Thursday, June 7, 2007

Ay Calor! Los Panatlones, Como Queman!

The idea of a testable and universal biological manifestation of lying seems plausible enough, although my understanding is that all the current leading candidates -- polygraphs, microexpressions, pupil dilation -- remain unproven (to say the least). And I am damn sure that free-style lie detection -- observation of demeanor, eye contact, gut feelings -- is an unmitigated failure. This fact functions as a damning indictment of the American jury system, or more specifically, the tradition of discretion given to finders of fact on the ground that they can observe the witnesses and need not rely on a "cold record."

How good are we humans at using demeanor evidence to detect truth-telling or lying? A considerable amount of research has been conducted in recent decades on non-verbal behavior and the detection of deception. The findings indicate that demeanor cues often reduce accuracy in detecting deception, by distracting people into looking at cues they think are associated with lying and overlooking cues that actually are. Table 1 summarizes the findings of a meta-analysis of a number of experiments. Observers receiving no information at all would have standard deviation scores of 0.00 (sheer guesswork). Where no speech is provided, but only non-verbal cues, observers do no better when they see the face and nothing else (0.05), noticeably better when they can see the body and not the face (0.43), and when they can see both together, their performance falls a bit (0.35). Apparently, facial cues provide little help and sometimes do more harm than good. By contrast, subjects given transcripts alone are better at detecting deception than any of the conditions we have considered thus far (0.70). Speech sounds alone, with no visual cues at all, raise performance further (1.09). So much for the notion that non-verbal channels of communication carry more information than verbal channels, for human deception detectors to rely upon. Adding body cues to speech raises performance to its height (1.49). Adding facial cues to speech-with or without body cues-drags performance below what it was with speech alone.

But even if it were possible to find some universal phsyical manifestation of deception, the speaker's ability to know that he or she is actually engaged in deceptive conduct will remain a major limitation. As Richard Wiseman tells us in the Guardian's article -- the title link of this post -- lying is ubiquitous. It is simply not possible to imagine that if people are telling 14 lies a week, most of them are not spontaneously told -- in most cases, there is then probably time only to compare the lie to the truth and note its passable similarity. Accordingly, there is probably not time, in most cases, for the liar to have formed a firm opinion about whether or not it is truely a lie. If whatever bodily manifestation of lying eventually emerges fails to sweep in this huge gray area, situations where the liar is unsure whether to disbelieve himself, it will be next to useless. If it does sweep this in, it will be dangerously inaccurate.

Hat tip to slickdpdx at

No comments: