Wednesday, July 8, 2009

Some Myths About Behaviorism

I've been revamping my website, and in the course of upgrading one of the articles in my list of the Top Ten Myths about dogs, I came up with some interesting, and I think, valid criticisms of one of the current trends in dog training.

Some Myths About Behaviorism
Dog trainer and behavioral expert Patricia 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."

Of course what McConnell means is that when an animal of any kind finds that a behavior produces positive results, it will have a tendency to choose that behavior over and over again. And that's true. But the implication is that there is only one type of training that works for all dogs (i.e., the "cookie-cutter" approach), and that all training should be based strictly on giving a dog rewards for good behavior.

What's wrong with using rewards?

Nothing. But for most behavioral science-oriented trainers that usually means food, partially because the foundation of behavioral science is built almost exclusively on the behaviors of albino rats locked inside Skinner boxes, and partially because it's usually the easiest and quickest way to get a dog's attention. As for those rats, their only motivation for learning to press a lever was supposedly to get a food pellet. But dogs aren't rats. Plus we don't normally train them inside boxes in a research lab.

"Yes," positive trainers would argue, "But whether the incentive in dog training is a treat or being given a ball to chase, it still boils down to one thing: positive reinforcement."

I agree. But as soon as we get locked-in to the idea that the linear, cut-and-dried precepts of behavioral science can show us all the answers, we don't keep our minds open to other possibilities. And far too many trainers these days consider food to be a universal reward. And that tiny little flaw in thinking keeps some dogs from ever being fully trained. If you're a dog owner who's been to a +R trainer and you tried to follow the protocols they gave you but found they didn't work, what was the first thing they said in reply? Probably: "Up the value of your treats!" (I had a client who complained, "What am I supposed to do, slaughter a cow and take the carcass with me on our walks?")

It's also instructive to understand that behavioral science techniques are notoriously ineffective when it comes to curing serious behavioral problems. The best proof of this is The Dog Who Loved too Much by Nicholas Dodman, even though Dodman didn't consciously write the book as a critique of behavioral science but as a justification for using drugs. But if behavioral science techniques were really effective we wouldn't need drugs except where there's a definite physiological cause of the behavior.

It's also interesting that Patricia McConnell has a much better success rate in solving behavioral problems than Dodman does. Much better. I think there are probably two reasons for this: 1) McConnell genuinely loves dogs while Dodman reportedly doesn't even like them, and 2) McConnell's protocols for solving behavioral problems also include teaching obedience skills, Dodman's doesn't. (Since all obedience behaviors are based on the predatory motor patterns of wild wolves, and since the prey drive is the key organizing force behind all canine social behavior, it stands to reason that teaching obedience skills will have some positive effect on bringing a dog's emotions back into alignment with his owner's needs.)

Going back to Skinner, I think we need to consider that when he proposed his theories it was widely believed that animals didn't even have emotional lives. With some of the recent advances in neuroscience, and the discovery of the same emotional circuits that exist in both the non-human and human brain, we now know that animals can be very emotional. This is especially true of dogs. Yet the behavioral science approach is based almost exclusively on changing a dog's behavior with little thought given to the underlying emotional cause of that behavior.

Some in the field would disagree. They would say that they're very conscious of how emotions affect behavior. I have no doubt that that's true. But the techniques they use are still based on a clinical, unemotional, Skinnerian foundation, one that's simply not geared to change a dog's emotions as much as it is to change his behavior. That kind of thinking is built in to the system despite the fact that all behavior, learned or instinctive, is the end product of emotion. In fact without emotion there would be no such thing as positive reinforcement. This is not something that factored in to Skinner's equations at the time he made them. It should be factored in now, but from my observations that rarely happens.

Meanwhile in Natural Dog Training our focus is always on changing the dog's emotional state first because we know once we do that and bring the dog's emotions back into balance, the right behavior will always follow.

It was also believed during Skinner's time that the foundation of all animal behavior was geared specifically around the survival instinct, so when his rats pressed the lever and "learned" to make food appear, it made sense that their only impetus for doing so was based on their own survival: food is necessary for survival, therefore food is a primary reinforcer. But with the current trend in science to find and understand the roots of "biological altruism," the tendency in social animals (and even in some non-social species) to give up what's in one's own "self" interest in order to help another animal in need, the primacy of the survival instinct is starting to seem a bit mothworn if not badly outdated. Biological altruism is a huge puzzle because it implies that a very important aspect of Darwinism (and one that has a domino effect on behaviorism as well), may not, in fact, be what it seems.

