Teaching dogs to sniff out
corpses or drugs or bombs has traditionally been more craft than science. But
some novel synthetic substances may soon change that.
copyright Jessica Snyder Sachs, as originally published in Discover magazine
FOUR YEARS AGO, standing under the Arizona sun, Detective
Mark Green thought about the search ahead and felt a little queasy. Four eyewitnesses
had each told the police a similar tale of young children murdered, eight years
earlier, on a moonlit desert night. On this day the Phoenix police would search
for their remains, reportedly buried somewhere on this desolate plateau
His partner, Green remembers, was far more enthusiastic--his shiny brown coat
was twitching with excitement, his tail whacking against Green's leg as they
stood side by side. Judge, a chocolate Labrador retriever, was accustomed to
sniffing out dope, but recently he'd been learning a new scent, that of a
human corpse. His education, though, was somewhat experimental: he had learned
this scent not from real bodies but from an artificially concocted perfume that
purportedly captured the smell
Green now broke open an ampoule of
that perfume and gave Judge a whiff. "Back!" he commanded--as in, "Here's what
I want; now bring it back!"
Judge traced a switchback pattern across the baked red earth, his nose jumping
like a rabbit's. He paused to smell
the flattened remains of
something furry, then moved on. After 30 minutes, he slowed, swept his snout
back and forth, and started furiously digging. "Good boy!" said Green, bouncing
a red chew-toy in front of
his partner's nose. The Labrador bounded away with his reward.
Green brushed over Judge's scratch marks and took the dog several hundred yards downwind to repeat the search. Within a few minutes Judge returned to the same spot, again scratching and barking. Now the humans dug. They found old diapers and shreds of rotted clothing.
Unfortunately, the site was ground zero for an overpopulated pack-rat colony. It looked as if any flesh and bone, had it been here, had been eaten or carried away. Still, the human odor remained, according to Judge, who returned to claw, bark, and bite at the unearthed clumps of clay.
A forensic pathologist from the University of Arizona in Tucson arrived and pointed out a subtle depression in the desert floor. Not natural, he said, but more like the settling that would follow the filling of a wide hole. The digging continued with backhoes, and the police combed through the clay for more evidence. One of the alleged victims, a preschooler, had been described as wearing cowboy boots. Even if the leather was gone, the metal shanks should have remained. The police found nothing, but by finding those scraps of cloth, Judge became one of the first dogs in the ranks of the pseudoscent-trained.
In the wake of the Oklahoma City bombing and the Kobe earthquake, sniffing dogs have become a common sight on television. What the pictures don't communicate, though, is how difficult it is to train a dog to track a given scent. The dogs have to be worked at least once a week, and if the scent in question is that of a corpse, the trainer's life can get complicated. Carrying around the odor-laden ooze from a corpse is not a great way to win friends. "When I'm on a three-day trek in the desert, the real stuff will get me kicked out of camp pretty quick," says Green. Even training a dog to recognize a drug like heroin is problematic.
To acquire and use illegal drugs, a trainer has to plow through mountains of paperwork; moreover, a dog can easily overdose if it gets a snoutful of the stuff.
Trainers have therefore tried to replace the real stuff with substitutes. "For heroin and cocaine, we mixed up a paste of powdered milk, vinegar, and a little quinine," says Texan Billy Smith, who began training drug-sniffing dogs in the 1970s. Similarly, dogs slated for search-and- rescue missions are trained on everything from roadkill to hair and nail clippings to their trainers' own blood. Sometimes the substitutes work. But just as often, they don't.
A small cadre of chemists and biologists believe that science can make the training of dogs easier and more reliable. Their most visible handiwork, commercially available pseudoscent, is manufactured by the Sigma Chemical Company in St. Louis. Over the past five years, Sigma has developed a unique product line that now includes Pseudo Corpse I (for a body less than 30 days old), Pseudo Corpse II (a formulation designed to mimic the dry-rot scent cadavers attain after a month), Pseudo Distressed Body, and Pseudo Drowned Victim. Pseudo Burn Victim is in the planning stage. Sigma also sells a pseudo powder explosive and a line of pseudo illegal drugs.
