This story is from Texas Monthly’s archives. We have left the text as it was originally published to maintain a clear historical record. Read more here about our archive digitization project.

“There must be some mistake here,” said the young woman with the wraparound eye shadow and nervous hands. “I don’t know anything about this. Believe me, I’m telling the truth.” We were seven in a cluttered conference room at the Harris County Courthouse—three prosecutors, the young suspect, her two friends, and I—engaged in one of criminal justice’s more elemental processes: the discernment of truth from falsehood. It’s a ritual repeated countless times in any day, at any courthouse. If most of the blank faces that show up in that room every morning wind up pleading guilty, most of them also claimed at one time or another that they weren’t.

I was there to test my own powers of lie detection for a couple of reasons. For one thing, ever since becoming hopelessly addicted to the Perry Mason television series as a youngster, I have wondered if Mason’s knack for selecting the lie out of a dizzying array of convincing alibis had any kinship to fact. TV melodramatics aside, it is possible for a person to divine a lie without the aid of physical evidence or scientific device. But can it be done with certainty? A percentage of defendants can be judged guilty or not guilty on physical evidence alone, but a lot of cases are more circumstantial: they come down to the crook’s word against the state’s.

The bare facts of this particular woman’s case seemed to militate in her favor. She was charged with “unauthorized use of a motor vehicle”—meaning that if you drive around in a car that you know is stolen, you’re in just about as much trouble as the fellow who swiped it. The woman had been picked up one morning about two, after police officers noticed a pair of suspicious-looking cars parked in a shopping center lot in Southwest Houston. One car was unoccupied, though a front door was ajar; the woman and a boyfriend were in the other car. Running a quick computer check on the vehicles, the police discovered that both were stolen.

The woman admitted she had been driving one of the cars for a couple of months. But she claimed she didn’t know it was stolen. Yet another male friend had allowed her to use it. She was adamant about her innocence to the point that she had decided to take the risky move of talking to the prosecutors about it. That, among other things, had led me to an initial judgment that she was probably telling the truth.

As the interrogation progressed, the prosecutors homed in on the crucial issue: how could she not have known that the car was hot? “This guy was a boyfriend, wasn’t he? How well did you know him?” She responded that she knew him pretty well. “What’s his last name?” “I don’t remember,” she said flatly. Now I was interested.

The woman stumbled and stuttered. She had dated the guy for a while, but she couldn’t explain why she didn’t know his last name. No, he never said anything to her about the car’s being hot. In fact, she’d been with him in Livingston County a month or so before, and had been stopped for speeding. The cop had run a computer check on the car: nothing. “So, you see, there was really no way I could have known,” she said. I didn’t believe her anymore.

I was bothered by a couple of things. She was increasingly restive; her voice quaked and her hands were all over the place. She’d obviously lied about her relationship with the boyfriend, so she was lying about her relationship with the car. Simple. “She’s lying like hell,” I told a prosecutor after the suspect left. “I’d nail her. I really would.”

The young suspect eventually pleaded guilty to the charge and accepted a misdemeanor punishment—though the crime remained on the books as a felony. I felt cocky. I’d sniffed out a liar in less time than it takes to read the sports page. Later that afternoon, though, I observed what surely has to be the most sobering sight in any courthouse—that long line of somber-faced, chained-together inmates being led from the holdover tank back to the county jail—and I wondered: what if my instincts were wrong?

Of course, the courts don’t rely on instinct to determine someone’s guilt or innocence. But one of the grander ironies of criminal justice is its refusal to embrace new lie detection technology the way it has other gadgetry. While the courts have begun to admit all manner of exotic hair and skin comparison tests, the system has been more than reluctant to recognize the polygraph exams to be admitted as a matter of course under certain conditions, but others, including Texas, will accept the results in court only if the test has been taken voluntarily and only if all parties agree to allow the jury to hear it—and sometimes not even then. Lying is still judged largely by instinct. In many criminal cases, that would seem to be enough. Most crooks are not clever liars; indeed, most criminals’ lies are as simple-minded and spontaneous as the crimes they are meant to cover up.

