Ask Me No Questions…

Lies are part of everyday life. Research published in the Journal of Basic and Applied Psychology indicates that a person in active conversation in a social group typically tells 2.92 lies every 10 minutes. We’re not talking about serious deceits or money-inspired cons – at least not very often. Overwhelmingly they are insignificant white lies intended to lubricate conversation or to avoid potential conflicts. They can be anything from false compliments (“your hair looks nice”) to false explanations about why one didn’t answer the phone earlier. The 2.92 number doesn’t even take account of “lies of omission,” i.e. intentionally giving a false impression by leaving out pertinent information. Most of us acknowledge this foible in ourselves, and might even debate the term “foible.” We still likely consider ourselves honest “in ways that matter.” Nonetheless, the taboo of dishonesty we learn as children (and at the same time learn selectively to violate) still can make us feel guilt when we are forced to face our prevarications.

An acquaintance of mine makes a sport of catching people out. If you slip in conversation by giving two contradictory statements of where you ate lunch yesterday, most of us (perhaps not a jealous spouse, but most of us) will just let that slide without mention; we know you just wanted to frame a restaurant recommendation or some such thing. Not this fellow. He’ll acquire a satisfied grin and say, “But I thought you said”… etc. This is socially objectionable and explains why he seldom is invited to parties. There are people, however, who detect lies not just to be annoying but for a living. Actually, “lie detection” is a misnomer. No one who starts out uncertain of the facts can detect a lie. He or she, if competent, merely can detect emotional responses that might indicate a person is lying. Then again, they might indicate something else. People get emotional for lots of reasons. Sometimes the truth triggers emotions.

The man who literally wrote the book on lie detection is Paul Ekman. A professor emeritus at UC, Ekman consults with police, national security agencies, private companies, and legal specialists on such things as facial cues, voice inflections, and polygraphs. In his book Telling Lies he describes experiments which demonstrate that most people are terrible at detecting lies. They misread body language and voice patterns and rarely beat pure chance results on controlled tests. Police superficially appear to have a higher rate of success than the general population but only because their job breeds such cynicism that they often assume everyone is lying; you won’t miss the liars if you accuse everybody. If you include their false positives – people assumed to be lying who are telling the truth – police typically are not any more accurate than the general public. Contrary to the claims of some enthusiasts, polygraphs also are not much better than chance in the hands of the average polygraph operator. As Ekman repeatedly points out, the machines do not and cannot detect lies directly. They merely record emotional responses. They do not even specify which emotion the subject is feeling, just its intensity. It is easy to misinterpret the data.

The good news – or at least the useful news – is that by combining the polygraph with a serious study of expressions, micro-expressions, and questioning techniques, the success rate at lie detection can be boosted. An example of a useful questioning technique for a polygraph test would be as follows. Rather than ask a murder suspect what weapon he used, recite to him a list of possible weapons (assuming you know the answer yourself but that the public doesn’t know); a guilty suspect will almost surely have an emotional response when he hears the right weapon while an innocent person likely will have the same response to each weapon named. It is not always possible to frame questions this way, of course. The police do not always have sufficient knowledge to do so. But with training a minority (about a third) of trainees can achieve success rates as high as 80%.

Therein lies the promise and the rub. Even this elite cadre is wrong 20% of the time. Most so-called experts are wrong far more often than that. A 20% error rate sounds a lot like “reasonable doubt,” and the odds are against getting an expert as good as that. False positives, and they are common, can cause innocent people a world of trouble. Further, just as a person can be trained to detect lies better, a person can be trained to tell them better too, including to polygraph machines. Lawyers train their witnesses in a limited way all the time. Courtrooms are ideally designed to aid liars anyway. Cases come before a jury long after the event, so the emotional responses of the witnesses have dulled. Furthermore the type of questions is restricted and the witnesses are coached. It also has been demonstrated that some people are natural liars who regularly defeat interrogators and polygraphs; on the other hand, some other people can’t tell convincing lies to save their lives, and training helps them very little.

Ekman’s conclusion:

“…I do not believe that judgments about who is lying should be allowable evidence in court. Such judgments, however, may provide a sounder basis for deciding, at least initially, whom to investigate further…”

This seems to be a sound policy recommendation. He has similar reservations about polygraphs used by private companies for employment reasons. They simply are not reliable, especially in the hands of the typical operator.

While a skilled analyst’s 80% success rate may not be proof, it is worth noticing. Ekman’s techniques are useful and his books are intriguing reads. But, if you learn his methods, I urge against using them to confront people with their harmless fibs in social situations. The number of invitations you receive to parties will fall drastically.

Lies