You can hide those lying eyes

Near the end of the second season of the TV show Prison Break (Episode 15: “The Message”), there was a bit where two escapees release a taped confession.

The authorities start examining the tape, paying particular attention to the way the eyes of the people in the tape are moving. When Lincoln Burrows says, “I was sentenced to death for a crime I did not commit. I did not murder Terrance Stedman,” he glances up and left. There,” one investigators says. “The eyes. See them?” Another says a moment later, with some confidence, “He’s lying.”

Oh, television, you led me wrong again.

The notion that eye movements are particularly revealing about people’s thought processes appears to trace back to neurolinguistic programming. This is a discipline that I’ve heard other people refer to occasionally, and I got the impression it had enthusiastic devotees outside the scientific community and effectively no support within the scientific community.

A new paper by Wiseman and colleagues indicates that my impression was not far off the mark. The paper takes aim at the notion that you can infer anything about whether someone is telling the truth by watching their eye movements.

You can’t.

The new paper has three experiments, but the first one is key.

People in the experiment were told either to tell the truth or lie up front. They went into an office came out, and were filmed asking three questions about stuff in the desk drawer of the office they were in. Then, the video recordings were watched and coded for eye movements. The eye movements were not correlated with whether the person told the truth or not.

Are we done here? Well, not quite. The second experiment looked at real time judgments, rather than video analysis, as neurolinguistic programming says lie detection can be done “live.” Unsurprisingly, doing it on the fly is not more accurate than detailed frame-by-frame analysis.

The third experiment tried to get away from the lab setting. After all, one might argue that the effect might not be seen in a relatively safe setting of a psychology experiment. Wiseman and colleagues used videotapes compiled by other researchers of people who were making public please for return of a loved one who had been abducted. In half the tapes, there was good evidence that the person making the plea had committed the crime him- or herself. In other words, this is a massive lie with real consequences, not just an innocent fib.

Still nothing.

This was not a huge surprise to me. I did a decent amount of reading on research about detecting deception for an ethics paper I wrote. There’s a lot of interest in detecting lies reliably, and no technique is even close.

Oh, and in Prison Break? The characters were coached by someone who had done the “lie detection” before, so were able to give signals the exact opposite of what was expected, and fool the authorities into thinking that they were lying, when they weren’t.

Still no word on whether a smile is a thin disguise, though.

Reference

Wiseman R, Watt C, ten Brinke L, Porter S, Couper S-L, Rankin C. 2012. The Eyes Don’t Have It: Lie Detection and Neuro-Linguistic Programming. PLoS ONE 7(7): e40259. DOI: 10.1371/journal.pone.0040259.t003

External links

Neurobonkers take on the paper