That’s not to say that dogs should be automatons that simply factor in stimulus and response. They should still be somewhat freethinking. I use a clicker during training to mark behaviors—a simple kind of association tool that allows me to communicate to the dog that the behavior it just exhibited was a proper/acceptable one. In the beginning, I hold a treat in my hand. I sit there looking at the dog, he looks at me, I click, and then he gets the treat. Naturally, the dog will start to offer up behaviors, hoping that one of them will get me to reward him. I don’t even have to give him a command, and I don’t. If he goes into full down position, I click and reward. If I want to teach him to bark on command, I wait until he barks, and then I click and reward.
As another example, if I want to get a dog to go into his crate, I place a crate in the middle of a clear room—one with no other distractions; if the dog takes a step toward that crate, I click and reward it. If he takes another step, I repeat the process, and so on. When we get to more complicated tasks, that same process applies. With the crate, repeat that often enough, and the dog will go in that room, see the crate, dash into it, lie down, and wait for the treat/reward. This is a win-win situation. The dog gets a treat, and I get the desired outcome.
Obviously, I could put a prong collar on a dog, give no treats, and I could get to the same place by punishing him for not doing what I wanted. He will learn just as quickly what the answer is, but there’s an enormous paradigm shift between the positive and the negative. You’ve got a dog that is doing what he’s “supposed” to do because he doesn’t want his ass kicked, versus a dog that is doing the right thing because he wants to, because he wants that reward. One builds trust, the other destroys it.
What’s important to remember is that that kind of training is, in a sense, an act of coercion—getting an animal to do what we want it to do, not letting it do what it naturally wants to do. The other side of the training coin, one that isn’t a win-win, and one that I don’t like to see used, is the old spiked/pronged/electric-shock choke-collar approach. I mean that literally and figuratively. A spiked training collar, a whip, a stick, or any other tool to mete out punishment to show a dog who’s boss, to punish it when it doesn’t do what we want, may eventually get a dog to do what you want, but at a great cost to the dog. So why do it?
You can use one of four methods of operant conditioning to train a dog. The first is positive reinforcement. Keep in mind that the word “positive” in this case refers to addition—giving something. The classic example of this is a rat in a maze receiving a reward, a food pellet, when it successfully reaches the end of the maze. The opposite of that is, obviously, negative reinforcement. Here, “negative” refers to taking something away, or subtracting. Again, using the rat-in-the-maze experiment, if a loud noise is made while the rat moves through the passages, and then it is silenced when the rat succeeds, you’ve used negative reinforcement—you took something away but still rewarded the rat by getting rid of that awful sound.
The third type of operant conditioning is positive punishment: you add a stimulus, usually something painful, when a behavior you don’t want occurs. An electric-shock collar is an example of that kind of operant conditioning. Negative punishment, as you can probably guess, is taking something away when an undesired behavior occurs. Your kid acts up, so you take away her cell phone.
For 90 to 95 percent of the time, when training dogs—either pups or adolescents in early or advanced training stages—I use positive reinforcement. I equate relying heavily on the other three to this: you want your kid to learn algebra, so you set a book down in front of her and demand that she does the equations starting on page one. You don’t provide any kind of instruction or encouragement. You only use positive punishment or negative punishment as consequences. She can’t solve the first problem, so you slap her hand. Repeat and repeat. Sure, at some point your kid might learn algebra, but it will take a hell of a lot longer and be more painful for you and for her if you do it that way. Far too many trainers and handlers spend far too much time using positive and negative punishment. They may get results, but the essential bond of trust between a trainer and a dog won’t exist. A dog may do what you want out of fear or to avoid pain, but that kind of relationship is one that will always be out of balance. Just like the Navy SEALs are a team and watch out for one another in every way because of their trust and respect for one another, the same has to be true of how a military working dog and his handler operate.
Another takeaway from that algebra analogy is that you almost always have to go into a situation when training a dog carrying along this assumption: they don’t know anything but body language and the environment. They are not stupid, they are not incapable of learning, but they do need to be taught. Invariably, because dogs learn through association and repetition, the learning process will be longer than we might like. For example, once dogs enter into our advanced training, even though they’re already titled, it may take one to two years before they are ready to be deployed. That’s a lot of time and innumerable repetitions before they are truly ready to do the important work necessary of them.
Besides my respect and love of dogs and the value we place on the human/canine bond, I spend so much time using positive reinforcement simply because I can’t imagine how hellacious my day would be if I spent the vast majority of it punishing the dogs. How awful would that be? I can’t imagine the toll that would take on you mentally, and it’s no wonder that I see people, both in the military-working-dog community and pet owners, allowing their emotions to get the better of them. One of the most frequent mistakes I see people make with their dogs is that when “correcting” them, they get the dog to either stop doing what they didn’t want it to do or get it to do the thing they wanted, and they continue to either berate or punish their dog. That’s incredibly confusing for a dog. You told me to stop, I stopped, and now you’re still screaming at me or correcting me. Does that mean that you don’t want me to do what I’m now doing?
