noncrime-relevant events. No difference in P300 responses to crimerelevant and irrelevant events should be observed in innocent suspects. One of the advantages of ERPs is that they have been proposed as a measure resistant to manipulation.
Several studies have been conducted to assess the validity of the P300 as a guilt detector. Farwell and Donchin (1991) conducted one of the first studies on the use of the P300 to detect the presence of guilty knowledge. The study consisted of two experiments:
- In the first experiment, participants role-played one of two espionage scenarios that involved exchange of information with a foreign agent, during which they were exposed six critical details (e.g., the colour of the agent's hat).
- In the second experiment, participants were asked about details of minor offences they had committed in their day-day lives.
In the first experiment, using P300 as the measure, 18 of 20 participants were correctly classified in the guilty condition, and 17 of 20 were correctly classified in the innocent condition. In the second experiment, all four of the guilty participants were correctly classified, and three of the four innocent participants were correctly classified.
Although the results look impressive, there are several limitations to this study:
- First, guilty participants reviewed the crime-relevant details just prior to taking the CIT
- There were no aversive consequences linked to performance in this study
- The sample size, especially in the second experiment, was very small
Over the past decade, a possible new lie-detection technique using functional magnetic resonance imaging (fMRI) has been examined. fMRI measures the cerebral blood flow (a marker of neurological activity) in different areas of the brain. Researchers have used fMRl to determine which areas of the brain are associated with deception in a variety of deception paradigms, including forced-choice lies (i.e., responding yes when the truth is no), spontaneous lies (i.e., saying Vancouver when the true answer is Toronto), rehearsed lies, faking memory impairment, and concealed information tests (i.e. lying about hiding stolen money)
The most consistent finding from the studies is that the lie conditions produce greater activation in the prefrontal and anterior cingulate regions as compared to truth conditions.
These findings and others indicate that brain-imaging techniques can differentiate which parts the brain are involved in lying and may even indicate which areas are associated with different types of lying.
A limitation of this research, however, is that it is typically based on averaging fMRI data across multiple participants, which constrains its use for detecting deception in individuals. In addition, studies have used healthy volunteers who have been screened for neurological and psychiatric disorders, who have been instructed to lie, and who do not face any serious consequences.
There are also concerns that brain-imaging evidence may have a particularly powerful influence on juror decision making. Simply seeing pictures of brain images makes people believe that scientific results are more valid compared to when other images (such as figures or tables) are presented.
Recently, McCabe, Castel, and Rhodes (2011) examined the influence of evidence from the polygraph, fMRl lie detection, or thermal imaging on verdicts in a mock juror trial. Evidence from fMRI lie detection resulted in more guilty verdicts compared to the other types of evidence presented. However, the researchers found that if expert testimony was included that questioned the validity of fMRI lie detection, the number of guilty verdicts was reduced to a similar level as in the other conditions.
Box 4.2 Cases in Forensic Psychology
Brain Fingerprinting: is This Admissible in Court? Harrington v. State (2003)
Terry Harrington was a man convicted of murder who attempted to use the results of brain based deception testing to prove his innocence.
The case that put brain fingerprinting in the news was Harrington v. State (2003). On July 22, 1977, retired police officer John Schweer was shot and killed while working a security guard for a car dealership in Iowa. Seventeen year-old Terry Harrington and Curtis McGhee were arrested for the murder. At his trial, Terry Harrington claimed he was not at the crime scene and several witnesses testified that Harrington had been at a concert on the night of the murder.
The prosecution's key witness was another teenager, Kevin Hughes, who testified that he was with Harrington and McGhee on the night of the murder. According to Hughes, the three teenagers decided to steal a car. They went to the car dealership. Hughes testified that he waited in the car while Harrington, who first removed a shotgun from the trunk, and McGhee went around a building at the car dealership.
Huqhes claimed he heard a gunshot and that Harrington and McGhee came running back to the car. Hughes testified that Harrington had stated he had just shot a cop. Both Terry Harrington and Curtis McGhee were convicted of first-degree c murder and sentenced to life in prison without the possibility of parole.
