Scientifically, it’s a pipe dream. It would be convenient if we could train law enforcement officers on how to tell if people are lying, but so far there is no way to do so. That doesn’t stop people from trying, though.
We’ve discussed lie detector tests before. In theory, polygraphs measure changes in breathing, blood pressure and skin conductivity, tracking allegedly tell-tale signs of lying. In reality, polygraphs may be no more accurate than chance, especially depending on the operator. Worse, they are prone to false positives, making it seem that people are lying when they are not.
There’s a relatively new tool out there that promises to allow law enforcement to spot liars. It’s called “Scientific Content Analysis,” or SCAN. Investigators around the country, including some with the FBI and CIA, have bought in.
The trouble is, there is absolutely no scientific evidence that SCAN works. Evidence from SCAN is not admissible in court because it doesn’t meet the low threshold for scientific evidence. But police use SCAN anyway, often to determine if people they interview are lying. And that can affect the entire investigation.
In a 2016 study in the journal Frontiers in Psychology, four researchers examined SCAN and the claims of its manufacturer, the Laboratory for Scientific Interrogation. They flatly found “no empirical support” for the technique. Indeed, much of SCAN is directly counter-intuitive.
How does SCAN allegedly work?
According to the manufacturer’s website, using SCAN is easy. Simply give the subject writing materials, ask them to write down their version of what happened, and then analyze their statement using what the company claims are scientifically supported techniques. (The website offers no citations to any research.)
Some of the techniques are odd. For example, when the subject uses pronouns instead of skipping them, uses the past tense or directly denies involvement, this is seen as evidence of truthfulness. Signs of untruthfulness supposedly include making corrections or using synonyms, such as saying “kids” at first and “children” later on.
Others are outlandish. For example, the creator of SCAN believes that, whenever someone describes opening or closing a door, this means they were probably a victim of childhood sexual abuse. Or, when a woman refers to herself as an “individual” or a “person,” it indicates a likely problem with sexual identity.
Again, there is no reason to believe SCAN can identify liars at a greater rate than chance. It seems surprising that anyone would believe that it could.
Yet law enforcement agencies around the country and world are spending thousands of dollars to have detectives trained in SCAN analysis.
How many false leads have been driven by SCAN? How many truthful reports have been ignored?