Friday, February 8, 2008

Of Philosphers and Flatworms

This post is part of a blog carnival hosted by Neil Sattin. It was originally posted on Feb. 8, 2008. It was modified and re-posted later that week, on Feb. 15.

It’s In His Blood, It’s In His Heart:
Why the Wolf Model Is Still Relevant to Dog Training

I'm not a fan of the dominance or pack leader approach to dog training. I've been writing for years about its potentially harmful and unscientific nature. But in the past six or seven years, many in the positive training movement have been denying or denigrating the importance of wolf behavior when it comes to understanding dogs. On a certain level, this makes sense; there is a wide gulf, behaviorally speaking, between them. And yet the two also share a very long evolutionary history. And some of the traits, unique to wolves—hunting large, dangerous prey by working as a cohesive hunting unit, sublimating their urge to bite into ritualized social behaviorsare what enabled dogs to domesticate us (or vice versa), and helped them morph into becoming the most diverse and socially adaptable species on earth.

Do All Animals Learn the Same Way?

Behaviorist and positive training maven Patricia McConnell (who actually holds a degree in anthropology, not behavioral science) is among those denying the importance of the wolf model when working with dogs. (This despite the irony that one of her booklets is titled, How to Be Your Dog's Pack Leader!) She, and many others in the positive movement, have been putting out a continuous public relations campaign to try to steer people away from dominance techniques, and convince them to use positive reinforcement instead. That's a fine goal, but in my view, they're being deliberately dishonest about it. 

For instance, McConnell wrote in Bark Magazine not too long ago, “The process of learning is pretty much the same whether you’re a pigeon, a planarian [flatworm] or, come to think of it, a philosophy professor.”

Doesn’t that strike you as just a little bit off?

Of course, what McConnell means is that when an organism of any kind finds that certain behaviors produce positive consequences, that organism will have a tendency to choose those behaviors again and again (which is true—kind of).

The Torturous Origins of "Positive Training"
The idea that animals learn through positive consequences comes from experiments done by B.F. Skinner in the 1930s, where he half-starved some rats and pigeons, locked them inside boxes, then waited to see if they’d learn to press or peck a bar in order to obtain a food pellet. Eventually most of them pressed or pecked the bar, found out that doing so got them goodies, and began doing it again and again with the presumed expectation of getting more goodies.

Skinner called this “operant conditioning” and expanded on it by providing the animals with a “discriminate stimulus,” meaning the food would only be given when a light was flashing, for example, which is akin, on a very remote level, to linking a pup’s behavior to a verbal cue. (Remember, you’re not training your puppy while he’s locked in a box, in a controlled laboratory setting; you’re doing it in the real world with thousands of variables you’re unaware of, and your dog probably is).

Skinner was quite pleased with himself following one experiment where, after he’d stopped supplying food pellets to some pigeons, to see how long it took before they’d stop pecking the bar, even though the light was still flashing. The poor birds wore their beaks down to nubs. Skinner never achieved that same level of “learning” with rats, because for birds, pecking is a predictable fixed-action pattern (or modal-action pattern), related to food; it’s more or less stamped into their DNA. Rats, meanwhile, don’t have an instinctive “bar-pressing” behavior.

Another development came when Keller Breland, who studied under and later worked with Skinner, did an experiment with a group of animals who weren’t locked inside Skinner boxes. They were free to roam around a large, barn-like structure. He conditioned them to learn a simple behavior that gained them a food reward. And most, if not all, of the animals quickly learned to do whatever behavior they were taught whenever they were hungry. After a few days, though, a funny thing happened: the raccoons began “washing” their hands, the chickens began pecking at the floor, and the pigs began rooting around in the dirt. They all stopped producing the conditioned behavior in favor of their own food-related, fixed-action patterns, even though those behaviors weren’t rewarded. The real kicker is that the experiment with the pigs had to be stopped or they would have starved to death! As a result Breland said, “There are definite weaknesses in the philosophy underlying these techniques,” and suggested that animal trainers be on the lookout for what he called “instinctive drift.” (“The Misbehavior of Organisms,” American Psychologist, 1961.)

