Archive for September, 2013

“How to beat a lie detector”

September 2, 2013

An Indiana prosecutor announced last week that a federal judge should seek extensive jail time for a Mr. Chad Dixon. By day, Mr. Dixon is your workaday Joe American: small-town Indiana, owner of a small family business, father of three struggling to pay his bills, just like everybody else. But when night falls, Dixon sheds his charming exterior in favor of a nefarious scheme that has everyone from the FBI to Customs to “an unnamed intelligence agency,” from the Air Force to “69 to 100 communities,” in utter disarray, shattered by Dixon’s masterful con game:

Prosecutors are asking a federal judge to send a “strong message” by sentencing an Indiana Little League coach to prison for trying to teach as many as 100 people across the country how to beat lie detector tests.

In a test case aimed at deterring other such polygraph instructors, prosecutors have urged the judge to sentence Chad Dixon to one year and nine months in prison, citing a “career of criminal deceit” that included teaching the techniques to child molesters, intelligence employees and law enforcement applicants.

In short, according to the Indiana prosecution, a blog post with this title is a potential threat to the very fabric of American society, that the only thing between us and general disarray is the hammer of justice itself, the polygraph – the “lie detector.”

Now, the real story beneath the hyperbole of this article is that Mr. Dixon’s extracurriculers viz-a-viz the polygraph are more interesting for his actively coaching criminal suspects to lie to investigators than for their general strategies for how to beat a lie detector. But the prosecution has stated its goal of discouraging people from teaching others techniques for beating the lie detector, and so the question is raised – what’s wrong with teaching people how to beat a useless investigatory tool?

The colloquial name for the polygraph machine, the “lie detector,” is a misnomer, since the polygraph does not detect lies or any other mental states. Rather, the polygraph reads a select few of the enormous number of physiological indicators of stress, and from this stress the prosecutor receiving the information from the polygraph examiner reading a select few indicators of stress infers that the test-taker is either lying or telling the truth. The extremely attenuated connection between a prosecutor’s interpretation of a polygraph reader’s interpretation of a polygraph machine’s interpretation of a polygraph taker’s stress levels forms the basis for all polygraph evidence, which is why most employers are banned from using it in pre-employment screening, five states have banned its use by employers entirely, and virtually every state at least restricts the ability of prosecutors to bring in polygraph evidence.

The most common limitation on polygraphs is to admit them by stipulation only – that is, only if both parties agree before the test is administered that, no matter the outcome, the outcome will be offered for the jury. This creates a certain gamble for either side – the prosecution might request a polygraph in order to try to “prove” that one party is lying, only to have that person “pass” the polygraph, thereby undermining the prosecutor’s case, and vice versa with the defense. It also gives the party being tested ample time to follow famed Soviet spy Aldritch Ames‘ advice on how to beat a polygraph: “get a good night’s sleep,” eat hearty breakfast, and go into the test on minimal stress. Because how’s the stress-detector going to infer lies from nonexistent stress? (An interesting consequence of the ability to beat the polygraph with low stress is that, at least in theory, sociopaths should be immune to the polygraph.)

In contrast to the states, where local laws often draw bright lines over how and when the polygraph is admissible, federal jurisprudence has created a complex weave of rules and exceptions over how and when either side can bring the polygraph into a federal case. At the axis of this confusing web is Rule 403, one of the Federal Rules of Evidence, forbidding the introduction of information whose value as evidence of a fact or crime (“probative value”) is outweighed by its tendency to confuse the jury or prejudice them against the defendant. Where questions of a scientific nature are concerned (and the question of whether or not physiological indicators of stress can be used to reliably infer deception most certainly is a scientific one), the test of probative value you need to pass is called the Daubert standard.

Prior to the Daubert standard (ironically created by a case overturning a different case that excluded polygraph testimony as unscientific), the reigning sole condition for whether or not expert scientific testimony was admissible was whether or not a given practice was accepted as scientific by the community that uses that practice. If you kind of squint and don’t think about it too hard, this old standard (the “Frye standard”) seems reasonable enough. Fingerprinting is likely thought to be scientific by forensic experts, right? But on the other hand, homeopathy is likely thought to be scientific by homeopaths, which is the community that uses homeopathy, and so on the old standard, evidence derived from the methods of homeopathy was theoretically admissible under the old standard.

The Daubert standard is more amorphous but I think significantly more scientifically meritorious than the Frye standard. It has several components, the most important of which are that scientific testimony must rely on methodology derivable from the scientific method, and that “the scientific method” is determined (non-exclusively) by such factors as falsifiability, the success of the evidence under peer review, a known error rate, acceptable standards for the control of such practices, and the acceptance of the scientific community of that methodology (formerly, this was the sole relevant criterion). So how does the polygraph hold up?

Sadly, the Supreme Court chose to duck the application of Daubert to the polygraph, declaring instead that local rules control and that the polygraph is admissible sometimes and leave it to the circuit courts to make that call, but at least five of the circuits have held that polygraph evidence is almost universally excludable, especially when the party importing it to trial is not doing so under the stipulation method discussed above. This is perfectly consistent with the known ambiguous-at-best statistical evidence for the actual reliability of the polygraph. More general objections – such as that stress does not always mean lying, especially in stressful situations like police interrogations, or that there are Constitutional questions as to whether or not a criminal defendant, or even his lawyer, or even trial judges (who are left as gatekeepers of scientific evidence by the Daubert standard), generally have the scientific literacy necessary to understand the problems with polygraph evidence or that the context in which it is being used varies its results – hardly seem necessary when the hard data itself tells us that the polygraph is unreliable.

That is why Rule 403 is the nucleus of arguments that tend to exclude polygraph evidence – because Rule 403 knows that, even if the other side has all the time in the world to explain the scientific shortcomings of polygraph evidence, the jury always stands to be confused or prejudiced against people who “fail” the “lie detector.” There’s a certain Cold War charm, an association with the gadgetry of espionage, connected to the polygraph. Television shows often depict its results as a smoking gun, with the polygraph working like a seismometer for deception. The mere transformation of “stress-detecting polygraph” into “lie detector,” bypassing miles and miles of missed inference between them, is itself evidence that culture attaches far more weight to these devices than they deserve and, as such, they tend to be more prejudicial than probative.

As to Mr. Dixon’s classes, they hardly seem worth the thousand dollars a pop that he apparently was charging his clientele (many of whom he apparently advised not just to “beat the lie detectors” but to “lie to the detectives”) since there are so many legal avenues to avoiding it entirely. If being a sociopath or a well-trained Soviet spy or a yogi isn’t your thing, you can always simply refuse to stipulate to one, or craft a clever Daubert motion to exclude it.