Tuesday, February 21, 2012

How to be a better liar

Rice by babbagecabbage, Photoshoppery by yours truly.

Listen up, liars. Society is onto you. Well, its machines are anyway. Computers can spot your fake online hotel reviews, and they know when you’re shaving ten pounds off your weight on your Match.com profile. They’re sharpening their skills daily and, with their help, even the human brain (a device notoriously terrible at truth detection tasks) might get wise to your chicanery. So if you’re not going to change your wicked ways, you can at least try to improve your technique a bit.

But how? Researchers say the telltale signs of deception are difficult to hide. Even in written form – where the author has the opportunity to edit his or her appalling untruths – liars still leave behind linguistic clues. But I say, what kind of defeatist attitude is that? We can make our résumés more appealing to employers by replacing passive phrasing with exciting ACTION words. Why should lying about the rest of our lives be any different? Before we throw in the towel and resort to honesty, shouldn’t we at least try to apply what science has learned to our own deceitful endeavors? Of course we should. Let’s have a go at it.

Mind your language
Certain outward signs can draw attention to a poorly executed lie: fidgeting, stammering, sweating, shaking, heartbeats audible from two rooms away. Humans took note of this and have concocted lie-detecting technologies that attempt to exploit such physiological cues. The earliest example usually given is China, circa 1000 B.C., where suspected criminals were asked to place rice powder (or rice, by some accounts) in their mouths and then spit it out during interrogations. The idea behind this test was that the stress of lying dried out one’s mouth. Thus those answering honestly should be able to spit out more than liars, whose mouths the starchy stuff would stick to like feathers on tar.

The modern polygraph, which measures pulse, blood pressure, and other indicators while suspects answer questions of varying stakes (“Is today Tuesday?” vs. “Did you kill your wife?”), is basically a twentieth century upgrade of the rice test.* Both rely on finding signs of nervousness exhibited by guilty liars. Problematically, both also risk merely capturing the anxiety of innocent people freaking out over being accused of a crime.

A more novel way of discerning between factual and fabricated statement involves not measuring the body, but analyzing the words used by the speaker (or writer). While not initially conceived as lie detectors, computer algorithms that examine linguistic patterns have been used experimentally to search for hallmarks of deception. Linguistic Inquiry and Word Count (LIWC), software designed by James Pennebaker, Roger Booth, and Martha Francis is one such tool. It sifts through documents (written text or transcribed speech) and tallies the instance of various word categories, including significant but harder to control “function words” - pronouns, articles, and the like.

LIWC was most recently unloosed in the jungle of online dating and proved superior to human judges at assessing “trustworthiness” from the content of daters’ profiles.

Thinking cues and feeling cues
Having spent some time tackling this idea of liars’ linguistic cues, scientists have come up with a few observations. The giveaways can be divided into two types: cognitive (thinking) and emotional (feeling). These distinctions are based on the proposed causes of the cues. The idea is that lying is 1) morally troubling and 2) intellectually difficult. Let’s tackle the second one first.

Because liars have the challenge of creating and managing the details of a fictional tale, their accounts should theoretically reflect this by being shorter and simpler. Experimental analysis has found that false statements generally do have lower overall word counts. They also contain fewer exclusion words. These are words like “except” and “but”, which are used to make the kind of fine distinctions in stories that can be a headache to keep track of when they’re not actually true. For instance, when playing hooky from work, the cognitively-taxed liar might claim, “I have a cold.” or even, “I have a cough and a sore throat.” But you can improve upon this by throwing in an exclusion, “I have a cough and a sore throat, but I’m not feeling feverish.” Just make sure you have more symptoms than exclusions, else your employers will think you’re being lazy.

Due to ease of handling, motion verbs are also common in dishonest accounts. “I fell and sprained my ankle; I won’t be coming in today.” “The car ran over a nail and got a flat; I won’t be coming in today.” And so on. You may not want to lean too heavily on motion verbs. Try adding a little detail about what it was that caused you to fall in the first place, or the crummy pot-hole-strewn road you driving on. Be careful though. As a child, my instinct when lying was to construct elaborate narratives with well-developed characters and lengthy passages of dialogue designed to answer questions no reasonable person would think to ask. It was perhaps overkill, as I can report anecdotally that my results were not stellar.

Emotional cues stem from feelings of guilt and fear related to lying, rather than its cognitive demands. These are especially important in our modern computer-driven world, because they’re harder for the liar to filter out, even when given the chance to edit their work (the online dating profiles found more emotional than cognitive giveaways of deceit). Unhappy words and negations both crop up more frequently in the ramblings of the untruthful. Though you might have a hard time eliminating these entirely when you’re lying about a dour subject. I mean, is it better to say, “I feel lousy” or “ I don’t feel well”?

Perhaps we should focus instead on paring down instances of “distancing”. Apparently, liars want to detach as much as possible from their ghastly falsehoods and often do so by avoiding first person pronouns. So make sure to use, “I” and “me” whenever possible.

Of course, as with physiological measures, none of the above pitfalls are likely to affect an evil genius (clinical definition: high functioning antisocial personality with IQ of 140 or greater). But, given that you’re sitting around reading this rather than out devising a kryptonite trap for Superman, you’re probably not in that category.

Man vs. machine
Something else you might want to keep in mind in all this wordsmithing, is who or what you are trying to dupe. So far, everything we’ve discussed relates to outwitting computer algorithms. But humans and machines rely on different cues to decide whether or not to trust you. And even though humans suck at spotting a lie (we tend to fare at about the rate of chance), that shouldn’t stop you from peppering your fictions with the very elements we incorrectly perceive as signs of honesty.

The only linguistic trait that both evokes distrust in humans and actually correlates to lying is word count. Shorter stories are more likely to be false and more likely to be perceived as false by a human audience. † But from that point on, our poor species gets lost and gravitates toward linguistic patterns that have nothing to do with honesty. While we like long descriptions with plenty of details, we also want individual sentences to be on the short side. Got it? Use lots of words. Make short sentences.

Also popular is concrete language. Abstract or convoluted sentences inspire suspicion (not to mention boredom). And do try to use the word “we”. The plural first person pronoun “we” makes listeners feel included, whereas the second person “you” or third person “they” makes us feel like outsiders. To test this, I made a point of using “we” in describing human lie-catching ineptitude, so that you wouldn’t think me a snob accusing you of being a bumbling simpleton. Did it help? Were you filled with trust? Perhaps an urge to lend me money?

And Now the Caveats
WARNING: Don’t try this in a language other than English. All the studies I looked at were conducted in English and, as anyone who has struggled to learn a new language knows, grammar varies considerably between languages. Culture likely has an effect too. Who knows if first person pronouns have the same appeal outside of the egocentric U.S.

Also, results may vary with the degree of the lie being told. A lie created to conceal a major transgression will likely leave more clues in its wake than one told to praise an unremarkable meal.

And while we’re on the subject, those big lies about where we were and who did or didn’t kill aren’t the most common type of dishonesty. The majority of our deceptions work to mask our socially unacceptable opinions and feelings, or to hide our perceived shortcomings. Given the banality of our lies and the effort required to tell them convincingly, it might be easiest to just fess up to those unpopular attitudes and lackluster achievements.

* Bonus trivia: The man who invented the blood pressure measuring component of the lie detector, William Moulton Marston, was also the creator of Lasso-of-Truth-wielding comic book heroine Wonder Woman (under the name Charles Moulton).

† This is not to say that any short sentence is false. It’s all relative. And, relative to a truthful account on a similar topic, an untruthful one is likely to be shorter.