If we apply this lesson to philosophers and flatworms, we can see that McConnell’s idea really is off. Different species have different evolutionary histories, different morphologies, different develop- mental processes, different environmental stressors, thus different, fairly predictable predispositions to certain fixed-action patterns.

So no, they don’t always learn the same way.

They Dont Give These Shows to Chimps!
It seems to me that the mindset that gives us McConnell’s logic, also tells us that the wolf model no longer applies to dog training. For instance, positive training guru Ian Dunbar says that since humans share roughly the same amount of DNA (98.6%) with chimps as dogs do with wolves, then, logically speaking, trying to train dogs by studying wolf behavior is like learning how to raise a child by watching chimps, to “see how they do it.” This despite the fact that a mere 12,000 - 120,000 years of evolution separates dogs and wolves, while 6.5 million years separates us from chimps. And even when you parse that comparison down by the numbers of generations rather than the number of years, there’s still a significant difference. And here’s something even more disingenuous on Dunbar’s part: we don’t, in fact share 98.6% of our DNA with chimps; we share 98.6% of our nucleotide sequence. And as cognitive scientist Daniel Povinelli, of the University of Louisiana, puts it: New research has shown that rough similarity in our nucleotide sequences obscures the fact that the same genes may have dramatically different activity levels in the two species. So even where humans and chimpanzees share genes in common, it turns out that there are what can only be described as major differences in gene expression.” In other words chimps and humans aren’t anywhere near as alike as Dunbar would have us believe, nor even remotely as alike as dogs and wolves actually are to each other.

Meanwhile Dunbar’s analogy also crumbles when we consider that by some scientific forms of reckoning dogs are actually a sub-species of the wolf (canis lupus, canis lupus famliaris), while chimps and humans (pan troglodytes, homo sapiens) aren’t even in the same family. For instance, although I dislike Jay Leno*, I watched a wildlife segment the other night, just to see a gray wolf on The Tonight Show.

It was shocking to behold this gorgeous, majestic animal. We’ve all seen cheetahs and alligators and grizzlies on talk shows before. And they look dangerous and exotic and scary. But what was so shocking about seeing a wolf climbing into the chair next to Jay’s (technically Johnny’s) desk was that he looked very wild and yet very much like a cross between a big sweet German shepherd and a giant malamute. Leno’s wildlife expert even warned people not to try petting a wolf, if they should ever see one up close (presumably in a sanctuary; it’s doubtful you’d get a chance in the wild). 

The point is, we feel awed by something so wild and dangerous as a grizzly bear. The hairs on the back of our neck stand up when we see one in such close proximity to a human being. But while that same wild, dangerous energy is present in a wolf’s appearance and movement, he also looks so familiar and comfortable, like you actually could go up and pet him or kiss him on the nose. I think that’s what was so shocking about the wolf I saw on Leno.

Granted, you’d never mistake a pug or a dachshund for a wolf. But as Letterman likes to say, “They don't give these shows to chimps!” And no matter how much closer Jay Leno, for example, is to a chimp in both appearance and intelligence, than Letterman is (or Johnny Carson ever was), there’s still no danger of mistaking him, or any other talk show host, for a chimp the way there is of mistaking a wolf for a dog.

(I’ve gone off topic, but I did say that I dislike Jay Leno, right?)

The Wrong Model
Personally, despite what I perceive as Dunbar’s intellectual dishonesty, I agree, at least partially, with his point: that it’s perhaps unwise to try to copy wolf behavior when training our dogs, particularly when most of the behaviors we’re told to copy—the alpha roll, or being the pack leader—don’t actually exist in nature. (See “Is Your Dog Dominant, or Just Feeling Anxious?”)

