Are There Logical Errors Involved? We believe that pro- ponents of the polygraph commit two types of logical error. The first involves pointing to individual cases in which they claim a “lie detector” test either revealed a liar or forced a confession from a reluctant suspect. This is nothing more than “proof by testimonial,” as we see all the time in ads for “miracle” weight-loss products or engine-oil additives. The fact is that testimonials are no substitute for a controlled scientific test.
The other main logical error is oversimplification. By focus- ing on apparent successes, they may gloss over the failures— which, in the case of the polygraph, can be quite serious. As we will see, the polygraph failures can lead to a surprisingly large number of honest people being identified as liars.
What Is the Evidence? Without a doubt, wrongdoers sometimes confess when confronted with polygraph evidence against them. Yet, critics have pointed out several problems with the polygraphic procedure that could easily land in- nocent people in prison and let the guilty walk free (After- good, 2000). For example, polygraph subjects know when they are suspects, so some will give heightened responses to the critical questions even when they are innocent. On the other hand, some people can give deceptive responses be- cause they have learned to control or distort their emotional responses. To do so, they may employ simple physical move- ments, drugs, or biofeedback training—a procedure in which people are given moment-to-moment information on certain biological responses, such as perspiration or heart rate (Saxe et al., 1985). Either way, a polygraph examiner risks incor- rectly identifying innocent people as guilty and failing to spot the liars.
Important statistical issues call the polygraph procedure into further question. Even if the examination were 95 per- cent accurate, a 5 percent error rate could lead to the mis- identification of many innocent people as being guilty. To illustrate, imagine that your company arranges for all 500 of your employees to take a polygraph test to find out who has been stealing office supplies. Imagine also that only about 4 percent (20 out of 500 people) are really stealing, which is not an unreasonable estimate. If the lie detector test is 95 percent accurate, it will correctly spot 19 of these 20 thieves. But the test will also give 5 percent false positives, falsely fin- gering 5 percent of the innocent people. Of the 480 innocent
polygraph A device that records or graphs many (“poly”) measures of physical arousal, such as heart rate, breathing, perspiration, and blood pressure. A polygraph is often called a “lie detector,” even though it is really an arousal detector.
false positive Mistaken identification of a person as having a particular characteristic. In polygraphy, a false positive is an erroneous identification of a truthful person as being a liar.
very well, according to reports by the American Psychological Association and by the U.S. government’s Office of Technol- ogy Assessment. In general, like the polygraph, these instru- ments seem to be more accurate than mere interviews, but they also suffer from a high false-positive rate.
More recently, researchers have turned to brain scan- ning techniques to see if they can catch liars (Ross, 2003). A certain brain wave pattern known as P300 has been linked with a variety of attention-getting cues, such as hearing one’s name, but studies show it can also be evoked by fibbing. In addition, fMRI images show that lying activates all the brain areas involved in telling the truth, plus several more (Langleben et al., 2002). This suggests that lying is not something com- pletely separate from the truth but an operation the liar must perform on the truth, says psychiatrist Daniel Langleben, all of which raises the concern that there is too much hype and too little solid evidence behind brain-scan-based lie detection (Gamer, 2009; Stix, 2008). In addition, some neuroscien- tists worry about the ethics of peering directly into people’s brains to “read” the neural traces of their private thoughts (Pearson, 2006).
The potential advantage of these newer brain-scan tech- niques is that they bypass the anxiety-response pathway used by polygraphy. By registering neural activity, they get much closer to the person’s actual thoughts. But how well do these alternative methods work? Not well enough for the police and the courts—yet.
Finally, Paul Ekman—the same one who studies universal facial expressions of emotion—has found that liars often display fleeting “microexpressions” and other nonverbal cues. In one study, Ekman and his colleague Maureen O’Sullivan found that some people are especially good at detecting deception, but they are a small minority. In their tests, most people perform at about the chance level. Still, Ekman and O’Sullivan hope to learn what those most skilled at detecting deception look for and teach that to police officers and other concerned with crime and security issues (Adelson, 2004).
employees, the polygraph will inaccurately implicate 24 as liars. That is, you could end up with more people falsely ac- cused of lying than people correctly accused of lying. This was borne out in a field study of suspected criminals who were later either convicted or declared innocent. The polygraph re- sults were no better than a random coin flip (Brett et al., 1986).
An equally serious concern with polygraphy is that there are no generally accepted standards either for administering a polygraph examination or for interpreting its results. Dif- ferent examiners could conceivably come to different conclu- sions based on the same polygraph record.
What Conclusions Can We Draw? For these reasons, the U.S. Congress has outlawed most uses of polygraph tests in job screening and in most areas of the government, except for high-security-risk positions. National Academies of Science (2003) has gone even further in a report saying that the polygraph is too crude to be useful for screen- ing people to identify possible terrorists or other national security risks. Your authors agree.
As far as criminal investigations are concerned, we find a patchwork of laws on the admissibility of polygraph evidence among the states. Few have gone so far as imposing complete bans and 20 more allow such evidence only on agreement of both sides—although, in a few states, polygraph results are still routinely admissible in court (Gruben & Madsen, 2005). So where do you come down on the issue of using the polygraph in criminal cases? Does the fact that a “lie detector” test can sometimes force a confession from a suspect justify its use?
Alternative Approaches to Deception Detection The reining-in of polygraph testing has spurred the develop- ment of alternative means of detecting dishonesty (Capps & Ryan, 2005; Lane, 2006). Much of this work has been de- voted to paper-and-pencil instruments that are often called “integrity tests.” How well do these instruments work?