How to Detect a Liar

Posted in: Criminal Procedure

A highly admired evidence scholar and colleague in the academy told me something surprising a few months ago. He said that jurors have an easier time distinguishing truth from falsehood when they read a transcript of testimony than when they listen to and watch the testimony directly. I have not yet identified the studies to which he referred, but I want to offer a theory of why such findings, though perhaps at first surprising, make sense.

Why Surprising?

Ordinarily, having more data is better than having less. A transcript provides much less information than live testimony does, bare words compared to sounds, facial expressions, pauses, uncomfortable laughter, and tears. The richness of a live performance is why people are willing to pay much more money to watch a play than they are to read a page or an iPad screen. Don’t mock jurors miss a lot when they simply read a transcript?

Another intuitive reason to favor live testimony over naked transcript is that we think we can feel when someone lies to us. We look into their eyes and take their measure. We cannot work our magic if we neither hear nor see the emotion in their voices and on their faces.

But what if the following were true? What if voices and faces gave you additional information only in the sense that ambient noise in a study room gives you additional information when you are trying to solve a math puzzle? Noise can make it harder for you to concentrate on the issue at hand. And if face and voice are like noise, then they will draw our attention away from what we should be thinking about. Unlike noise, moreover, we may well imagine that we are attending to exactly the right details when we are in fact distracted from those very details by the appearance of relevance.

Humans may generally be poor at pure lie detection, i.e., at knowing whether someone is lying by listening to how they say what they say or by watching them speak. What humans are better at doing is figuring out whether a story makes sense. You hear that someone learned that his rain boots would arrive in the mail tomorrow instead of today, and he immediately set fire to his house. That sounds crazy. Who would do that? It would have to be something other than an ordinary case of disappointment that would motivate such behavior. We can evaluate narratives and, with focus, conclude that one or another makes no sense, perhaps because the ordinary reaction to action-X is action-Y, and one of the parties carried out action-Z instead.

If we are able to concentrate on each witness’s narrative, we can figure out, at least in a number of cases, who is telling the truth and who is lying. Some stories simply are not plausible. And we can tell which stories they are at times, even before we can specifically identify why. People experiencing this sense of dissonance might say that “something smells fishy,” an indicator of suspicion without an immediate sense of what the problem is. The question, then, is why seeing and hearing a witness’s testimony would interfere with this ability to pick up on, and then identify the source of, deception by the witness.


When people lie to one another, they often try to distract the target with non-answers to questions calling for incriminating information. “Did you have an affair with X?” gets the response “I barely know X.” “Did you read my private messages?” gets “I’m insulted; do I look like the sort of person who would invade your privacy?” Anyone who has had the experience of repeatedly being lied to becomes keenly aware of the gap between answers that sound responsive but aren’t and answers that truly respond to questions. And though some inveterate liars will stick to the lie, even when it must go from implied (“I barely know X”) to express (“I did not have sexual relations with that woman”), some will fess up once they must do so in order to avoid explicitly lying. In that case, it will be important for the audience to notice that the answer is evasive.

On the page, distractors (like errors) are easier to spot. Anyone who has read a transcript of testimony or a meeting can pick up mistakes that seem absurd, given the intelligence of the speaker or speakers. But listening to the same testimony, one could miss the errors. There’s so much else to pay attention to when a witness testifies. I would surmise that what is true for errors is true for lies. We’re so busy looking at whether the person makes eye contact, seems calm, and speaks at the right pace that we miss the fact that his story makes no sense and must therefore be false.

Liars can distract with more than their truthful-witness-like behavior. They can also tell the jury an interesting and suspenseful story, and they can thereby or otherwise make the jury like them. I remember being on a grand jury years ago where the defendant testified (a rare event), and his story was ridiculous. Looking at it on a bare piece of paper, no one would have believed it. Yet he was attractive and charming, and all of the jurors (including me, I am embarrassed to admit) voted not to indict, the only such outcome in thirty days of grand jury service in which we reviewed about five cases a day. A witness can seduce his jury. If the jurors like looking at him and listening as he tells his story, however implausible it may be, they will want to accept what they hear. Perhaps they will truly believe it. And perhaps they will simply stop caring about the truth, because they want to be allied with that witness, whatever it takes.

Federal Rule of Evidence 403 permits the exclusion of evidence whose probative value is substantially outweighed by the risk of unfair prejudice and other harm. We ought perhaps to think of the face, the voice, and the gestures of witnesses as falling into this category of excludable material. If the research truly is out there, then we might reach better, more accurate results if testimony took place outside the presence of the jury and if jurors read a transcript of the trial. Less dramatic, to be sure, but perhaps more accurate in the end.

Comments are closed.