Yes, some traditional trainers (like Cesar Millan and the Monks of New Skete) are still locked into the mistaken idea that dogs “think” they’re part of a hierarchy, and need an alpha wolf to control them. But that’s an outdated, and wrong-headed model. Hierarchical behaviors are only seen in captive wolves, and village dogs, etc., animals living under stress. Wild wolves don’t have pack leaders, per se, or form hierarchies; they’re more harmonious, less at each other’s throats, which is almost entirely due to the way they hunt together—something captive wolves are unable to do. This is true even at Wolf Park, where although the wolves are given an opportunity to “humanely hunt” buffalo (they’re allowed to chase them but never get a chance to bite and kill). As a result, even those wolves are often antagonistic to one another (i.e., they form what appear to be hierarchies based on captivity stress). Thats because they never get that final payoff through their teeth and jaws. Thats the ultimate stress reducer. And its why most dog owners have baskets full of bones and chew toys.

Who Knows, Maybe Dogs Domesticated Us...
In recent years a new theory about how dogs became domesticated has arisen, suggesting that dogs are not predators at all, that they became domesticated because they scavenged at human encampments, and somehow, through this kind of rat-like behavior of eating our shit and garbage, they somehow wormed their way into our hearts. Sounds lovely, right? (And a bit unlikely if you ask me.) It also denies a few simple questions that almost every dog owner inevitably asks: “Why does my puppy shake his head around when he has a toy in his mouth?” or “Why does my puppy chase leaves when the wind blows, or run after anything that moves?” or “Why does my puppy stalk the cat?”

The answer is that dogs are really predators at heart, and the heart of the puppy is the clearest window into that predatory nature. Think about it: a puppy is attracted to everything in the world through his teeth. He seems utterly driven to grab, bite, nibble, mouth, and chew everything he can. Do kittens do that? Gerbils? Nope, just dogs.

It’s true that wolves are generalists. They don’t just hunt large prey, they’ll also scavenge if necessary. So it’s not much of a stretch, I suppose, to think that it was only this aspect of the wolf’s nature that created the dog/human bond countless years ago. But I like to think that instead of individual wolves being attracted to us through our garbage, early man was attracted to wolves because of the way they hunted. After all, we were social animals, they were social animals. It’s a pretty good bet that we identified with them on some level. (The animal lore of many Native America tribes tell us that there’s a very strong likelihood that this is, indeed, true.) Even today dog owners form strong feelings of identification with their dogs.

There’s another element to this, by the way, which is that in most predator families the young animals are kicked out of the group once they reach adolescence; they’re not allowed to stick around with mommy once they’re big enough to take care of themselves. But wolf offspring don’t go out on their own until they’re at least 2 years old. If we look at the way wolves continue to nurture and take care of their young, we have to wonder why this is. What is the adaptive purpose, if any for this difference? It’s dangerous for most predators to try to live together; they have a tendency to attack one another. And what was the actual mechanism for the evolution of this continued nurturing behavior in wolf families?

I think the answer is simple: oxytocin, the nurturing hormone. I think i
t’s quite probable that at some point in time wolf pups kept producing this nonopeptide long after cougar and jaguar cubs did. This hormone, which also acts as a pheromone, is said to create feelings of trust and even love in others. And if our human ancestors were in an environmental niche where they had close contact with wolves, they would’ve been affected by it as well, which would also explain why we became so attracted to wolves intitially, and invited then into our campfires, etc.

So my theory is that in some long ago ecological niche, before humans thought themselves superior and separate from other animals, we were probably sharing a habitat somewhere with a group of wolves, whose young came equipped with strong doses of oxtytocin. This would’ve made us feel trusting of them, plus it would’ve given us an opportunity to observe the way they hunted together. If we were still struggling with the idea of how to create weapons to assist us in our need to hunt, and saw how wolves managed to kill large prey without them, and if we saw how successful they were at killing the kind of big animal with lots of meat on its bones that made our mouths water, the kind of animal we would have been hesitant to hunt on our own for fear of being knocked senseless by its hooves or gored by its horns, and if we were hungry, and out hunting rabbits one day, and saw how successful these wolves were at killing something that could feed our families for a week, there might’ve been a pre-historic light bulb that popped over our heads saying, “Oh, so that’s how you do it. You work as a team!”