Throughout his 25 years of imprisonment, Terry Harrington maintained his innocence, but all his attempts to appeal his conviction were unsuccessful. From his prison cell, Harrington heard about a new technology that might help his case. He contacted Lawrence Farwell, a cognitive psychophysiologist and head of Brain Fingerprinting Laboratories.
On April 18 and 25, 2000, Farwell came to the Iowa State Penitentiary to test Harrington to determine if he had knowledge of the crime scene and of details about his alibi (the concert he claimed he attended). Farwell measured the amplitude of Harrington's P300 brain potential to irrelevant and relevant crime scene and concert details. According to Farwell, Harrington's lack of P300 response to crime-relevant details indicated that Harrington had not participated in the murder. In contrast, Harrington showed a prominent P300 to alibi-relevant information.
Although the results of the brain fingerprinting were entered as evidence, the judges relied on other evidence to overturn the murder conviction. During the hearing, three of the prosecution witnesses recanted their testimony. Kevin Hughes stated that he had made up the story about what happened the night of the murder. Hughes claimed that he lied to obtain the $5000 reward being offered about the murder and to avoid being charged with the crime. In addition the police failed to turn over all the police reports to Harrington's defence lawyer. These reports documented the police investigation of an alternative suspect. On February 26, 2003, the Supreme Court ox Iowa overturned the murder conviction of Terry Harrington and the case was remanded for a new trial. On October 24, 2003, the Pottawattamie County Attorney announced that he was dropping the murder charges against Terry Harrington. Harrington and McGhee sued the prosecutor and the State of Iowa and received a joint $12 million out-of-court settlement.
Verbal and Nonverbal Behaviour Cues to Lying
On average, North Americans tell one to two lies per day. However, a small number of people are prolific liars. The most common method of deception detection is through the analysis of verbal characteristics and nonverbal behave iours.
The underlying assumption is the same as that for polygraphy: The act of deception produces a physiological change compared with telling the truth.
The argument here is that it is more difficult for people to control aspects of their nonverbal behaviour than their verbal behaviour. The typical experiment involves one group of participants (called the message source) who are told to provide either true or deceptive messages.
- For example, DePaulo, Lassiter, and Stone (1982) asked participants to honestly describe people they liked and disliked.
- They also asked participants to describe the same people dishonestly (i.e., to pretend to like the person they disliked, vice versa).
- Another group of participants was asked to detect when the message source participants were truthful or deceptive.
- Participants who were instructed to focus their attention on the message source participants' tone of voice were more successful at detecting deception than those participants given no special instructions.
Researchers have also assessed facial cues and other nonverbal cues to deception.
If a liar is not feeling excited, scared, or guilty, or when the lie is easy to fabricate, behavioural cues to deception will likely not be present.
When people are attempting to conceal an emotion, the true emotion may be manifest as a micro facial expression. The microexpressions are brief facial expressions reflecting the true emotions the person is experiencing. In response to terrorists concerns, the United States has been training security officers at airports to use this technique (reading concealed emotions in people) to identify potential threats. Matsumoto, Hwang, Skinner, and Frank (2011) recommended that during interrogations investigators pay attention not only to what a suspect says but also to the suspect's facial expressions.
Stromwall, Hartwig, and Granhag (2006) explored the role of stress by creating a realistic deception scenario that used experienced police officers, employed long interrogations, and generated suspects who were motive driven and had adequate time to prepare their deception. Participants were offered $30 to tell a biographical story to a police officer and were randomly instructed to be honest or deceitful. To create motivation and higher risk, participants were offered an additional $20 if they were able to convince the officer that they were being truthful. Liars felt more anxious and stressed during the task when compared to truth-tellers. No differences in nonverbal behaviours were observed. For verbal strategies, the majority of truth-tellers claimed to ”keep it real” (50%), whereas liars would “keep it simple (46.7%).”
Table 4.3 describes the types of verbal and nonverbal indicators used to detect deception (pg.105). The verbal indicator that has been most strongly associated with deception is voice pitch. Liars tend to speak in a higher-pitched voice than those telling the truth. Most studies have found increased use of speech disturbances (“ah,” “umm”) and a slower rate of speech during deception
Summary, it appears that cognitively more difficult lies (lies in which you have to fabricate an answer) may be associated with one pattern of speech disturbances, whereas cognitively simpler lies (lies in which you must conceal something) may be associated with a different pattern of speech disturbances.
