A recent study has overturned what we think we know about lying.  Most of us have a pet theory about how to tell if someone is dissembling to us.  We may develop that theory from observations of those people we know well and see regularly, but we tend to generalize from those ‘tells’ we’ve uncovered from that unscientific daily research and make them a universal theory.

So we might imagine that liars have twitchy, evasive eyes, or the opposite – they stare at you too fixedly.  Or we might believe that it’s a particular set of the jaw, or the lips, or eyebrows.  Or perhaps it’s more generally nervous behavior – fidgety hands and the like.  Or sitting too still.  Whatever the particular theory, it’s usually based on close observation of people we know.

And we do get lots of practice.  On average, we’re lied to 200 times per day.  Mostly harmless, kind lies (no, that doesn’t make you look fat) – but lies nonetheless.

Here’s the problem with our pet theories and observed behavior:  it’s usually wrong.

The average person – you and me – tested rigorously on how well we detect lies fails to do better than chance.  That’s well established over many studies and lots of attempts by people like the CIA, FBI, NSA, KGB, NGU, and so on to work out reliable ways to detect lies.  It’s even relatively easy to outwit lie detectors, the gold standard of lie detection, by training yourself in breathing techniques and symptom suppression.

Is there any way to improve on this sorry record of underachievement?  It would be nice to be able to bump up our accuracy rates to something better than chance.  It might improve our poker playing – and it might just depress us, knowing that we do look fat in that article of clothing after all.  Either way, the truth could set us free, and as such it would be good to know.

How do you get better at detecting lies?  Here’s where the new research offers some surprising advice:  stop looking.

Instead, listen to your potential fibber.  It turns out that if we can’t see the face, but rather focus on the voice of the person in question, our accuracy rate improves.

I love this research finding, because it points to something I’ve been writing about for a long time:  the importance of the voice.  Many public speakers who come to me for coaching know that nerves show up early and clearly in the voice.  You’re familiar with the phenomenon:  the shaky voice, the spasmodic swallowing, the dry mouth.  The adrenaline cycle produces some unpleasant and audible reactions.

That’s important, because earlier research has shown that the voice is where your authority shows up and unconsciously influences the audience to accept you as a leader – or not.  If your voice is shaky and subpar, then you’re not going to be able to become the authority that your role as public speaker requires you to be.

Speakers, tune up your voices.  And for everyone else, if you want to become a little better at detecting the lies around you, shut your eyes and listen.

Come find your voice at our Powerful Public Speaking workshop on March 31st, in Boston.  Sign up soon — only a few spaces are left.  



  1. I just gave a presentation to youth on finding their story to nail an interview. I asked the group about a challenge they overcame, and the wisdom they learned from it. A young man (maybe 18) took the mic and with a shaky voice he talked about being bullied on the playground in elementary school. His mouth was dry. He had to support himself against the podium. It could not have been more authentic.

    Certainly the body language (covering the mouth, heart or the stomach) could be a sign. Or, standing unevenly. Maybe darting unfocussed eyes. What other voice cues do you recommend we look for in someone who is lying? Very interesting topic!

    1. The best way to detect lying is to first of all establish a baseline of behavior with the person when you know they’re relaxed (and telling the truth). Once you know a person’s standard behavior, it’s much easier to detect variations, and thus possible lies.

Leave a Reply

Your email address will not be published.