It might have even taken us a few generations of letting the wolves do the work and then scavenging from their kill site. If so then the human/wolf dynamic as proposed by the current “dogs as scavengers theory,” would be totally reversed. We may have very well let the wolves do the killing, and even let them have the organ meat; we just wanted to scavenge that fresh, juicy muscle tissue (“Mmmm, juicy muscle tissue...”). Another possibility is that we might’ve even helped with the hunt. Being taller, with long arms what we could wave around in the air, we could’ve been very usefull at scaring the prey animals into running, one thing that’s sometimes difficult for wolves to accomplish. So we could’ve very well assisted them, rather than what was proposed in the old theory, that we taught them to hunt for us. But if the symbiotic relationship started because oxytocin made us trust them, which enabled us to we recognize that they had superior hunting skills and were primarily interested in organ meats, well, once they were through eating, we would’ve politely scared them off so we could take the rest. And they probably wouldn’t have minded too much. After all, that’s how symbiotic relationships operate.

So my theory is that we didn’t domesticate wolves, they domesticated us. (Though it was probably a two-way street, I like putting it the way I have because of how my own dog, Freddie, had such a domesticating influence on my life.)

Why the Prey Drive Is Important in Training
What’s so special about the prey drive that makes it important in dog training? Well, imagine that you’re a cheetah hunting an antelope. It takes a lot of focus and energy to take down your prey, plus an enormous amount of emotional flexibility. You have to be able to adjust your movements, your emotions, and your level of energy instantly whenever the prey animal changes course or the terrain goes from open plain to a riverbed or a stand of trees, etc. And when you get in close enough for the kill you also have to be extremely aware of the prey animal’s hooves and horns.

Now imagine that you’re a wolf hunting a deer. You can’t do it alone so you’ve got some of your buddies along with you. Unlike the cheetah you don’t have enormous muscular power in your shoulders and haunches, you don’t have claws, and even though you’ve got sharp teeth, your jaws aren’t as powerful as a big cat’s. A cheetah doesn’t need help with his prey, but for if you’re a wolf it’s very unlikely that you could kill such an animal on your own. (It’s been done, but only rarely.) So even one-on-one with the deer, you’re at such a disadvantage that your ability to adjust your movements and emotions, your focus and your energy, while hunting has to be double that of the cheetah if youre to succeed. Now add the fact you’re hunting as a pack, which means you also have to focus on what everyone else is doing. The pack can’t succeed if no one is paying attention to how the others are behaving. You have to be able to adjust your movements, emotions, your focus and energy by another factor for each additional pack member involved. In terms of the level of energy exchange taking place and the emotional flexibility necessary to succeed, it could require ten times more than when a cheetah or other big cat is hunting its prey.

There’s another factor, too. Kevin Behan explains it this way in his groundbreaking book, Natural Dog Training:

“When I talk about flexibility, I don’t just mean the individual’s ability to react to change; I mean that all the members adjust to change as a group. This kind of collective coordination is the bedrock of sociability. Normally this might be thought to fall more under the realm of communication, learning, and intelligence than instinct. My premise, though, is that the prey [drive] coordinates behavior and controls the learning process. It exerts an influence that exaggerates slight differences in each individual’s temperament into gross differences of behavior, thereby producing the phenomenon of specialization. As an individual learns one role in the hunt, indirectly he’s halfway to learning another. Each job is not so much a skill as a different emotional state of uninhibited-ness. In such a flexible system of learning, where each job is emotionally linked to another, there can be social migration through ‘ranks,’ both upward and downward, as the emotional environment of the group adapts to retain the overall balance and synchronization. So while learning is dynamic and responsive to outside elements, its also predetermined.”

