Planned terrorist attack in the works for the Netherlands discovered – in a bar?
That’s right, in a bar, with documents labeled Top Secret right in front of the terrorists. Only it wasn’t a real plan, and they weren’t real terrorists (although it was a real bar). They were 12 students from Maastricht University in the Netherlands who were carrying out an experiment to show that someday a certain test (called a concealed information test) could surpass existing lie detector techniques used in uncovering information from groups of suspects.
Each student was given an envelope marked “Top Secret”. Inside are plans for a terror attack somewhere in the country and they are instructed to tell no one. Later, the students were arrested under suspicion of terrorist activities and interrogated using the concealed information test (CIT). The test uses sensors to measure the skin moisture levels. Then math techniques (with a little help from artificial intelligence) analyze the findings to work out what the students were hiding.
A group CIT is able to uncover just one piece of information that is being concealed by the entire group, compared to the traditional lie detectors that concentrate on individuals. If a group of individuals are concealing a planned terrorist attack, say, in, I don’t know, New York City, a group CIT would pick that up.
Ewout Meijer, who led the experiment, says that many of today’s security threats are from terrorist groups involving multiple suspects.
Lie detection techniques usually compare a suspect’s responses to a series of questions about a crime against responses to unobjectionable questions. These responses are recorded by a polygraph, which measures the suspect’s pulse, respiration rate and skin conductance. The problem lies (couldn’t resist the pun) with the detectors, which are notoriously unreliable.
In this experiment, Meijer used a CIT to see if a suspect reacted strongly to any particular question in a series of similar ones. Meijer’s team questioned the 12 students while measuring their skin conductance by attaching two small electrodes to their fingers. The students were then presented with a series of questions that listed possible cities, dates, and specific targets for the planned attack, and included the correct information for each. The students were told to answer no to every question.
If the suspect is guilty, a question that identifies an element of the plot will give off an independent nervous system stimulating response, which is stronger than for other questions in the series. This triggers a spike in both skin conductance and an EEG signal that serves to act as a sort of flag. An innocent person’s responses would not spike at all.
The skin conductance test results were analyzed by software that contains mathematical models called Bayesian networks that deal with probability which is the base of artificial intelligence. When all the data was compared, the software found with high probability that all of the suspects knew that the attack was planned for August 15 in a particular department store in Rotterdam.
This all sounds well and good, but if the individuals in a group only have small bits of information about a planned attack, then Meijer admits that this method would be far more difficult to analyze.
Still, Frank Horvath of Michigan State University in East Lansing, says the technique could be useful for fighting terrorism. “It won’t necessitate tactics that we are all familiar with and find distasteful, if not worse,” he says.
If there can be a group CIT, I don’t see why there can’t be one for a single suspect. At least then we’d have something that could be admissible in a court of law.