Someone accused of a heinous crime they didn’t commit will likely be scared and confused, and after hours of intense questioning, will also be weary and sleep-deprived. Which means they may make poor decisions about whether to continue speaking or to have a lawyer present. Add to this mix the claim that evidence has been found again them and one can end up with the terrifying reality of a person admitting to something that they didn’t do and which will deprive them of their freedom and reputation. This is even more likely if they can be persuaded that they will be allowed to plead guilty to a lesser charge than what they are being accused of.

In an article for Debunking Denialism, Emil Karlsson wrote that authors of the police manual Criminal Interrogations and Confessions insist that if law enforcement officers ask certain questions of suspects and study their behavior and responses, they can make accurate determinations of guilt or innocence 85 percent of the time.

However, the study on which that assertion is based had no control group and no reliable way to determine the actual truth of the criminal cases where these techniques were used. Further, research shows that alleged signs of deception, such as nervousness and darting eyes, may not be that at all. Still, confirmation bias and subjective validation will make the percentage of successes seem greater.

The most common method of getting suspects to confess is through the Reid technique, which combines a hostile interrogation which assumes built and lying about evidence against the accused. In traditional good cop-bad cop fashion, there will eventually a more sympathetic ear offered to the accused as the interrogator tries to understand the reasoning behind the crime or to mitigate its circumstances. Then, contemplating the consequences of, say, being convicted of first-degree murder and pleading guilty to manslaughter, the accused may break down and make a false confession.

The introduction of manufactured evidence is crucial. Experiments have shown that false evidence used against the accused can double the number of persons who confess. In one such study, subjects filled out a computerized survey and were warned that if they hit the alt key, the machine would crash and the data be lost. If a subject were wrongly accused of doing this, half of them confessed to having done so. But when a purported eyewitness mendaciously claimed to have seen the alt key pressed, the confession rate rocketed to 94 percent.

And once a confession is made, the damage is usually irreversible. Karlsson wrote that studies utilizing mock jurors show that “confessions have an extraordinary high impact of decisions. Even when conclusively proven to be coerced, jurors are not able to discount their influence and thus cases where coerced confessions are presented and jurors are explicitly instructed to ignore it have a higher conviction rate than the same cases without a confession.” Even if the confession is false, proven to be coerced, and buoyed by no other evidence, the accused is much more likely to be convicted. If the defendant is painted by the prosecution or police as being unstable, that makes it even worse.

A lab study using an actual case demonstrated this. Researchers broke volunteers into four groups: A control group given the real story; a group given the real story but with a false confession thrown in; a group that got the real story but with an irrelevant testimony from police about the suspect’s emotional state; and a fourth group that heard both the false confession and irrelevant testimony.

The base rate conviction rate was 53 percent, a false confession increased that to 63 percent, while irrelevant testimony reduced it to 48 percent. But if hearing both the false confession and irrelevant testimony, mock jurors voted to convict nine times out of 10.