A Consistently Erroneous Technology

James Randi

I’ve often been asked to participate as an advisor in court cases or investigations where I might be useful due to my experience as a conjuror, but I have always refused when required to accept evidence obtained via the polygraph, or lie detector as it’s commonly known. However, if I were to be called upon to deny that this silly device is effective or dependable, I’d have no hesitation in doing so. The evidence is just so much against this technology, it’s difficult to believe how long it has existed as a supposedly valid notion.

Look at the history of so-called “lie detection.” The device itself is a nightmare of tubes, wires, electrodes, and moving styluses, something right out of a Bugs Bunny production. By measuring and displaying changes in the subject’s respiration, heart rate, blood pressure, skin conductivity, and other variables of the human body, a complex series of graph lines is generated, and a technician is—theoretically—able to decide whether the answers to a set of questions were honestly given or not. We need not get into more involved aspects of the procedure such as who comes up with the questions to be asked—though that is a matter of primary importance, of course. Let’s examine opinions of the “professionals” who should know.

On November 5, 2002, on his Pentagon letterhead as Assistant Secretary of Defense for Command, Control, Communications, and Intelligence, this memorandum was sent by John P. Stenbit to the directors and administrators of all major military offices of the Department of Defense:

SUBJECT: Continued Use of Polygraph Techniques
I have recently reviewed the report of the National Research Council on the existing scientific evidence on the validity of the polygraph technique. While the report contained many findings that may ultimately lead to improved methods for the detection of deception, I think it is important to emphasize that the National Research Council found that none of the potential new technologies for the detection of deception showed any promise of supplanting the polygraph technique for screening purposes in the near term.

I believe this could have included Tarot cards and Ouija boards as well and made just as much sense. Note that it doesn’t say that the polygraph works at all! The truth is that it’s a useless high-tech assemblage that has consistently failed double-blind tests of its efficacy, but the way the above paragraph was composed, there is a strong suggestion that the flummery actually has performed as claimed. This next paragraph is even worse, repeating the same inane claim and further implying that the polygraph really works:

As the Department continues to research alternative technologies in this critical area, I believe it is important to remember that the National Research Council Report determined that the polygraph technique is the best tool currently available to detect deception.

“The best tool”? No, I’d vote for a pair of dice. Next, all in one paragraph, Secretary Stenbit referred to the polygraph as “an important tool”—twice!—when it is virtually useless, according to that same report by the National Academy of Sciences/National Research Council (as we shall see). This is simply a blatant denial of fact to bolster an erroneous—though official—opinion:

In the coming months, our country will face many unique challenges to our national security. The polygraph technique remains an important tool to detect deception in selected national security and law enforcement matters. Where appropriate and authorized, I recommend that we continue to use the polygraph technique as an important tool in our total decision-making process.

Late in 2002, as if in reaction to Stenbit’s comments, the U.S. National Research Council, a branch of the National Academy of Sciences, issued their official report on the use of the polygraph. Secretary Stenbit completely ignored this report’s key finding that the use of the device was unjustified—and thus even dangerous to national security, nor did the report characterize polygraphy as “the best tool currently available to detect deception,” as Stenbit had written. In their summary, the Council actually concluded: “Overall, the evidence [for polygraph validity] is scanty and scientifically weak… and some potential alternatives to the polygraph show promise, but none has yet been shown to outperform it.”

Este artículo también está disponible en español.
Haga clic aquí para leerlo.

Since the polygraph works about as well as the Oracle at Delphi—another shaky premise, though much older—this is not at all supportive. Regarding polygraphy technology in general, the NAS/NRC stated that “…there is essentially no evidence on the incremental validity of polygraph testing, that is, its ability to add predictive value to that which can be achieved by other methods.”

Let’s examine the realities, folks. A CBS News account of a suicide bombing at a CIA base in Afghanistan stated, regarding the careless acceptance of a spy on the base: “The double agent was brought onto the base without first being given a polygraph test, one of the basic tools in establishing a spy’s trustworthiness.” This is nonsense; the polygraph simply does not work.

Aldrich Ames

The history of this farce is long and involved, with uninformed “experts” raving about success and law enforcement agencies smugly accepting results as if they were valid. In addition to the Afghanistan case above, some very strong examples of this error stand out: In 1985, use of the polygraph did not expose the fact that Larry Wu-Tai Chin, a Chinese language translator working for the CIA in a critical capacity, was selling crucial information to China and had done so for thirty-three years, despite having been regularly subjected during that time to polygraph tests. In 1994, Aldrich Ames, a high-ranking CIA analyst, routinely passed all his polygraph exams, though for years he was a master Soviet spy who passed information on to his real employers. (Aldrich Ames even confirmed this directly to the Skeptical Inquirer. After Sandia National Laboratories Senior Scientist Alan P. Zelicoff published a denunciation of polygraph testing in SI in 2001, Ames wrote to SI from prison agreeing with the article and calling polygraphs “junk science” and “a superstition.” SI published his letter [vol. 25, No. 6, 2001]; see also Morton Tavel, MD’s, January/February 2016 SI cover article “The Lie Detector Test Revisited.”) In 2001, Robert Hanssen of the FBI went similarly undetected despite the regular periodic “screening” of FBI employees using polygraphs. His treachery was described by a review of FBI Security Programs as “possibly the worst intelligence disaster in U.S. history.”