Police interrogation, like criminal lying, has been mythologized by literature and TV. Few true-life interrogations include scorching overhead lights, intricately acted good-guy, bad-guy routines, or sobbing, eleventh-hour breakdowns by suspects. Most are simply rather informal conversations during which the interrogator does more listening than talking. Object number one is to allow the suspect to tell the truth if he’s inclined to.

Houston detective Gil Schultz, a ten-year veteran of almost daily interrogations of criminals, says this is most easily accomplished by treating the suspect like a human being. “Nothing will catch him off guard more than that,” he says. “A lot of crooks want to get it off their chest. Your chances are better if you treat the guy nicely, call him by his first name. You want to be sure he knows you’re in charge, but browbeating won’t work. A tough guy will get into a macho thing with you.”

If the crook is not inclined to confess, the interrogator may have to resort to any of a number of simple gimmicks. The object of the interrogation then becomes allowing the crook to lie, hoping he can be caught in his deception later. Some interrogators choose to catch the suspect with contradictory evidence after he has lied: “Don’t tell us you weren’t there, Larry. Your prints are all over the doorknob.” Others may bait the crook by confronting him with some nonexistent piece of incriminating evidence; if the suspect tries to explain it away, the interrogator knows he’s got a liar on his hands. There are myriad other tip-offs that tell a smart interrogator that his suspect is lying: conspirators whose alibis are perfectly matched; the conspirator who refuses to speculate on who might have committed the crime and says something like “That’s for the police to find out”; the crook whose recollection of the hours just before or just after the crime is very specific but whose memory of the time frame coincident with the crime is vague; the thief or robber who volunteers restitution to the victim, saying, “I didn’t do it, but I’ll pay him back if it’ll help.”

In any case, if the suspect sticks by his denial, the best the police can do is pinpoint enough contradictions in his story to suggest to a jury that he’s not telling the truth. But is the suggestion of deception, evidence merely indicative of lying, enough for twelve laymen sitting on a jury? Why won’t the courts turn to other methods?

Some of their reluctance to admit polygraph exams has to do with the libertarian streak in our collective consciousness. Even in this era of fervent law-and-order righteousness, there is a resistance to endangering in any way the rights of that one in a hundred who is innocent. The polygraph machine can’t guarantee accuracy; it is no more sophisticated than similar medical equipment. It’s a simple contraption designed only to measure spontaneous anxiety. Two large tubes are wrapped about the subject’s upper and lower lung area to measure respiration; a cuff is attached to his arm to measure blood pressure and pulse rate; finally, small electrodes are attached to the fingertips to detect “galvanic skin response,” or sudden perspiration, the most reliable physiological reaction. These raw data are fed through a mass of wires to the so-called lie box, which regurgitates it to the polygraph operator in the form of undulating inked lines on graph paper. The idea is as simple as the hardware: because certain physiological symptoms of anxiety—sweating, increased blood pressure, and slow, shallow breathing—are commonly associated with lying, it stands to reason that if a suspect answers a question truthfully, the undulations on the graph paper will remain relatively stable. If he les, one or more of the pens will jump farther and faster on the paper.

Because the lie detector apparatus is simple to operate, it has been subjected to a great deal of misuse. Private industries—notably security services and retail stores—have begun administering exams to prospective employees, presumably to test their general honesty. Writer Dick Reavis “beat the box” three times recently while researching a story on the private security industry in Texas. Given what Reavis was asked—and how—I’m not surprised. He was asked many questions about his background and some other questions clearly designed to find out if he was basically honest. The polygraph simply cannot detect such things; it can ferret out only a specific lie about some significant wrongdoing that occurred in the past, and then only in the hands of a skilled operator. It may take three or more tests to detect a lie. The polygraph examiner can’t isolate lies if he conducts a rambling fishing expedition; he is not clairvoyant, and he cannot predict future dishonesty. And no examiner can reliably detect dishonesty when there is nothing at stake.

