Taking a polygraph test is commonly referred to as being put on "the box." Last week the Supreme Court hammered a nail into the box by upholding a ban on the use of so-called "lie detectors" in military trials. As a scientist who studies lying, and as a consultant to the government on polygraph tests, I am relieved by the court's 8-1 decision in U.S. v. Scheffer to keep this unreliable physiological test out of courtrooms. We should breathe a collective sigh of relief that the responsibility of juries to make judgments about guilt and innocence will not be diminished.

Airman Edward Scheffer was convicted by a military court-martial of illegal drug use after he tested positive for methamphetamine on a urine screening. At his trial, Scheffer sought to introduce the results of a government polygraph test that indicated he had not knowingly ingested illegal drugs. The military judge refused to admit this evidence, but an appeals court overturned the decision. The government then appealed to the Supreme Court, which held that, given the lack of scientific consensus on the reliability of polygraph evidence, its exclusion does not violate a defendant's constitutional right to present a defense.

A defendant has a 6th Amendment right to present any "relevant and reliable" evidence, but the right is not unfettered. If one substituted "astrological chart" for "polygraph," very few would think it reasonable to present such evidence to a court. Polygraph tests as evidence of "truth" are unreliable because they measure physiological changes that have only a tenuous link to honesty. A defendant's nervousness when asked about a crime is irrelevant to whether he or she is telling the truth. People do not show a unique and universal bodily reaction when they lie.

Scheffer's polygrapher asked him questions about his use of illegal drugs, about other crimes and about unrelated matters. The polygraph recorded his physiological reactions -- changes in heart function, perspiration, and respiration -- as he answered each question. The polygrapher assumed that because he responded less anxiously to questions about the crime than to unrelated questions, he was truthful. But this assumption makes no sense. One need not be a psychologist to know that someone who is lying may be unemotional, just as someone with nothing to hide may nonetheless be anxious.

There are many possible reasons why anxiety is not related to lying. Defendants, for example, may become emotionally inured to questions about a crime that they have answered (however falsely) many times, but may react emotionally to unexpected questions. Such people will be incorrectly judged truthful by a polygrapher. Indeed, the most callous, cynical liars, those most indifferent to the consequences of their crimes, may be the biggest beneficiaries of lie-detector testing. Anyone with a little knowledge about what polygraphers are trying to do should be able to "beat the box."

After decades of research, much of it summarized in a study I led for the Congressional Office of Technology Assessment, it is clear that there is no adequate way to validate that a polygraph detects truthfulness. Typically, the only cases we can confirm are those in which the subject confesses. As Justice Stephen G. Breyer noted in oral argument in the Scheffer matter, the lack of an absolute criterion of truth makes it impossible to validate polygraph testing. Thus, a polygraph test does not shed light on Scheffer's veracity, nor does it help us understand the more complex cases of O. J. Simpson and the au pair Louise Woodward, both of whom took lie-detector tests. By objective criteria we will probably never know the truth in these cases, and lie detector results do not change the situation.

"Lie detector" is a misnomer for a device that is no more than a fear detector. Its continued use in the absence of scientific evidence of accuracy testifies to our wish for a simple answer to complex human questions. Wouldn't it be wonderful if a truth machine really existed? Or would it? If we could get into people's heads, as the polygraph falsely promises, it would be truly frightening. Most people engage in casual lying on a daily basis, and a technology that could detect these lies would open up our most private experiences to public inspection. It is just as well that the complex workings of human perception, thought, and memory make the prospect of a lie detector fanciful. The writer is adjunct professor of social welfare and psychology at Brandeis University.