Puppy Brain Scans Could Help Pick the Best Dog Bomb Sniffers
Credit to Author: Lily Hay Newman| Date: Sun, 27 May 2018 11:00:00 +0000
If you've been in a large crowded place with a fair amount of security, you've probably seen bomb-sniffing dogs at work. (You may have even petted the puppers.) Dogs have long been used for detecting contraband and explosives, but attackers have made advances in body-worn explosive technology, forcing law enforcement to evolve too. This means enlisting a new type of canine defense: Vapor Wake dogs. These highly-trained, specially-selected animals, typically Labrador retrievers, sniff the air as people walk by to detect even faint trails of explosive particles that could indicate a body-worn bomb. This allows just a few canines to check tens of thousands of people. You may have already seen one out in the wild, at work for Amtrak, the New York Police Department, or at a certain Thanksgiving Day Parade.
Because this technique is so specialized, the dogs involved must undergo intensive training from a young age. So before investing in that two-year process, behavioral researchers at Auburn University (where the Vapor Wake training technique was originally developed) want to know if they can identify the behavioral and neurological indicators of which puppies will make good Vapor Wake dogs. To do that, they're putting really cute puppies in fMRI machines. You know, to save the world.
"The best dogs, the ones that become the Vapor Wake dogs, by about six months and sometimes as early as three months we start seeing differences between those dogs and the dogs that don’t graduate to that same level," says Jeffrey Katz, a cognitive behavioral researcher at Auburn who is leading the research into Vapor Wake puppies. "We're starting to see a number of factors that are predicting success. The newest are in cognitive tests at different time points that involve various domains—physical skills, social skills and some general descriptor tasks. Think of it as a battery of intelligence tests for dogs."
In one cognitive experiment the researchers put a treat on top of a small box and have a dog approach and eat the treat while a person stands by watching. Then researchers put a treat inside the box, but the dog can still get it out. Finally, they put a treat in the box and lock it so when the dog approaches she can't access it. In the first two phases of the test, researchers look at how quickly the dogs get to the treat; in the third phase, it's about how quickly and how often the dog looks between the nearby person and the box to try to signal to the human that they need help getting the treat out. By 11 months, dogs that spend more time "gaze shifting"—looking at the box and then looking at the person—are more likely to eventually become Vapor Wake dogs.
The research is also expanding into neuroimaging to see if measuring brain activity from various stimuli can reveal anything about which dogs will succeed as Vapor Wakes. Researchers have been investigating what parts of dogs' brains are active while they process different types of information. In some experiments they show the dogs pictures of familiar and unfamiliar people making happy, angry, and neutral faces, and then measure the dog's response. Though the imaging work is still in the early stages, Katz says it has shown some promising correlations between measurements from the behavioral studies and certain types of brain activity.
The group hopes to form a roadmap for developing neurological indicators of success from these types of connections. And another long-term goal for the research would be to expand into genetic studies of the dogs as well. If the researchers could uncover genetic markers that predict which puppies will graduate from Vapor Wake training it would be the easiest and cheapest indicator to check at scale.
Research on using fMRI in dogs is still evolving, and experts say it is important to be wary of over-extrapolating. When studies use pet dogs as subjects (a common practice to increase test pool and reduce the costs of boarding dogs), they inevitably have wide variation in their histories and home environments, which can impact findings in unexpected ways. Vapor Wake puppies come from a more controlled environment, but still have variations as their training goes on. And even basic things like training a dog to lie still for an fMRI in the first place can skew the population sample in studies, because dogs that can't stay still can't be scanned. fMRI studies in general, regardless of subject, are also expensive to conduct, which limits sample size. Katz's group has imaged 37 dogs so far, a reasonably robust number.
The benefit of using fMRI along with behavioral studies is the potential to catch indicators of success or failure even earlier. "The value of doing this type of research is if you wait or rely entirely on a dog’s behavior, it’s too late in a sense," says Greg Burns, a researcher at Emory University who has been working with dog fMRI for years, including in a 2017 study that looked for neurological indicators of successful service dogs. "Some switch has flipped in the dog’s brain and they’re doing something you can observe, but you’d like to have more insight beneath the surface before something is apparent from the behavior."
Burns notes, though, that it's valuable to combine behavioral studies with fMRI data to get a fuller picture and corroborate findings. "If we’re talking about detection work, the dog is not just a nose, it's not just a portable sniffing device," Burns says. "There's a sentient creature on the end of that nose and that creature has to communicate with human beings and do all sorts of cognitive processing."
The more options handlers have for assessing puppies and tailoring their Vapor Wake training, the more dogs can be deployed in the field for bomb detection.
Vapor Wake dogs also may be in the field a long time and their skills need to have staying power. VWK9, a private company, manages and reassesses every Vapor Wake dog each year to confirm its performance quality. So researchers are hoping they can eventually unlock the key to determining which dogs will succeed for the long-term. "fMRI research can have weaknesses—if you work data long enough you can get stuff out of it," Katz says. "So what's key is to be able to take fMRI data, behavioral data, genetics, all the different techniques that you have, and if they all start converging on the same explanation, then you start to have a good story."
https://www.wired.com/category/security/feed/