They found that dripfeeding evidence to the suspect boosted the detection rate to 68 per cent compared with 50 per cent – no better than chance – for the existing methods where all the information is revealed either early or late in the interview.

Their discovery has implications for any official interviews such as benefit fraud investigations, medical negligence and inland revenue cases.

The researchers say that revealing information gradually “exploits gaps in a deceiver’s account immediately inconsistencies begin to emerge, while at the same time providing innocent interviewees an early opportunity to convey their honesty”.

The study published in Legal and Criminal Psychology was carried out by Dr Coral Dando, Professor Tom Ormerod and Alexandra Sandham from Lancaster University, and Professor Ray Bull from Derby University.

The research involved a total of 151 people who were all asked to build part of a virtual Olympic stadium in an interactive computer game.

But some of the participants were also assigned the identity of deceivers or “terrorists” plotting to blow up the stadium and they had to hide their intentions by, for example, covering up the purchase of large amounts of “explosive” from virtual shops.

They were then interviewed about the game by a trained police interviewer who revealed relevant information either early on, at the end of the interview, or gradually. Thirty laypersons viewed a random selection of interviews and then judged whether the interviewees were telling the truth or not.

Their judgements were most accurate in the interviews when information had been revealed gradually and the laypersons were more confident about their judgements. In addition, the so-called “terrorists” said that the gradual disclosure of information by the interviewer was much more mentally taxing, making it harder for them to sustain the deception.