The fact is that not a single spy has ever been caught by a polygraph screening exam. In 2003, the National Academy of Sciences issued its final report titled The Polygraph and Lie Detection that found the majority of polygraph research to be, in their words, “unreliable, unscientific, and biased” and that in national security matters and for law enforcement use, the level of accuracy drops to such a level that “its accuracy in distinguishing actual or potential security violators from innocent test takers is insufficient to justify reliance on its use in employee security screening in federal agencies.”

The NAS also found the high rate of false positives obtained with the device to be unacceptable. Physicist and CSI Fellow Bob Parks, with his usual degree of wry humor, reported on this situation: “I have argued, however, that the small number of true positives is the real problem. I propose replacing the polygraph with a coin toss. That would identify 50 percent of the double agents compared to zero with the polygraph. The unfortunate increase in false positives constitutes collateral damage, which is inevitable in war.”

The media, too, are delighted with this electronic Bozo, which adds flashing lights, buzzing noises, and mystery to so many of their stories. In the beginning of 2010, a prospective adoptive-couple—named by the FBI as “persons of interest” in the disappearance of an eight-month-old baby in Arizona—appealed on television to be administered a polygraph test to determine whether they were telling the truth. The test was duly administered, and the following day the polygraph examiners announced the result: inconclusive. Still, the general public perception is that the polygraph is a scientific device that works. The media rarely if ever mention the strong, well-founded scientific objections to the validity of the thing. In fact, as wag Bob Park—again—comments: “The polygraph looks for spikes in blood pressure, heart rate, respiration and perspiration. In other words, you can’t tell a lie from the sex act.”

Said Stephen E. Fienberg, chairman of the statistics department at Carnegie-Mellon University in Pittsburgh, who led the panel appointed by the National Academy of Sciences to evaluate the worth of polygraphy: “It’s everywhere—every three- and four-letter agency you can imagine, including the US Postal Service.”

That panel’s report concluded that almost a century of research may have produced a pseudoscience good for tricking naive people into blurting out the truth, but not much else. Thus, Chairman Fienberg was rather surprised to find his panel’s report cited in favor of potentially raising the number of lie detector tests the Department of Defense (DOD) is allowed to administer annually.

In a report it filed with Congress in January, the DOD stated it had administered more than 11,500 of these tests in fiscal year 2002. That’s more than twenty a working day! Of that total, 4,219 were “counterintelligence-scope polygraph,” or CSP, exams, subject to a 5,000-exams-per-year limit under 1991 Public Law 100-180. In its own report, the DOD put Congress on notice that it might ask for authorization to conduct more than that number, and cited the NAS report in support, according to Steven Aftergood, who monitors polygraph policy for the Federation of American Scientists. That DOD report stated: “It is important to note that the NRC [National Research Council] Report also concluded that the polygraph technique is the best tool currently available to detect deception and assess credibility…. The Department will continue to use the polygraph technique as it has in the past, until improved technologies or methodologies are developed as a result of scientific research.”

Folks, anything would be an “improved technology” over this high-tech toy, the “best tool currently available”! Fortune cookies outperform it! Chairman Fienberg charitably called the DOD’s reference to the NAS report “disingenuous.” A DOD spokesman said it was drawn directly from the NAS panel’s conclusion that, while more promising technologies are on the horizon, none yet has supplanted polygraphy. He could have added that the Tooth Fairy has also not yet supplanted other means of providing funds to children. When asked, Don White, a spokesman for the Office of Inspector General (OIG) at the Department of Health and Human Services, would not discuss whether lie detector tests were part of their investigative procedure, but the DOD spokesman did name OIG as one of the government bodies that uses polygraphs.

To close this denunciation of this particular pseudoscience, I will first note that in the March/April 2013 Skeptical Inquirer, Dr. L.G. Wade Jr. commented on an article about phrenology, the “art” of reading bumps on the human head to determine character and talents. This was once actually used by employers, the military, and even in court proceedings to examine individuals, particularly in France, and was considered a real science, much as polygraphy still is today. Dr. Wade wrote:

I was first alerted to the fraud of the “lie-detector” by two cases: A county assistant treasurer who was falsely accused of embezzlement, and a distraught widower who was falsely accused of murdering his wife. Both of these accused “failed” a polygraph exam (whatever that means) and were immediately convicted in the newspapers, with their reputations destroyed. They were both subsequently exonerated. The polygraph measures indicators of stress, and honest people who are falsely accused of heinous crimes are likely to “fail” the exam. On the other hand, hardened criminals show little stress when they are asked about their crimes.

Unlike phrenology, polygraph tests are still used and accepted by law-enforcement and the government. They are still used and absolutely required for a wide range of employment with the FBI, CIA, and police agencies.

What should you do if required to take a polygraph test? If you agree to take one, you may be placing your reputation in the hands of an unwitting charlatan who can proclaim you to be guilty or innocent. If you refuse, you are assumed to be guilty. I would like to see CSI do a careful study and expose the use of this polygraph flummery. Our government should be pressured to abandon such pseudoscience.

Other involvements in court cases where my personal expertise has been sought were either very minor or very similar to those I’ve stated here. Most of them involved gypsy-style swindles, which are the sort most often encountered, I believe, because they take advantage of strong religious and/or specific ethnic superstitions. Again, religion is behind so many of our species’ problems of accepting reality.

Nothing new in that statement.

James Randi

James “The Amazing” Randi is a magician, investigator of psychic claims, author (Flim-Flam!, The Faith Healers, The Mask of Nostradamus, The Magic of Uri Geller), and the president of the James Randi Educational Foundation. He was a founding fellow of CSICOP. This article is based on a special presentation on investigating psychics he gave at the Fifth World Skeptics Congress, Abano Terme, Italy, October 8—10, 2004.