With pets, you end up with dogs that aren’t clear about what’s expected of them. That’s not good. With working dogs, when excessive punishment is used, you end up with either a handler-aggressive dog or a broken dog—not in body but certainly in spirit. We need dogs with intense spirits to take on the challenges of being outstanding detection and apprehension dogs. As I’ve pointed out before, the pool of suitable dogs is relatively small. To see a dog with the right characteristics ruined by poor training methods is both heartbreaking and an enormous waste of a valuable resource.
Negative and positive punishment can break the spirit of a dog and this is especially true of dogs that are just entering adulthood, the prime age for specialized training. It seems counterintuitive to me that we select and breed dogs to have a fierce and courageous demeanor and then try to take that spirit away from them, especially since they will need that kind of character to charge into the austere environments that we ask them to. With the dogs we train, we want and need them to feel like they are kings of the world. They better have a nasty, hard, confident attitude, since they may very well be taking on insurgents who want nothing more than to kill them and their human associates. They’ve got to have a pair on them that will allow them to charge through the gates of hell to go bite a guy who’s not going to just sit there and let them take him down.
So, if you’ve taught a dog from the age of seven weeks on that he’s below humans in the chain of command, and he’s then going to do whatever he’s told out of a desire to avoid pain, whenever the pressure of a human being is being placed on him, he’s going to fold. I don’t care how strong a dog is genetically, if he’s been trained in that dominated/coercive/negative way, he’s going to be ruined. He’s learned through his whole life that a human being can dominate him. Would you want to go into battle thinking that? Whether it’s a friendly game of tennis, softball, or whatever, if you go into it thinking you can be beat, chances are you will.
That’s not to say that there doesn’t come a point in
a military working dog’s training when the dog is old enough, powerful enough, that you as trainer have to set some rules and teach the dog certain manners and obedience, or you’re going to end up losing fingers or getting puncture wounds all up and down your arms. That kind of uncontrolled and naked dog aggression is counterproductive as well. Simply put, though: if you are counting on a dog to work to protect you, to apprehend human bad guys, they can’t be scared of anyone.
Sometimes the demands of the marketplace mean that a dog is rushed through the training pipeline too quickly. Some trainers make too many demands too soon on a dog after they first acquire him. Since there’s a demand for dogs to do all kinds of work for the military and other governmental agencies, there’s obviously a temptation to rush through the process. Using negative reinforcement may seem like one of those ways to get a dog to market sooner, but in the long run, I’ve seen that backfire. The same is true of not fully understanding some of the developmental stages that a dog goes through. Raising dogs and raising children are similar in some ways, but there are a couple of crucial differences. From birth to eighteen months, a dog goes through a rapid period of mental and physical growth, from helpless “infant” to an animal possessing the vast majority of its adult capabilities. With kids, recent studies have shown that they don’t reach full mental maturity until much later than we thought—into their mid-to-late twenties. Those are two very different time frames.
The window of opportunity that you have with a dog is much smaller, obviously, but that’s where the patience has to come in. Getting a dog and immediately thrusting it too soon into the kinds of specialized training we do can overwhelm it. Letting the dog acclimate to a new environment is essential.
Keep in mind that environment is so critical in training and in the field when these dogs are deployed. Exposure to various environments, and not just climate or topography, but indoors, outdoors, on stairs, in helicopters, on narrow walkways, in the dark, and so on, is something that I spend a lot of time on when I’m preparing pups for training. It’s not something I want to have to do when I acquire a candidate dog for SEAL training. That’s not to say that any good trainer couldn’t take an eighteen-month-old dog, or older, and eventually get the dog to be comfortable in those various environments, but it would take an awful damn long time to get to that point. And, as I stated above, you just don’t have that kind of time. That’s why I spend so much time with pups getting them over whatever apprehension they might have about unknown/strange environments. They have to associate overcoming that fear with something positive—a reward. Some of that is genetic, what we call the nerve of the dog, but dogs can also learn to set their fears aside. Both aspects—genetics and training—are required to be successful.
Detection work is very complicated, and as a result, the training has to be very segmented and very regimented. I’d say that there are as many as two hundred steps to doing it well, and as we progress through them, if we encounter any snags along the way, we’ll fall back a few steps and then work forward and hopefully past that problem area. Dogs will then have performed some of these exercises and discrete tasks with thousands of repetitions before being deployed. Because they learn through association and repetition, consistency is the key to training working dogs or pets. Anyone who owns a dog knows that his or her dog likes routine. It isn’t so much that they like it; it’s how they live their lives.
That’s why training is so time-consuming. I have to be consistent in how I approach every task. When working with dogs on their detection work, that means that I have to get a dog from his crate, get him to sit, go down on one knee, hold the dog by the back of the collar, show him the ball, rub it against his chest, and then feint a throw to let him know that it is now time for him for go to work seeking an object. All those things I did prior to that were kind of like priming the pump, letting the dog know that a reward—playing with the ball—is going to follow soon.