So viewed from Kevin’s perspective, the prey drive is vitally important in creating emotional and social flexibility in wolves. And if dogs share a genetic history with wolves, this should apply to them as well.

If you still believe the prey drive isn’t important in training, think of the way a dog behaves when he sits for a treat. He wags his tail and certainly acts happy, but his primary motivation seems mainly just to get the treat. Correct? How about a dog who obeys a command because he’s learned to “submit” to his owner. In that case you can tell fairly easily that his heart really isn’t in it. He’s primarily focused, at least more often than not, on avoiding a negative experience. Now think of the way a dog behaves when you take him to the park and he finds another dog there to play with. That’s pure joy. And think of the way your dog greets you at the door, perhaps with a toy in his mouth. He’s nuts about you. Or maybe there’s a dog you know who’s been trained with a tennis ball or Frisbee as his primary motivation and reward. In all three instances, as you look at each dog’s behavior, you can tell his heart is fully involved in what he does. That’s how obedience should be taught, in a way that energizes the dog and uses his whole heart.

While the dominance model of training seems to work for some people, and the Skinnerian model works for others, I think it’s important to look at this from a different angle. And despite the (yes) partial truth to Patricia McConnell’s statement, your dog is much more like a wolf than he is like a flatworm. I don’t know how anyone can contradict that. And contrary to Ian Dunbar’s argument, your dog is far more like a wolf than you’re like a chimp. Still, both Dunbar and McConnell are recognized experts in the field of dog training. They’re very good at what they do; they have much to offer. But the funny thing is, obedience training actually got its start as a way of duplicating some of the predatory motor patterns found in wild wolves. That’s a simple fact that McConnell and Dunbar seem unaware of: the down, the stay, and to a certain extent the heel and the recall, are all analogues to behaviors displayed by wolves while hunting.

Look, I’m not saying positive training can’t be a good thing. It’s a damn sight better than beating your puppy into “submission,” hitting her till she yelps in pain (as the Monks of New Skete once so cavalierly recommended), or stabbing him in the neck with your outstretched fingers. But has “positive” training really been “proven effective, scientifically” (as is often advertised)? Or does it sometimes work by accidentally activating a dog’s fixed-action patterns, the way Skinner did with his pigeons? I think it’s probably mostly accidental. But honestly? It could be a little of both. So why not do both? Use positive reinforcement if you like; but also use your dog’s prey drive. After all, for a dog there’s nothing more reinforcing than an opportunity to bite something in play. Besides, when you use your dog’s true instincts you won’t have to worry about “instinctive drift” or who’s alpha, etc. Your dog will simply obey you because he can’t help himself; it’s in his blood. And much more than that, it’s in his heart.
"Changing the World, One Dog at a Time"

*(Carson and Letterman went out of their way to help Jay Leno’s career, yet the first chance he got, Leno turned around and stabbed both of them in the back out of pure, naked ambition. Plus he's constantly stealing ideas from other comedians. In other words, he's the worst kind of show-business weasel. Plus, he's nowhere near as funny as he used to be.)


Annette said...

have to admit...I liked that! I am a total advocate of using what works with each individual dog and the person they are with. And I find the best trainers know this as well. Use some positive stuff, use some "encouragement" stuff. Use tricks, rewards, use whatever the dog likes and work with it. Good read.

heartdog said...

Great read. I'm vastly fascinated by dog training, and I raised my current dog, a GSD, with "positive training", and was pretty hardcore into that philosophy.It works well for teaching behaviors, just as it does for film animals, and zoo animals. But it's when I started doing Schutzhund, and utilizing prey and play along with a marker (both positive and negative), that training went to a higher level. Just using a tug, and engaging prey/bite, unlocked a much greater potential.
Using either the dominance model, or the "all positive" one rings false to me. Neither seems to tap into cooperation as much as this more natural approach.

LCK said...

Thanks. I appreciate your comments.