Verbal Cues to Lying
In a comprehensive meta-analysis, DePaulo and colleagues (2003) coded 158 cues to deception from 120 samples of adults. Most of the verbal and nonverbal behaviours coded did not discriminate between liars and truth-tellers.
One of the most reliable indicators was that liars provided fewer details than did truth-tellers. Liars also told less compelling accounts as compared with truth-tellers. For example, liars' stories were less likely to make sense (less plausible, lack logical structure, have discrepancies), were less engaging, and were less fluent than were truth-tellers stories. Liars were also rated as less cooperative and more nervous and tense than truth-tellers. Finally, truth-tellers were more likely to spontaneously correct their stories and more likely to admit to a lack of memory than liars were. Deception cues were easier to detect when liars were motivated to lie or when they were attempting to cover up a personal failing or a transgression.
Truth-tellers used more rhythmic hand gestures than liars, whereas liars made more speech-prompting gestures than truth' tellers when describing an event. These more subtle cues to deception may warrant further investigation.
Are Some People Better at Detecting Deception?
Across studies, the ability to distinguish lies from truth tends to be only slightly better than chance. A meta-analysis by Aamodt and Custer (2006) found that on average the accuracy rate for detecting deception for “professional lie catchers,” such as police officers, judges, and psychologists, was 55.5%, a rate that is not much more accurate than that of students and other citizens (who had 54.2% accuracy). Thus the accuracy rate for both professionals and students is just barely above what would be obtained from guessing (50%).
This poor performance in deception detection has been explained in three ways:
- First, people tend to rely on behaviours that lack predictive validity. Laypeople have a number of beliefs about lying. The most common stereotype about liars is that they avoid eye contact, police officers share belief in these stereotypes: They believe that two cues indicative of deceit are eye gaze and fidgeting. Unfortunately, these two cues have not been found to be related to deception
- Second, most people have a truth-bias. It's the tendency of people to judge more messages as truthful than deceptive
- Third, there are only small differences between truth-tellers and liars.
- With so few cues to rely on, it is a challenging task for people to identify liars
There are also cases where there are lie detection “wizards” who are really good at lie detection. It is too premature for government officials to start using them.
In a review of 40 studies, Vrij (2000) found a 67% accuracy rate for detecting truths and a 44% accuracy rate for detecting lies. In most of the studies, the professional lie catchers were not very accurate at detecting deception. The results also showed that in most studies, truthful messages were identified with more accuracy than deceptive ones. Thus, even professional lie catchers have a truthfulness bias.
Mann, Vrij, and Bull (2004) examined police officers' ability to detect lies and truths told by suspects during police interrogations. These police officers were able to reach accuracy rates similar to those of more specialized law enforcement groups, such as U.S. Secret Service agents (There are two potential explanations for the higher-than-usual accuracy:
- First, the suspects were highly motivated to lie, and research has shown that high-stakes lies are easier to detect than low-stakes ones.
- Second, the police were more familiar with the setting and type of individual they were judging, namely suspects.
Neither level of experience nor confidence in deception-detection ability is associated with accuracy rates. The reason that confidence is unrelated to accuracy may be that people rely on cues they believe are related to deception and when they see these cues, their confidence increase. However, since the cues people believe are related to deception are often not valid, their accuracy tends to be poor.
Research by Bond and DePaulo (2008) found that there are no specific traits related to detecting deception in others. They concluded that it “deception judgments depend more on the liar than the judge.”
Is it possible to improve professionals' deception-detection abilities? Although there have been only a few studies examining this, it appears that training programs can help professionals become more accurate. Porter, Woodworth, and Birt reported that a two-day workshop focusing on myth dissolution, knowledge of behavioural cues to deception, and performance feedback increased the detection accuracy of parole officers from below chance levels (40%) to 76.79%.
Thus, although detecting deception is difficult, it is possible to improve judgment accuracy through training. Future research is needed to design and evaluate what types of training programs are most effective at helping professionals become better at detecting deception.