Pseudoscent Variety Pack
"In theory, the pseudoscent is the way to go," says Larry Myers, a sensory
biologist and veterinarian at Auburn University in Alabama, "because the truly
difficult thing about training a dog to a scent is stimulus control." The ideal
compound, he says, should capture an odor signature common to everything you
want a dog to find, but nothing else. "You don't want a dog trained to find
explosives hitting on a can of
shaving cream." Even given the amazing sensitivity of a dog's sense of smell,
such things do happen. For example, Myers tells of a narcotics officer
who had trained his dog on drugs kept in plastic storage bags. "I'll be damned
if that dog didn't start alerting to the scent of Ziploc bags," says
Myers. A dog trained on street drugs can likewise get distracted by cutting
agents, homing in on baking powder in the fridge and ignoring uncut cocaine in
"Reliability is crucial," says Myers, "because today search dogs are being used in life-and-death situations." Among those who rely on such dogs is the Federal Aviation Administration, which deploys roughly 100 canine search teams to check suspicious-looking air cargo for explosives. The faa might be interested in using pseudoexplosives in the future--one reason being that real explosives have a nasty way of actually exploding-- and so it sponsors research on dog olfaction, including Myers's. But before he or anyone else is going to be able to come up with a reliable pseudo bomb scent, Myers says, there's a lot of basic science that needs to be discovered.
Researchers know that when a dog sniffs deeply and odor-carrying molecules flow into its nasal cavity, the shape of the cavity changes so that the molecules are focused onto a yellow, rippled, mucus-covered membrane, called the sensory mucosa, toward the back of the snout. So convoluted is the canine mucosa that if it were smoothed flat it would be several times larger than the dog's head. Because it has so much surface area, the mucosa can carry a vast number of odor-sensitive, hairlike cilia- -ten times more than are found in a human nose.
But beyond that, researchers know very little. They have yet, for instance, to define the limits of a dog's sense of smell. A dog may be able to track the day-old trail of a fugitive, yet when it comes to certain smells, such as that of acetone (the sweet smell of nail polish), a dog's nose is no better than a human's. No one has yet systematically sorted out just what a dog can smell and exactly how it does so.
Against this background of meager knowledge, Sigma chemists Thomas Juehne and John Revell created their first pseudoscents in 1989. Dog handlers working for federal agencies had come to Sigma asking for compounds for training narcotics dogs. Revell began with heroin and cocaine, each of which consists of a single big complex molecule. "With such pure, large compounds," he explains, "we knew we had to find some outer piece to work with, a little active site that might break off from the main molecule." Such a piece would probably be safe--that is, nonnarcotic--yet present in the air around the drug, so a dog could be trained with it to recognize the drug.
Fortunately, U.S. Customs had already done a lot of Revell and Juehne's work for them, analyzing the gases that float above both heroin and cocaine and isolating a variety of alcohols, alkanes, esters, and acids. All were readily available in Sigma's catalog of 35,000 laboratory chemicals. Revell and Juehne could proceed directly to a game of mix and match: they developed several test batches for each drug and sent them to six handlers with dogs already trained on real narcotics. Each handler was asked to try to have their dogs find a hidden sample. The dogs completely ignored some samples while showing keen interest in others, and from these Sigma created refined formulas.
Search dogs assist diving team in body recovery.
After a confirmation round with the veteran dogs, Revell sent the most
promising signature for each drug to a second set of handlers, asking
them to use it to train new dogs not yet exposed to the scent of actual drugs.
Reports came back that these pseudo-trained dogs were then able to locate the
real stuff. Voilà: Sigma had its first pseudos.
Developing a pseudomarijuana has been more complicated, says Revell. "Instead of a single pure compound, now we're working with a whole plant." To isolate the molecules in marijuana and determine their abundance, he uses both gas chromatography, which can separate chemicals based in part on how quickly they evaporate from a liquid, and mass spectrometry, which identifies compounds according to their atomic mass and charge.
Revell looks in particular for substances that will become gaseous even at low temperatures, since these would be the compounds most likely to waft from a hidden stash. "Unfortunately, we discovered that not all dogs alert to the same thing," he says. Though all the dogs had been trained on whole marijuana, they had apparently selected different signature chemicals to use for identification. Revell was able to produce a commercial pseudomarijuana by taking several of the most popular compounds and combining them in a scent cocktail, on which all the dogs hit. Still, he wants to tinker with the formula more, since Sigma has received occasional reports of the cocktail's not working. "The first came from Saudi Arabia," says Revell. "My hunch is there may be differences between marijuana varieties worldwide."
In 1990 dog handlers let Sigma know about the troubles they had training their dogs on corpses. Because this type of work comes in irregular spurts, handlers need to train their dogs continually--at least once a week, preferably more. Their substance of choice is dirt collected from under a corpse, which becomes infused with its putrid smell. Reflecting the callousness probably essential to the job, the handlers refer to this training aid by a number of names: "dirty dirt," "Mr. Sousa," or "Fred," as in "Fred B. Dead." Nobody likes handling the stuff. Trainer Carl Makins, of the Greenville, South Carolina, sheriff's office, keeps his double wrapped in plastic and locked inside a vapor-proof munitions cache. When he opens the box even for a second, he saturates the room with a sickeningly sweet smell. (Think skunk meets Montezuma's revenge.) But that's the least of a trainer's worries--there's also the threat of infection associated with hiv, hepatitis, and other diseases transmitted through body fluids.