Strangely enough, the clearest window into this puzzle (or perhaps not so strangely) is the domesticated dog. No species is more famous for its ability, let alone its outright unstoppable zeal for putting its own survival on the line in order to help those it loves. In the past few months alone (I'm writing this in July of 2009), there have been two videotaped incidents of dogs dashing into traffic in order to rescue a fallen comrade, one was on a freeway in Chile and other on the Major Deegan Expressway in the Bronx.

This brings up another point about the difference between Pavlov's and Skinner's era and ours (i.e., the early 20th Century v. the early 21st Century). Back in Skinner's day it was believed that all animals were vying for dominance within their own habitats as well as within their own social groups. And just like the beliefs about the survival instinct which accompanied and most probably engendered this Darwinian idea, the underlying principle was that animals always put their own "self" interest above all else. And that's simply not true. It's especially not true in dogs, and it turns out that it's not even true in wolves. And yet every single dog trainer who espouses behavioral science as the bedrock of all animal learning is still operating under this false premise. They don't accept the fact that sometimes the survival instinct simply isn't operational, which means that sometimes a primary reinforcer is not only not primary it's not a reinforcer. And yet we're told time and again: "Up the value of your treats!" (I'm not saying that Darwin's theory is wrong, or that evolution isn't a real process, or that treats aren't valuable in dog training, nor am I saying that the survival instinct isn't an important part of the evolutionary process, just that it's not as important as we once thought it was, which again has a domino effect on Skinner's theories about behavior.)

One of the clearest examples we see of a dog's ability to routinely override its own survival instinct is in dogs who do search-and-rescue work. With the recent explosion of interest in behavioral science some search-and- rescue dogs have been trained exclusively with food and clickers. There was great hope in certain quarters that this would be the dawn of a new age of perfectly conditioned working dogs. But in the end most of these dogs have proven unreliable, especially when forced to work for long hours, because they’ll often indicate a false positive just to get a treat.

"Dogs want rewards," says Dr. Lawrence J. Myers, of Auburn University's College of Veterinary Medicine, "So they will give false alerts to get them."

Giving a false alert is something a dog trained through his prey drive would never do; he wouldn't know how. He'll work for hours and won't quit until he finds exactly what he's looking for. Why? Because he's focused on hunting, not on getting an external reward. The only problem they had with the search-and-rescue dogs at Ground Zero was making them stop to rest. Those amazing animals would have kept working until they found a survivor or a body or just dropped dead themselves. Is that courage? Is that altruism? Or is that just the way dogs are?

Kevin Behan made a very insightful comment on his blog recently. (If you don't know this, Kevin spent a major portion of his career training police dogs and detection dogs, using their prey drive - not food rewards - as the focal point of learning.)

"Search-and-rescue dogs can search disaster sites whereas no other animal can be conditioned to do so, which is especially revealing since cats and monkeys are far better adapted, physically speaking, for such work. One can acclimate a police dog to love running up a metal fire escape with someone throwing metal pots and pans down at it. All these so-called negatives ... arouse [the dog's] prey-making urge to an even greater pitch."

Can you imagine what those pots and pans would do to cats and monkeys? Is it even remotely possible that they could be trained to run up a fire escape while you're throwing loud, clattering objects at them? Even if they did make it to the top, my bet is that their first priority after getting there would be to find a safe place to hide and not come out for days.

Monkeys clearly have more mental agility than dogs. And to a large extent, so do cats. And Kevin's insight is, as usual, dead-on; both species are also more physically agile when put in the kinds of situations that most search-and-rescue dogs find themselves in. So if learning is only about reinforcing the behaviors you want from an animal, and monkeys and cats are smarter and more physically capable of working in and around disaster sites, why can't they be conditioned to do it?

Because they don't want to. Dogs, on the other hand, live for this kind of stuff. Talk about treats, they eat this stuff up. When dogs are trained properly, through their prey drive, they're absolutely driven to find survivors at a disaster site, or to sit and stay and come when called, or to do whatever else you want them to. They'll do it: no questions asked, no treats expected. You can't condition that kind of willingness into a cat or monkey just as you can't condition it out of a dog.

