Homepage

Back

Archives

Sniffer spy rats stand up for undercover duty
Irene Burgo

Can you imagine a mischief of rats showing up for duty at a disaster scene or at an airport baggage depot? It's not as far-fetched as it sounds, and the rats won't be misbehaving. They'll be working. Computers conceivably could one day train squads of sniffer rats to detect contraband odors such as drugs or explosives and subsequently monitor their performance on the scene, similar to the way dogs are taught to detect odors.

Dr. Michael F. Brown, professor of psychology and director of the Graduate Program in Psychology, and his colleague recently tested the concept. Brown, whose expertise is in animal behavior, especially in spatial cognition and spatial memory, with Dr. James Otto, a professor of computer science at the University of Baltimore (who was a former Villanova faculty member in the College of Commerce and Finance's decision and information technologies department) recently concluded a research project for the Defense Research Projects Agency, DARPA, to teach computer-directed rats to sniff and detect odors.

"Our goal was see if it was feasible to use rats in the way that officials have traditionally used dogs for detecting contraband odors, prohibitive foodstuffs, or other substances that are of interest to authorities," explained Dr. Brown. "Rats have some advantages over dogs and also over using instrumentation to detect substances." Rats are smaller than dogs and could be used in specific applications where dogs could not be used, for example, because of the fact that the dog's presence would be obvious. "Rats are conceivably better at screening odors than machines because instrumentation does not detect all substances of interest," said Brown. "And once they are taught to detect odors, they are almost always correct in discriminating the scents."

Brown and his colleague taught laboratory rats to rear up on their hind legs when they smelled simulated cocaine or other mimic odors in the lab. The rats were fitted with a harness and lead containing a motion sensor, which alerted the computer to their location while the professor remotely monitored the results. The rats first were trained in a Skinner Box--an instrumented conditioning chamber--to alert by demonstrating a prolonged rearing response when they detected odors. If they correctly responded when they detected or did not detect odors, the rats received rewards of automatically dispensed sugar pellets. After learning to alert, the rats were further trained in a more open setting of an arena, roughly one square meter in size. The arena contained about 25 cups where food pellets were randomly placed within the cups containing the simulated chemical odors. After repeated trials in the area, the lab rats were able to discriminate odors, such as almond extract and motor oil, and alerted by standing upright more than 90 percent of the time.

Sniffer rats have several advantages over dogs. Because of their smaller size, they can fit into places where dogs can not be sent, they are cheaper to train and might be more suited to dangerous situations in military operations such as mine field detecting.  

Brown noted that the results of the research were favorable, but he cautions that it will be a long time before the CIA or FBI hires the rodents as special agents. "The concept makes sense and the techniques are workable, but the research is at the very early stages of development," Brown noted.

The challenges of working with sniffer rats in the real world are more rigorous and complicated than in a lab setting. To make rats into real-life sniffer spies would require more extensive research and some type of remotely controlled state-of-the art guidance technology that has not yet been engineered, according to Brown. For example, the professor believes rats could be further trained to go where directed, if highly sophisticated sensors were developed that would allow the person in charge to remotely locate and keep track of the rat and detect their responses This kind of  research is nowhere near reality. "In theory, the technology certainly could be developed," said Brown, "but it might cost a fortune."

An article on Brown's research was published in January 2002 in the journal Nature Science Update.