Hearing such complaints, Patricia Carr, Sigma's liaison to the dog handlers, went to Revell and Juehne and said, "Give me body in a bottle." At first they looked at Carr as if she were crazy, but eventually they warmed to the idea. That's not to say that they allowed any ooze in their lab, let alone in their gas chromatograph. "That would have been difficult for me," Juehne says with an audible shudder. Instead he searched through scientific journals and found that the human body had been well quantified in various states of decomposition.
Five to fifteen minutes after death, protein synthesis in the body grinds to a halt. With nothing to maintain the protective lining of the gut, digestive enzymes eat the body from the inside out, splitting proteins into amino acids. At the same time, the body's resident bacteria, unhindered by an immune system, feast on the amino acids and skyrocket in number. As the bacteria produce chemicals such as ammonia and ptomaines (with such apt names as putrescine and cadaverine), they produce the distinctive smell of decaying flesh. Each stage of decomposition produces distinct peaks and ebbs in the levels of various chemicals, including the ptomaines, which is a great help to both the pathologist who wants to determine the time of death and the chemist trying to emulate the smell of it.
Juehne cataloged the chemicals most likely to be in the air or soil around a decaying human body--both fresh (Pseudo Corpse I) and well aged (Pseudo Corpse II). Among these, he looked for chemicals that might set the smell of a human corpse apart from that of an animal. "I needed something unique about the human body versus a dead animal," says Juehne.
Juehne's preliminary guinea pig was Revell. Revell had joined up with Sigma after seven years in a forensics lab, where he often worked alongside coroners at autopsies and crime scenes. "Basically," says Revell, "once Tom had a list of potential compounds, he began running them by my nose and asking, 'Does this smell like a corpse?' I'd say, 'Yeah, that's close,' and he'd disappear back into his lab to refine it."
Juehne diluted his scents to a level indiscernible to humans and sent them to a half-dozen dog handlers. The first batch was well received; a more refined brew drew raves. "I started out their biggest skeptic," says Billy Smith, "but as soon as we hid this stuff in a sandbar, the gators stole it. Then we put some in a tree, and the coons stole it; in a log stump, and the buzzards stole it."
Another tester was Caroline Hebard, a New Jersey mother of four who has been honored internationally for her canine search-and-rescue work. "Yes, this works," she told Sigma. "Now give me something for live folk."
But not just ordinary live folk. Over the years, as Hebard and her dogs sifted through the rubble of earthquakes and explosions, she saw that she needed to train her dogs to tell her if they were smelling buried trauma victims or the workers around them. "There's a certain scent, kind of sour and sweaty, around someone in shock," she says. "Anybody who's familiar with the smell in an ambulance knows what I mean." To fill the request, Juehne hit the journals again. There he found detailed analyses of compounds our bodies secrete onto the surface of our skin. "I needed to find a universal human scent, something nondiscriminatory with respect to a person's diet, sex, or age--from that baby-fresh smell of a newborn's head to the musty odor of Grandpa in the nursing home."
After another game of chemical mix and match, Juehne sent Pseudo Distress for field-testing. It reportedly sailed through all trials, with claims that dogs trained on the stuff were proving their worth in actual rescue situations. And not only did Pseudo Distress help dogs track people in shock: handlers report that it's good for finding frightened children in the wilderness and adrenaline-charged escapees in prison air ducts.
The company added its most recent pseudoscent--Drowned Victim--by reformulating its corpse tinctures into a granulated capsule that sinks in water. "The first batch was like Alka-Seltzer," comments Hebard. "It had the dogs jumping to bite overhanging branches." What trainers needed was a scent that would collect in a thin film just on top of the water's surface- -as true cadaver scent does--so dogs could follow its concentration gradient to the source. Accordingly, researchers at Sigma made a slower- dissolving capsule and filled it partly with salt grains to make it sink. "That did the trick," says Hebard.
For all the testimonials to the pseudoscents' effectiveness, there is still plenty of room for skepticism. There are no statistics from a controlled test of pseudoscents with large numbers of handlers who themselves did not know where the samples were hidden. Nor has the accuracy of dogs trained on pseudos been reliably compared with that of dogs trained on the real thing.
"We need to separate the science from the mumbo jumbo," says Myers. To begin with, he says, nobody yet knows what a dog is physiologically capable of smelling. A simple analysis of a drug or a decaying body won't tell you which chemicals are of canine interest.
That question is among those Myers is trying to answer at Auburn's Institute for Biological Detection Systems. Myers founded the institute in 1989 to study everything from actual canaries in coal mines to microbes that glow when exposed to pollutants. But for now, 90 percent of his grant money arrives earmarked for studying canine detectors.