As a trainer, you also want to watch the response of the dogs to the anticipated bit of play that is to follow. Early on, when you’re not wanting to do much of anything to deter his enthusiasm, you get used to the idea that you’re holding a ball, and involuntarily you begin to blink, and by the time your eyelid has completed its trip, a sixty-five-pound dog has launched itself with all four paws from a standing start to your eye level, with his jaws snapping in anticipation and uncontrolled glee. I’ll get to how these dogs respond when they get to do their favorite training—bite work—in a bit, but these multi-function dogs spend the majority of their time detecting odors while working in the field. Because of the demands of being able to detect multiple odors, a large part of their specialized training time is spent on this skill.
The object of this training is to take full advantage of a dog’s innate ability to detect odors far better than a human can. The estimates vary, but a frequently cited estimation of their superiority over us claims that a dog’s sense of smell is a thousand times more sensitive than that of humans. Part of the reason for this is that dogs have more than 220 million olfactory receptors in their noses, while humans have only 5 million.1 In 1999, researchers at Auburn University’s Institute for Biological Detection Systems conducted experiments to determine how those canine receptors work and at what threshold (in parts per million) of an odor a dog can still detect a specific odor when a target odor is intermixed with other odors. Their intent, essentially, was to determine how little of a substance needs to be present before a dog can no longer detect it. According to their report, “The dog’s limit of detection (absolute threshold) has been determined for four compounds to date. These sensitivities are in the tens of parts per billion (ppb) for methyl benzoate, cyclohexanone, and nitroglycerin. The sensitivity to DMNB, a detection taggant, is much greater at 500 parts per trillion (ppt). There is no reason to believe that these thresholds are not representative of thresholds for other compounds.”2 In other words, the dogs could detect these explosive compounds when only 10 million or so parts per billion were present. They could also detect another substance that is used in the making of explosive compounds (a detection taggant), which many governments, including our own, require be placed there as tags or identifiers to “label” the active ingredients of the explosive. These taggants are volatile chemicals that are released into the air and are designed to essentially “serialize” certain compounds our government wants to keep track of. Because they are designed to be detected, those taggants can be present in even smaller quantities (500 parts per trillion) than the chemicals that make up the actual explosive.
Okay, that’s the science lesson for the day. The important thing to remember is that we’re talking about really, really, really small quantities of those chemicals being present in order for a dog to detect them. When you get into the difference between millions and billions, you know that you’re talking about, in this case, incredibly small quantities.
That Auburn study also went on to talk about the need to not train dogs in terms of the weight of the explosives being hidden for them to detect. What’s important is that the chemicals give off molecules of odor and that the farther the dog is from that hidden cache, the fewer of those molecules are available. Because of their keen sensitivity, they will hit on that small number of molecules and then move toward the area of greatest concentration—the explosives themselves. That’s pretty common sense, but we use that notion of the odor cone all the time in creating detection scenarios. More on that in a bit.
The main question is, How do you train a dog to detect anything? Exposure (or association) and repetition come into play again here. This time, though, you don’t need thousands of repetitions. In my experience, exposing a dog to a specific chemical signature for, let’s say, TNT, to get them to recognize it and become familiar with it and to be able to detect that specific scent only takes a handful of times.
Think of it this way: how many times did you have to be exposed to skunk odor before you could positively identify that smell?
To
do this kind of initial detection exposure, trainers use lots of methods, but I employ one that I refer to simply as retrieving. We don’t want dogs to randomly detect any odor, obviously, so we have to get the dogs to respond to specific odors/chemical signatures. Most of those chemical signatures we want them to detect are for substances which they normally wouldn’t encounter and haven’t encountered at any time in their lives prior to their training.
The obvious point about odor detection is that you have to introduce the dogs to that chemical signature somehow. The medium you use to expose the dog to that odor is essentially irrelevant. You can enclose it in a PVC pipe with holes drilled in it, boxes, towels, or anything else that will retain that small bit of the explosive material. I use rags, bits of terry-cloth towel, but I’m extremely careful with how I use them—not because they could explode, but because they can become easily contaminated. I wear latex or nitrile gloves whenever I’m handling the samples to avoid my odor signature from getting mixed in with the explosive material. I’ll place a small amount (several pounds) of explosive in a plastic box along with a quantity of towels. After about a month, the explosive odor has worked its way into the fabric. At that point, I can then separate the towels and seal them individually in an airtight container (think Tupperware) and label it.
Early in the exposure regimen to imprint an odor, I won’t be too concerned about the environment in which I do the work. That will come later. I first take a dog out, along with his handler, and we work on the simplest level. I throw the towel, and the dog retrieves it. His reward, when he brings the rag back to his handler, is that he gets to play tug for a little bit with the towel. We repeat those retrievals fifteen or so times, and by the end, the dog is familiar with that odor. I use a new towel each time to make certain that the dog is fixed on that odor and not the combination of dirt, grass, spit, the handler’s odor, or anything else. It takes the dog about ten minutes to become imprinted on the odor, far less time than it takes me to do all the prep work.
Trident K9 Warriors: My Tale From the Training Ground to the Battlefield With Elite Navy SEAL Canines Page 8
