Do polygraphs (lie-detectors) work?

Question

Polygraphs are often used in fiction, but also sometimes in court or by government agencies and corporations. Physiological reactions from the subject are recorded and an operator interprets them to decide whether the subject is telling the truth or is lying.

The basic premise sounds plausible at first glance, but how reliable are polygraphs really? Are there scientific studies on the accuracy of lie detection using polygraphs?

How do they compare to experiences interrogators using just their observations, without recorded physiological data, to judge if the suspect is truthful?

Answer 1

Nope, polygraphs cannot tell whether you're lying. ^{1, 2}

The polygraphs tries to measure nervosity but registering sweating and, I think, light trembling. They always calibrate the polygraph for each individual by asking questions the subject has no reason to lie about, to generate what is called a honest baseline.

The concept has some damning flaws, as no such physiological traits have been shown to consistently correlate with lying. Not all people get nervous about lying! And quite a lot of people would be nervous in an interrogation even when they were telling the truth.

There are actually a few signs that have been seen to correlate with lying, such as going at great lengths to avoid self reference ("then this happend", rather than "then I did this"), but nothing of the kind that could be picked up by a polygraph.

Richard Wiseman has this to say on lie detection through body language (from the book 59 Seconds):

For successful lie detection, jettison the behavioural myths surrounding the Anxiety Hypothesis, and look for signs more commonly associated with having to think hard. Forget the idea that liars have sweaty palms, fidget and avoid eye contact. Instead, look for a person suddenly becoming more static and cutting down on their gestures. Also, learn to listen. Be on guard for a sudden decrease in detail, an increase in pauses and hesitations and a sudden avoidance of the words 'me', 'mine' and 'I', but an increase in 'her' and 'him'. If someone suddenly becomes very evasive, press for a straight answer.

Answer 2

This answer is really a follow-up to David Hedlund's (correct) answer and the resulting conversation in the comments between David and Sklivvz.

A large portion of a polygraph's effectness comes from the fact that the subject believes that it works. In some sense, it's a very fancy prop/placebo. A polygraph test (in an NSA / DoD setting) goes down something like this:

-- The questioner asks you a bunch of questions. Due to the general stress of the situation and the sensitive nature of the questions, you are generally nervous (sweaty palms, etc).

-- After going through all the questions, the operator leaves the room for a bit to "analyze the results". He then returns and accuses you of lying (or omitting details, or fudging the truth). He then gives you a chance to rectify your answers.

-- Figuring that things will go better if you come clean now, you try to guess what triggered the polygraph. Since the operator gave no indication as what the problem was, you basically spill your guts in a scattershot approach.

-- Rinse, wash, repeat.

Obviously, the operator needs to be a skilled cold-reader. Also, a trained liar will know that the polygraph isn't actually doing anything, and so step 3 will fail.

So do they work? In a technical / scientific sense, no. In a pragmatic sense, sure.

Answer 3

Lie detectors measure stress. They are effectively useless in detecting lies. Some people can lie without thinking about it, and some get nervous when they're trying to tell the exact truth.

What they are useful for is finding sensitive subjects. A polygraph expert can ask questions and note what subjects seem to be stressful. This can allow the expert to narrow in on promising lines of inquiry, which is a lot more efficient than asking all sorts of questions.

Answer 4

It all depends on how you define "work". Do breathalyzers work? Do pregnancy tests?

Polygraphs tests work with a reasonable accuracy for some topics, but not for others. They are not accurate enough for legal purposes, but they are accurate enough if you are trying to confirm other data and use them as support in the confirmation (non-legal).