Who can we believe? The recent debacle of the discredited accuser in the Dominique Strauss-Kahn case has put the issue of credibility front and center of our national (and international) discourse. There is a science of credibility. It is part of the research on the psychology of deceiving and detecting deceit, which was my primarily scholarly preoccupation before I pointed my passions to the study of singles.
A remarkable series of studies was just published in the most recent (July) issue of Human Communication Research. The studies, by Tim Levine and his colleagues, document the power of one particular factor - a person's looks (demeanor) - on other people's judgments of whether or not that person is lying.
There are people who just look honest and trustworthy - both when they really are communicating honestly and deserve to be believed, and when they are lying. Then there are the more unfortunate types who just chronically look dishonest - both when they are lying and deserve to be disbelieved, and when they are telling the truth. That's what I mean by "looks" in this post. The authors call it "demeanor."
Maybe you have heard the claim that there are some people who are superb lie-detectors. The TV show Lie to Me features a lead character based on the scholar, Paul Ekman, who makes that claim most insistently. Not everyone accepts that claim uncritically, as Charles Bond and I detailed in this collection. (A few other relevant books are here.)
To be a great lie-detector, it is not enough to know a lie when you see (or hear or read) one. You also need to recognize a truthful statement. Skillful human lie-detectors can distinguish truths from lies.
Skillful lie-detectors should not be fooled by a person's looks. They should be able to see beyond a person's characteristic angelic or devilish look to discern when each person really is lying and whey they are really telling the truth. But can people do this? Suppose people without any special training or experience in the detection of deception get fooled by a person's looks. Will the same be true of people who work at national security-related jobs and who have had formal training?
When a person's characteristic look matches their actual truthfulness, the task of figuring out whether they are lying should be a (relatively) easy one. Sincere-looking people telling the truth represent one kind of match: both their look and their actual truthfulness suggest that they are telling the truth. The other kind of match occurs when insincere-looking people are lying. If you go by their characteristic look, you will think they are lying, and if you can recognize cues to deception, you will also know that they are lying.
The more challenging cases are the mismatches: sincere-looking people who are lying, and insincere-looking people who are telling the truth.
Levine had a collection of videotapes from more than 100 American college students who had an opportunity to cheat while playing a trivia game. Those who did well at the game could win a cash prize, so it was tempting to cheat. Plus, there was a fellow student egging them on (a confederate, actually, but the students didn't know that). Some took the bait and cheated - then lied about it when interviewed later. Others resisted the temptation and told the truth when interviewed.
The videotapes of all of the lying cheaters, and a comparable group of non-cheating truth-tellers, were judged by 64 undergraduates, who guessed whether each person on the tape was lying or telling the truth. From those judgments, Levine could determine who was most often believed and who was most often disbelieved. From all of the videotapes, he chose 20 to use in the key studies I'll describe below.
- 5 truth-tellers who were very often judged to be telling the truth (they were sincere-looking)
- 5 liars who were very often judged to be lying (they were insincere-looking)
- 5 truth-tellers who were very often judged to be lying (they were insincere-looking)
- 5 liars who were very often judged to be telling the truth (they were sincere-looking)
The researchers showed the 20 video clips to five different groups: two groups of American college students, one group of Korean college students, a group of university professors (perhaps they have experience with students who may have cheated), and a group of government agents from a security and intelligence agency. The agents had a range of experience at the job, and they all had training in deception-detection.
In the first column of the table below are the accuracy scores of all 5 groups at the easy task: judging the truthfulness of the people whose "look" matched their actual truthfulness. Everyone did well at that. Even the Korean college students, who were judging English-speakers even though English was not their native language, got 71% of the matched examples right. And look at those government agents - when a person's "look" and their actual truthfulness pointed in the same direction, the agents' accuracy at guessing their actual truthfulness was 96%!
% Accuracy (averaged across truths and lies)
Match Mismatch Who made the judgments?
79 36 U.S. college students
78 41 More U.S. college students
71 34 Korean college students
78 41 University professors
96 34 US government agents
But now look at the second column (above). When faced with the challenging task of judging mismatches, every group did poorly. In fact, they did worse than chance! Because half of the people on the tapes were lying and half were telling the truth, a person could get a score of 50% just by guessing. Instead, people in all five groups were more likely to be influenced by a person's "look" than by their actual truthfulness, when those two elements clashed.
I've saved the most powerful results for last. It turned out that about half of the agents had fewer than eight years of experience, and the others had more than 14 years of on-the-job experience. That's a big difference. So was it the less experienced agents who were pulling down the scores of the entire group?
The next table (below) shows the results of the less experienced and the more experienced agents. First let's look at the matches. The more experienced intelligence and security agents literally cannot be beat. If they are judging the truthfulness of sincere-looking people who really are telling the truth, or insincere-looking people who are lying, they are correct every single time. The less experienced agents are not far behind.
Now look at their accuracy at judging the truthfulness of the mismatches. Across all of the mismatches, both groups of agents are terrible - the less experienced get only 37% correct, and the more experienced agents do even worse, at 20%.
Finally, look at the very last row. The people on the tapes really are lying, even though they are characteristically sincere-looking people. Accurate judges of deception would see beyond their angelic looks and call them liars. The most experienced government agents did the worst at this task. They were correct only 14% of the time. That means that 86% of the time, they were fooled by the liars who have a sincere manner about them, and judged those liars to be telling the truth.
Less More What are they judging?
95 100 All matches
93 100 Telling truth, sincere-looking
97 100 Lying, insincere demeanor
37 20 All mismatches
54 36 Telling truth, insincere-looking
30 14 Lying, sincere-looking