The work begins in the sensory lab: wrapped in a baby-blue blindfold, a tan cocker mix lies on a padded table. Two students murmur reassuringly as they clip to the dog's scalp electrodes that will pick up general patterns of brain activity when she is presented with a test smell. They also focus an overhead camera on her head. One student then lifts a test tube suspended from a two-foot glass handle. As the tube nears the cocker's nose, an electroencephalograph across the room traces eight jagged lines to record a spark of brain activity. The overhead camera captures the slight movements of a sniff.
Myers's students are determining the limits of the cocker's sense of smell with a dilution of eugenol, one of the odor-producing molecules in cloves. Myers employs eugenol as a standard for determining whether his test dogs are having an off day, since dogs, like people, experience a range of colds and allergies that can interfere with smell.
If the cocker's sense of smell is up to snuff, the students test her ability to detect vanishingly small amounts of an explosive and then several of its volatile ingredients. Ultimately, Myers would like to isolate just one or two key chemicals that dogs can use to recognize the entire explosive. The result could be a safe, reliable pseudo.
Enlisted in the effort are two men that Myers admiringly calls the institute's control freaks: chemist Mark Hartell, an eager young doctoral student with a passion for ferreting out contaminants, and experimental psychologist Jim Johnston. Skinnerian to the bone, Johnston is likewise obsessed with purging contaminants--the type that creep in when humans bring subjectivity to the study of dog behavior.
To begin with, says Hartell, "what's in the list of ingredients is not necessarily what's in the air around an explosive. If the guy down the hall is wearing Polo, that doesn't mean the explosive you're studying is made by Ralph Lauren." Today Hartell is fueling his gas chromatograph and mass spectrometer with air drawn from the explosive under study. He's already identified dozens of airborne compounds, many of which were contaminants from the institute's house air. ("Dirty stuff," he comments.) Many of the other compounds break down too quickly for a dog to notice. That leaves a dozen or so worth examining.
The researchers use conditioning experiments to test these remaining chemicals. Their subject dogs do their work in the isolation of six wooden chambers--oversize Skinner boxes--in a room slung with computer wires and plastic air tubes. No human handlers here. "Uncontrollable variable," says Johnston--humans have a habit of unconsciously affecting the response of dogs by subtle changes in their appearance. Johnston prefers the objectivity of a computer program. Each chamber is equipped with a nose cup attached to an olfactometer, a glorified air pump that delivers a precisely calibrated flow of clean or scented air. Inside the chamber, slightly above the cup, are three levers. The dogs have been trained to press the right lever when they smell the explosive under study, the left lever when they get a puff of clean air, and the middle lever when the air contains a scent other than the explosive.
In chamber two, a white shepherd named Columbus begins her eight- hundredth session. (Each dog works one hour a day.) At the sound of a tone, she inserts her snout in the cup. At a second tone, she removes it and paws the middle lever: in other words, she smells something, but not one of the explosives she has been trained on.
So far, admits Johnston, none of the dogs trained to recognize the explosive have responded to any one isolated ingredient. But only a few chemicals have been tested as yet. If no one ingredient evokes a response, they will try two- or three-chemical mixtures. Developing a pseudoscent in this way is time-consuming, Myers admits, but it may help reveal the classes of chemicals to which dogs are most sensitive. To know whether dogs are indeed more attuned to certain compounds, and to identify which ones, would elevate canine training to a new and reliable height.
Till then, 1,850 dog handlers will continue to use Sigma's fascinating but sketchily tested perfumes. According to the company's customer logs, sales have more than doubled in the last three years. Often the trainers who buy them use them in combination with more traditional materials. "I like to mix it up," says Smith, who trains dogs initially with corpse pseudoscent, then graduates them to dirty dirt. Hebard combines pseudoscent with human hair for "a very strong response."
Many handlers, though, steer clear of the scents. "Using pseudos is like going to the firing range with blanks," argues David Frost, canine training supervisor for the Tennessee Public Service Commission. "The strongest thing we have going for us is the dog's amazing power to discriminate one thing from another. So why muck that up with anything but the real thing?" Using real drugs means leaping over a long series of bureaucratic hurdles, he admits, "but sometimes to do something right isn't convenient."
Still, the stories of success linger and tantalize. Out in Arizona, Detective Frank Shenkowitz, who inherited Green's grisly case, remains haunted. "I still go back there fairly often," he says of the plateau where the pseudoscent-trained Labrador Judge uncovered the decayed clothes. "You never know what the desert is going to toss back at you." A while ago, not far from Judge's find, Shenkowitz came across a tiny faded cowboy boot--the size a four-year-old might wear. "It doesn't prove murder," he says. "But I know Judge is reliable."
Science writer Jessica Snyder Sachs is the author of Good
Germs, Bad Germs: Health and Survival in a Bacterial World (Hill&Wang/FSG)
and Corpse: Nature, Forensics, and the Struggle to Pinpoint Time of Death