Reavis even got away with using a fictitious name; that alone would seem to be a damning indictment of the polygraph. But one of the main requirements for careful lie detection is that the polygraph operator know in advance what the suspect might lie about. For example, if a cop picks up a vagrant on the street, takes him downtown, and tells a polygraph operator, “This guy must have done something wrong; let’s find out what,” it’s doubtful the machine will produce the desired results. On the other hand, if the vagrant is picked up around the scene of a rape and is asked specific questions about the incident—times, places, colors of clothing—chances are good a polygraph test will reveal whether he is lying about his involvement. Reavis’s examiners had no specific reason to suspect that he might lie about his name. Hence the false name was an undetectable lie.

A good control question—one that establishes a context against which the suspect’s responses to specific questions can be compared—is vital. It is usually about a situation unrelated to the crime in both time and place but of a similar nature. For example, a store employee suspected of covering up a theft might be asked: “Between the ages of fifteen and twenty, did you ever observe something being done that was wrong but fail to tell anyone about it?” The polygrapher would follow that with: “Do you know who took the money from that safe on May 1?” Because the machine can measure only information based on simple autonomic responses normally associated with the act of lying, the skill of the interrogator in his questioning and interpretation is crucial to obtaining a valid conclusion.

Eric Holden, chairman of the committee on standards and ethics for the American Polygraph Association, asserts, “As in any other scientific procedure, a small probability for error exists. But if the subject is properly prepared, if appropriate questions have been designed and well defined, and if recognized procedures are followed, an examiner should be able to evaluate with a high degree of accuracy whether or not deception has occurred.” It’s hard to imagine that some operators haven’t shaded their questions—or their interpretations of the answers—to benefit whoever is paying the bill. Injustice, however, is done to the subject more often through sheer incompetence. Unless the operator takes care to acquaint himself with the subject first and to run through the questions repeatedly (a thorough test should take at least an hour and a half), a truthful person can come out looking like a liar, and vice versa.

Don McElroy, a veteran of 28 years in the polygraph profession, explained how this works one afternoon at his office in Irving. “Some people are just more anxiety-prone than others. Call them hyper or whatever,” he said. “Unless you’re careful with them, they could be telling the truth but showing lie-level anxiety on the box because of general anxiousness about being interrogated or being in trouble.” There is also a certain percentage of the population that will not test normally, the poorest subjects being psychotics whose mental limitations prevent them from responding in any ordinary way.

Even for people who are neither excessively anxiety-ridden nor functionally “guiltless,” too much coffee in the morning or one too many at lunch can affect the box. A bartender friend of mine, whose employer administered polygraph exams yearly, claimed he could consistently beat the box by getting half in the bag or dropping a Quaalude before the test.

People can beat the box, but the chances of doing so can be diminished considerably by a skillful and prudent operator. “You can’t simply walk in there and ask the guy his name and age, and then hit him with something like ‘Did you shoot that woman over on Main last Thursday night?’ ” says McElroy. “An honest person will be shocked by the question and can show as much anxiety as the real killer when he responds, ‘No.’ ”

McElroy argued convincingly for the use of the polygraph in cases that lack physical evidence or an outright confession. But I was a little skeptical. I have always harbored a secret ambition to beat the box. McElroy allowed me to test myself against it, warning that it might not reveal much either way, since I was not accused of anything and any lie I attempted would have little stake attached to it. We agreed on a simple test that gave me every opportunity to get away with a lie. McElroy wrote four numbers on a small slip of paper—fourteen, fifteen, sixteen, and seventeen—and asked me to circle one; I circled fifteen. “Now, I’ll ask you about the numbers one by one. I’ll ask you if you circled each of them. Say no to each question. Lie about the one you circled.”

First, he wrapped the respiration-measurement tubes snugly around my upper chest and upper abdomen; then he taped the electrodes to my fingertips and strapped the blood pressure cuff around my left biceps. It was an eerie sensation, and when McElroy activated the box, I saw that my body was already responding to being trapped in an apparatus that perversely resembles the electric chair. Just so I could see how the box would register my reactions, McElroy kept it in clear view, but in any standard test this would not be allowed. “Let’s test your general sensitivity,” McElroy said. “Not right now, but in a second I’m going to rap you across the knuckles with this pen.”

Before I had time to absorb the threat, McElroy raised the ballpoint and feigned a blow at my left hand, stopping the pen just short of my knuckles. One, two, three seconds passed, and then the four needles on the box almost ran off the page. “Your sympathetic system is extremely sensitive,” he said matter-of-factly. “You’ll do just fine.” I sighed with relief, and the needles jumped again. I laughed, and they jumped even higher.