If we apply this lesson to flatworms and philosophy professors we can see that Patricia McConnell's idea really is off, particularly since she's a dog trainer herself, and particularly since the quote in question came from a piece she wrote for a magazine devoted exclusively to canines.

Again, I'm not saying that conditioning isn't a valid form of learning. It is. It has its place; there's no question about that. But in some cases, at least where dogs are concerned, there may be a much better alternative. You simply have to open your mind a little to see it.

And no, all animals don't learn the same way. Dogs are different. And it's their very difference that can help us see some of the cracks in the foundations of behavioral science.  

LCK
"Changing the World, One Dog at a Time"

9 comments:

Karen said...

Lee, as always I love your thoughts about dogs and their training... (although I AM a Cesar Milan fan as well...) Everything you say just makes so much sense. I'm on a constant search of how to help my dogs... Thanks!

Lee Charles Kelley, said...

You have to be careful about following Millan's example. Much of the stuff he does with dogs only works temporarily because it essentially puts the lid on a dog's energy. That kind of pressure builds and either explodes at a critical moment or else it quietly eats away at the dog's biological system. Repressed emotions cause cancer and all sorts of other health problems.

Please be careful what you take away from CM and what you leave behind.

LCK

Teri said...

I find a lot of what you say very interesting with some great ideas but I admit to being confused as to why you would think that fulfilling the dog's prey instinct with tug toys, playing etc. is really all that much different from the premise of +R training? I totally agree that using food treats has become waaayyy overused and in fact my own dog (very prey driven black lab who is not very food motivated) improved his skills tremendously when I moved from food rewards to playing tug as a reward. But really aren't they both rewards/motivations for presenting the behaviours that we want?

I really think we all need to throw out the cookie cutter, one method works better then all others and move towards a balance of methods (and yes, I think that sometimes can include some "negative" methods) that work for each individual dog. My lab is prey driven and very energetic. The same can be said for my mom's dog who is a full littermate to mine. Yet each dog needs to be handled quite a bit differently for him to reach his best "self" (for lack of a better word).

PS - I have great respect for Patricia McConnell and think she has some fabulous insights into training, loving and living with our furry friends.

Skibum said...

Lee, Like most things in life one solution doesn’t work for all. I have a challenging Portuguese water dog and am looking for new things to try and approaches to take to work on his aggressive behaviour (very protective of the door and fence line in backyard). Being the “pack leader” or dominant (have ever you look at it) is having very limited success. I have been working on positive rewards and it has some good results in getting his aggressive behaviour under control on the walks. Still reacts badly to some dogs we pass.

I am taking some good things from the information you have posted. With 3 dogs in the house now is it ok to have community play time or do we need to do this one on one with the dogs (which will be a challenge). The other 2 dogs are 20 lbs vs the 60 lb PWD. Fact is that Cosmo is incredibility tolerant of his little brother. Cosmo is 3, Luke is 1 and our latest addition to the pack (my daughter’s dog) is 2. The smaller ones are both Cockapoos and both want to tug on the walks which I don’t want. I have tried a gentle leader on both Cosmo and Luke with good success although neither dog is really fond of wearing them.

Thanks for the info and insight. I plan to keep reading the info here and see what more I can learn and try.

Lee Charles Kelley, said...

Teri wrote: "I find a lot of what you say very interesting with some great ideas but I admit to being confused as to why you would think that fulfilling the dog's prey instinct with tug toys, playing etc. is really all that much different from the premise of +R training?"

There are two pieces to this.

1) Training

On a certain, surface level you're right. One of the problems though is that on the most basic, and most important level, neither a tug toy nor a liver treat is a reinforcer. The positive reinforcement for all behavior and learning comes through changes in the dog's energy state via release of internal tension. Things like treats and praise and petting and tennis balls are only means to achieve that end. But essentially the best, most complete forms of learning come through whatever is most effective at releasing the dog's tension at any given moment.

Remember, dogs have only had a few hundred years of being house pets. Before that they worked for a living, using their prey drive. Before that, they had to work for a living because they weren't dogs. And the prey drive in canines works very specifically on principles of tension and release.

