The American Indians did it by placing the suspect's finger in a bowl of water and watching the ripples. The Chinese preferred to put a few grains of rice in the mouth to see how damp they became. In medieval England, we would make suspects walk over red-hot ploughshares and see whether their burns healed; or we would tie them up and throw them into the water to see whether they floated or sank (the innocent sank).
People have always wanted to be able to detect when someone is telling lies. Nowadays, the legal system does the job by gathering and evaluating evidence. But we still like to judge a person's honesty by looking at their appearance and behaviour; their demeanour, in other words. And business life is no exception.
Managers want to identify lying by their clients, customers and colleagues. It helps, too, in the recruitment process: a survey last year showed that a quarter of CVs in the financial services industry contain irregularities, even those from candidates who had accepted job offers and knew their applications were going to be reviewed by experts.
The problem is that detecting untruth is harder than it looks. Research shows that most people are so bad at it they might just as well flip a coin. That may be because trivial lies are a necessary part of normal social life. ‘How are you?' ‘Fine, thank you.' That's two lies: the questioner doesn't really want to know, and the respondent doesn't really want to say.
Given all that, it is no surprise that people have sought technological help in detecting lies. But the original lie detector, the polygraph, is easily tricked; some say you should put a drawing pin in your shoe and press it to mask your real reactions. The polygraph is also extremely intrusive and impractical.
But who wouldn't like to be able to detect dishonesty over the telephone? That's the claim of voice analysis. Invented in the 1960s, it detects changes in vocal quality caused by the body's ‘fight or flight' response to a threat. An Israeli company, the delightfully named Nemesysco, has licensed its system around the world. In Britain, DigiLog has successfully marketed it to insurance companies, including esure, Halifax, Provident and Highway.
DigiLog insists that it sells a ‘solution', not just a piece of kit. Call-centre operators learn to conduct structured interviews as well as operating the equipment. A baseline is established in opening chit-chat and then, when the serious questions start, the frequency patterns of the voice are compared with that baseline to discover minute changes in pitch. According to Lior Koskas, DigiLog's director of business development: ‘You train the operator to recognise certain trigger words, mark them as a signpost for deception and then look for a correlation between what you spotted and what the technology spotted. Once you have a match, it's likely to be a problematical area.'
Koskas claims the software identifies ‘a cognitive conflict between what your mouth is saying and what your brain knows to be the truth'. It will find lies, but it is better (‘100 per cent,' says Koskas) at identifying truth. Insurers use it to separate out genuine claims and fast-track them. Suspect applications are dealt with by traditional means.
The system costs between £3,500 and £10,000 a year per seat, depending on how many units are deployed, but Koskas makes much of its cost-effectiveness. ‘We always say: if you can't show 10 times return on investment, we will recommend you not to do it.'
He says one client, Highway Insurance, has saved £5 million in three years. Where it formerly found frauds in 5% of its claims, it now finds them in nearly 20%.
Halifax, DigiLog's biggest client, with some 28 installations, says it finds the system 60% more efficient than traditional validation. All its customers are informed about the system. Even so, more than a fifth of those screened go on to withdraw their claims. Sceptics, however, note that when faced with the traditional polygraph and told it was infallible, criminals often confessed. The mystique of the machine may be as powerful as its technology. Direct Line, meanwhile, says bluntly that it tried the system and ‘it was not found to be useful'.
DigiLog, though, is ambitious. Teamed up with Capita, it is targeting the public sector. It even suggests insurance companies might use it to weed out potential fraudsters at the application stage. Back in Israel, Nemesysco has also been expanding its technology's range, with a ‘Love Detector' kit. Point your mobile phone at your love-object and it will tell you whether you are wasting your time. None of this seems calculated to enhance the reputation of voice analysis as a business tool, but the principles behind it have achieved considerable respectability. All research in this area is contentious, but experiments have consistently identified a rise in vocal pitch as a mark of lying.
There is, though, a real problem for anyone trying to identify deception through demeanour. It was acknowledged by the doyen of deception studies, professor Paul Ekman of the University of California, in a 1997 paper: ‘There is no behavioural sign of lying itself.' Lying is an intellectual act, and no-one can read minds. Crude methods of lie detection, including the polygraph, simply detect stress and anxiety.
Most of us make the same mistake. We believe that liars avert their gaze, that they adopt uncomfortable postures and are ‘shifty', that they make a lot of hand movements and that they smile too much. We think they cover their mouths and eyes while lying. Those Native American and Chinese lie-detection techniques are also based on detecting stress. A shaky finger makes ripples in the water; a dry mouth leaves grains of rice unmoistened.
In the past, the police have used some of these ideas. ‘It's the holy grail,' admits detective chief inspector Gary Shaw, who is in charge of interviewing techniques at the national police college in Bramshill. He is sceptical, however: ‘What people think is an indication of lying is based on nothing,' he says. Today he teaches structured interviewing, based on detailed preparation, allowing interviewers to catch out their subjects by probing deeply into their stories. Old-fashioned detective work, in fact, in which looking into people's eyes plays no role.
The truth is that holding a steady gaze is the first skill an effective liar learns. Nonetheless, some persist in seeing the eyes as the window to the soul, looking hard for dilation of the pupils.
