Will science ever be able to detect the truth about lying?

Attempts at detecting someone telling lies are as old as humanity itself – one theory about the origins of human intelligence is that man's relatively large brain evolved out of the need to recognise deception in other individuals living within the same social group.

Finding out the truth from unwilling interviewees came centre stage this week from two unrelated spheres, highlighting the difficulty of detecting deception in skilled, well-trained inveterate liars.

The sole surviving gunman of the Mumbai attacks is undergoing interrogation in India that will include the injection of a so-called "truth serum" in the hope of eliciting information about his past and his associates that he would otherwise not volunteer.

Meanwhile, it emerged in the Queen's Speech on Tuesday that the Government is considering the introduction of lie-detector tests to expose untruthful benefit claimants.

Lie detectors, or polygraphs, do not in fact detect lies. They monitor the physiological changes to the body – such as heartbeat and skin conductivity, or sweating – that may, or may not, be associated with failing to tell the truth.

The idea behind the device is that there are involuntary actions that occur when someone experiences the stress of telling a lie, which can be detected by sophisticated machinery.

Polygraphs are widely used in the US but have been rejected in Britain because of their unreliability. Apart from whether they actually pick up on hidden signals of lying, their accuracy is predicated on the skill of the person interpreting the machine's signals.

In extreme situations, torture and fear are some of the oldest means of extracting the truth. But that has the disadvantage of leading people to say things under duress that they know the torturer wants to hear – as well as being a breach of human rights.

Another method is to ply someone with alcohol, a technique immortalised in the Latin phrase in vino veritas – in wine there is truth – which is ascribed to the Roman scholar Pliny the Elder. Josef Stalin is said to have feigned getting drunk with comrades so to hear what they really thought. The earliest attempts at putting lie-detection on a scientific footing go back to the start of the 20th Century. Robert House, a US obstetrician who had noticed the effects of an anaesthetic drug called scopolamine. One of his patients was in a state of "twilight sleep" after being given scopolamine. Dr House asked her husband to find a weighing scale for the newborn but he returned empty-handed, whereupon his wife told him exactly where it was while still apparently asleep. Dr House wanted the drug to be used as a way of supporting people's claims to be telling the truth.

US police from the 1920s onwards began to use drugs for interrogation. They experimented with the psychoactive barbiturates sodium pentothal and sodium amytal. However, it soon emerged that the drugs were being misused on suspects. It became clear the drugs had the same problem as torture – they made people say things that they thought the interrogators wanted to hear.

During the Second World War the first intravenous anaesthetics – the same class of barbiturates – were used on traumatised soldiers who had lost their memories to get them to remember what happened to them.

After the war, some American doctors continued to use sodium amytal and sodium pentothal on psychiatric patients to get them to talk about hidden memories. Some of these doctors also took their expertise in this area to the police and the US Government.

The most notorious search for a universal "truth serum" was the secret MK-ULTRA project of the Central Intelligence Agency, which started in 1953. The CIA tested many drugs, including the hallucinogen LSD, sometimes without the knowledge or consent of people who unwittingly took part in the experiments.

Officially, the law-enforcing agencies of the American and British governments do not use drugs during interrogation and no court of law in either country would accept statements that have been made under the influence of a so-called "truth serum".

Scientists have produced fairly convincing evidence that drugs such as sodium pentothal do not extract truthful memories. Instead, they tend to make interviewees more talkative in a way that makes them suggestible to cues elicited by interviewers for what they would like to hear.

"After 9/11, there were discussions in the national papers about whether it's a good idea to interrogate suspects using these drugs," said Alison Winter, a science historian at the University of Chicago, writing for the magazine Scientific American.

"Every time there is a desperate need for information from people, you get speculation about whether these drugs are going to get that information. But you also get consistent warnings that the information may be less reliable than what you would get without the drugs," Dr Winter said.