Now we were ready. “Did you circle the number fourteen?” “No.” The needles remained stable, jumping only slightly in response to my overall nervousness. “Did you circle the number fifteen?” I paused to psych myself for the moment; be calm, say this no just like you said the last one. “No,” I said. One, two, three seconds. The needles seemed stable, and for an instant I was confident that I had beaten the box. Four and then five seconds. And then the damnedest thing happened; the needles jumped, not berserkly, like they had when McElroy ambushed me with that false threat of harm, but enough to distinguish the response from my first, truthful no. Oh, well, I thought as we proceeded to sixteen and seventeen, maybe my system is generally agitated; surely my responses to the next two numbers will resemble my deceptive no enough to muddle any firm conclusion. “Did you circle sixteen?” “No.” “Seventeen?” “No.” Now I found myself trying to conjure up some anxiety, to generate some stress, to keep that lie from sticking out.

But when I glanced at the four needles, their undulations were maddeningly tiny and uniform. The box had me. It had not only isolated my lie, it had done so when I had nothing at stake and when I knew I was going to lie for several minutes in advance. While aware of the machine’s fallibility, I did wonder as I left that day if those needles wouldn’t have made me more certain of my instincts about the woman suspect. If that machine could pick out my harmless fib with such certainty, surely it could have told me a little bit more about the truthfulness of her claim that “there must be some mistake.”

I cannot advocate the completely unrestricted use of polygraph exams as evidence in court, but I do think the resistance to virtually any use of them in Texas is probably a case of misguided good intentions. Polygraph testing is a reliable form of lie detection, given a competent operator and a thorough exam. Some states have sought to ensure competence by imposing strict educational requirements on operators who test criminal suspects; others have refused to admit lie detector results unless a prescribed number of exams have been administered. In the mid-seventies, the New Mexico Supreme Court held that to deny a defendant the right to admit favorable polygraph results, if properly and competently arrived at, was a violation of due process; a New Mexico appellate court ruled that the prosecution should have equal rights of admissibility. A Massachusetts court placed polygraph exams under strictures similar to those applied to confessions. If the exam is requested by the defendant, and he knows that it will be admissible regardless of its outcome, and if the test is competently administered, the judge may admit it “in the proper exercise of his discretion . . . to be considered with all other evidence as to innocence or guilt.”

If polygraph exams given under carefully regulated conditions are, in fact, a valid procedure, why shouldn’t test results have as much weight as fingerprint or hair comparison or ballistics testimony? As the ratio of crimes and criminals to cops and courts continues to widen, criminal justice is going to have to enlist the aid of technology. With less and less time to devote to each case, cops and prosecutors and judges and juries might be tempted to be more cavalier in using their instincts. That probably doesn’t bother the ever-growing they-all-did-something-wrong-so-who-gives-a-damn lobby, but it bothers me, because I learned just how big the stakes are. Not long after my encounter with the young woman, I was chatting with the same group of prosecutors about another case in which lying was the crucial issue. A young ex-con with a record including a minor narcotics offense and an auto theft was charged with passing a bad check for about $300. He claimed in his statement to the police that it was all just a misunderstanding, that he was a regular customer at the bar where he wrote the check, that he’d cashed checks there before, that the thing just bounced and he was more than happy to make good on it. I surveyed the file. Recalling what I had been told about thieves who offer restitution, I began to look on the suspect with a jaundiced eye. His record bothered me, too. It’s hard not to believe that one who’s done it before probably has done it again. Finally, his explanation rang false. “Nail him,” I said confidently, closing the file. “He’s lying.”

A week or so later, one of the prosecutors gave me an update on that case. “You remember the rubber-check writer? He wasn’t lying. He was telling the truth, if you can believe it.” He proceeded to explain that an investigator had meticulously checked the fellow’s alibi and that he had, in fact, made a good-faith effort to reimburse the bar owner. “It just goes to show,” he said, “you never can tell.”

“Yeah,” I said, remembering how those four needles had not only jumped at my lying no but remained placid at my truthful ones.