Another school, based on ‘neuro-linguistic programming', thinks you can detect lies from eye movements. Looking up and to the left means you are using your imagination; to the right, you are consulting your memory. It's all neurological nonsense, and completely unproven by experiment.
If you really want to detect lies through demeanour, there are three things to watch, according to professor Aldert Vrij of the university of Portsmouth. Liars must simulate or conceal true emotions; invent a false story; and control behaviour that might give the lie away. Dishonest thoughts are not detectable, but these actions may be.
Ekman has identified non-verbal signs of these mental processes, but they are not easily spotted, often appearing as fleeting ‘micro- expressions' that are best seen in slow-motion. If you are interested, however, he will sell you a computer program that will supposedly teach you the technique in an hour. It depends on the fact that genuine emotions are not easily imitated or hidden. Real smiles, for instance, use the muscles around the eyes as well as those around the mouth.
In real emotional outbursts, the facial expression comes first, then the words: in fake emotion, it's the other way round. Good actors can accurately mimic emotions, often by using ‘method' techniques to recall real personal events, but bad actors can't. You can, apparently, train yourself to spot these techniques by slowing down bad television: try Casualty.
The fact that liars are inventing, rather than remembering, may have visible consequences. They may take longer to respond, may hesitate or stumble: but that depends on how well they have rehearsed their story. Ekman's findings have not been widely duplicated, but he says that is because subjects in laboratory experiments have nothing to gain or lose by lying. For true emotions to ‘leak', the anxiety level has to be high, and that only happens when lying will cost you your job or your liberty.
Most psychologists, though, tend to discount non-verbal signs of lying. There is, they chorus, ‘no Pinocchio's nose'. But Pinocchio may give himself away by what he says.
Researchers who take this approach work from interview transcripts: they consider looking at their subjects to be seriously misleading. Some, like James Pennebaker, professor of psychology at the university of Texas Austin, use computer programs to examine the minutiae of grammar. He believes, for example, that liars use the word ‘I' less frequently, and insists that ‘liars can be reliably identified by their words - not by what they say but by how they say it'.
A senior lecturer in Vrij's department in Portsmouth, Lucy Akehurst, uses a technique called ‘statement validity analysis', which concentrates on content. She believes that true accounts of events have certain identifiable characteristics. In particular, true stories are detailed. Not only do they have more detail than invented stories, they include unexpected and unnecessary information.
‘We as psychologists tend to think in terms of scripts,' she explains. ‘If I asked you to pretend you were kidnapped yesterday evening, you'd be able to make up a story. You'd base it on your experience of life and what you'd seen on the telly and you'd use what we call a script. In your script you'd say, ‘They took me in a van, I didn't know where I was going, they put me in a room and they locked the door.' You'd very rarely say that they tried to lock the door and they couldn't find the key. It's things that ring true, if we can use that expression.'
This approach could hardly be more different from the old look-me-in-the-eye school, but Akehurst is anything but an ivory-tower researcher. She works with police in assessing the credibility of children in sex-abuse cases, and trains officers and social workers in truth- assessment techniques. Often, she admits, that means ‘dispelling some of the myths'.
Her advice to anyone trying to spot a liar is simple: ‘Close your eyes. Don't look at them. Listen to what they are saying. Non-verbal cues are very misleading and we use too many stereotypes that are not accurate.'
The naïve idea of ‘truthful' behaviour is dangerous. Individuals, and people from different cultures, have different modes of non-verbal communication. Introverts and the socially awkward look like they're lying, even when they're not. ‘And people behave differently in different situations,' says Akehurst. ‘The same person, if they were lying to their boss, might behave non-verbally in a different way to when they were lying to their partner.' Eyes can indeed deceive us - our own eyes.
But while statement validity analysis works with children, it may be less effective with grown-up liars. ‘It's like the polygraph,' says Paul Seager, senior lecturer in psychology at the university of Central Lancashire. ‘If you know the premise it's based on, I can teach you to beat it. And it's all out in the public domain… You build a bigger mousetrap and mother nature builds a bigger mouse.'
Seager's doctoral thesis examined whether it was possible to teach people to be better lie detectors. He concluded that it probably wasn't. There are, he says, plenty of people who think they are very good at it; but evidence suggests they are no better than anybody else.
Says Seager: ‘My advice would be, don't be over-confident in your ability. Give yourself a shot of reality before you try and detect someone's lies.'
Ten research-based tips for detecting liars from their demeanour (your results may vary)
1 Remind the subject that the stakes are high. If they have something serious to gain or lose, they are more likely to give away their true emotions.
2 Establish a baseline. Observe their demeanour when they are in a stressful situation but telling the truth, and watch for changes.
3 Look for lots of signs of lying, not just one or two.
4 Do not tell them what you know. You need them to struggle to invent a story that won't contradict what they think you know.
5 Listen for detail in their story. You want unexpected, irrelevant information and snippets of dialogue.
6 Listen to the pitch of their voice. Does it rise when answering certain questions?
7 Are they pausing to think before answering? They may be inventing, not remembering.
8 Do their smiles reach their eyes? True emotions affect the upper portion of the face, and those muscles are much harder to control.
9 Practise, and give yourself feedback on your successes and failures. Skill in lie-detection comes with experience.
10 Do not over-estimate your ability to catch liars. It's a lot harder than it looks.