Hunting large prey animals is risky, so the wolf's (or coyote's) survival instinct overrides the hunting instinct until the tension and pressure builds to the point that they have to leave the safety of the den to go out and hunt. Nature has designed things so that the ultimate tension release for canines comes through biting a prey object.

So when +R trainers believe there's no difference between food and a tennis ball, they're missing the most important element of learning for dogs: releasing internal tension by biting a toy.

This brings us back to search-and-rescue dogs. If there were no difference between food and tug toys in terms of their place in the general category of "reinforcer" why can't working dogs be trained successfully with treats and clickers alone? Why can't monkeys and cats be trained to search disaster sites?

The 2nd part of the problem with behavioral science techniques doesn't really relate to your question because it has to do with solving behavioral problems. But in the article I did make the point that McConnell is better at this than Dodman because she incorporates obedience skills -- which are predatory motor patterns -- into her protocols. So once again, her effectiveness is at least partially related to using the prey drive, albeit unconsciously.

I hope this helps!

Lee Charles Kelley, said...

Skibum wrote: "I plan to keep reading the info here and see what more I can learn and try."

Neil Sattin's dog Nola had some behavioral problems similar to what you're experiencing with your Portagee. If I were you I'd check out some of his posts on aggression, and on playing tug.

Have you tried the pushing exercise?

If not read "An Open Letter to New York Dog Trainers," and "How to Do the Pushing Exercise," here on my blog.

Also Neil's DVDs are now available. They could be tremendously helpful. Go to Neil's blog, or to my website, to read more and to order them if you want.

As for walking 3 dogs together, you kind of have to train each one separately to walk next to you. One thing that might also work is to constantly talk to them and praise them as you're walking. Occasionally throw in a game of chase or an about turn to keep their energy focused on you.

Best of luck,
LCK

boywunder said...

I just want to hop in here real quick. I have a dog named Roxy who is aggressive towards other dogs, and fearful of new people. I've tried both the Cesar Millan dominance stuff, which is how I was taught to work with dogs based on the first trainer I worked with. And then moved on to positive reinforcement when Roxy was becoming more aggressive as a result of the dominance work.

Well, I've been working with her now for the past 3 years. And you know what? 3 days with Kevin Behan using these methods that Lee is talking about and she's already showing amazing shifts. All because he's changing her emotional state. Not because he's rewarding her for good or desirable behavior, or correcting her for undesirable behavior a la Cesar.

To give a specific example of how holistic this process is, Roxy has always been whiny and "anxious" in the car. As soon as she hops in, the whining starts, and the excitability would ensue. And it wouldn't stop until the car would get up to highway speed. Well, we had been working with Kevin for 2 days, and then decided to take a drive to explore Vermont with our other 2 dogs too. So we get in the van, and not a peep from Roxy. Her body was SO soft and relaxed, and she was so calm and settled. She just sat in her seat for the entire trip, which was in total about 8 hours. Then, when we got back at night, the dogs in the barn started barking, and our 2 other dogs started barking back at them. But Roxy didn't make a peep. She just sat there, calmly observing what was going on around her.

I share this because when you work with changing the dog's emotions, other problems seem to go away. You're not treating a symptom or a specific behavior with this method. You're treating the whole dog. Dogs already know what to do, but we corrupt them. And the more we try to "correct" the individual problems, the more problems we create. We didn't work with Roxy's specific behavior in the car. But that behavior went away as a result of working with her deeper issues.

I know it's really hard to distinguish between the idea of using positive reinforcement or rewards such as toys vs engaging a dog's drive and then letting him bite something. But there is a huge difference. Seems like semantics. But believe me, after what I've been seeing and experiencing firsthand on Kevin's farm, the difference couldn't be greater.

Margot said...

Well said boywonder. I often try to explain the pushing exercise to people but it is tough. I say that it seems you are working on something completely unrelated but the issue you are trying to resolve just gets fixed. Your words are much better, will use.

Margot said...

Progress report: the dog whos owner I have suggested do the pushing exercise is already much calmer after 3 times in 5 or 6 days. She had some excessive affection happening and would grab your arm with her mouth to pull you to her. Yesterday when I visitted was the first time she just sat and looked at me with happiness with no instruction from her owner. After a short while she nosed my hand to get me to pat her.

Its great to see how it works every time.