Tuesday, November 23, 2010

Ian Dunbar Seminar: Is Learning Theory Useful for Dog Trainers?

Ian Dunbar is not a heretic- he wants you to know that. Operant conditioning is real. It’s been proven. The experiments behind the theory were repeated thousands of times, and they were validated. It’s great science.

It’s just not useful for dog trainers. Or at least, not most of it. Ian believes that only about 10% of learning theory applies to us because learning theory was laboratory-generated. That is, the experiments were implemented, monitored and controlled by computers, carried out using Skinner boxes, and the animals used were “simple” and had “few interests.” In contrast, we are humans, and we train real dogs in the real world.

If we choose to use learning theory in training, then we must learn to train like a computer. That’s not necessarily bad- computers have some pretty good traits. They are tireless. They are completely consistent in both monitoring the trainee’s behavior and in providing feedback. This allows them to have very clear criteria. By contrast, we humans often have unrealistic or unclear criteria, and are inconsistent in our observations and feedback. So, why wouldn’t we want to train like a computer?

Well, computers as trainers have some drawbacks. They can not qualitatively assess an animal’s performance- they can’t see cute or flashy behaviors and train that into the final product. While they can give feedback, they cannot give instructive feedback. All a computer can do is say yes or no- provide a click and treat or a buzz and shock. They cannot explain why the animal was wrong or what he should do instead, and they cannot explain how important or urgent compliance is. Humans can.

Then there is the matter of reinforcement schedules. Ian identified seven reinforcement schedules: continuous, fixed interval, fixed ratio, variable interval, variable ratio, random, and differential. Ian explained that six of these seven schedules will maintain a behavior, but only one will improve behavior. The one that will improve behavior- differential reinforcement- is the one that computers cannot use. (Personally, I disagree. Computers may not be as good at it as we are, but there is no reason a computer couldn’t reward faster responses. In fact, they might be better at that than I am- I do not have a stopwatch in my head.)

Because we humans cannot be as consistent as computers, and because computers cannot provide instructive feedback the way we can, Ian sees no need for us to try to emulate computers. He finds this to be especially true because most dog owners don’t need nor want the precision that comes about from training like a computer. As a result, he really doesn’t have any use for the vast majority of learning theory.

So what does Ian like? Thorndike’s Law of Effect, which more or less says you should reward the good stuff and punish the bad stuff. Ian says this is simple, elegant, and pure. It doesn’t get into complicated and confusing types of rewards or punishment which cause endless arguments on the internet. Thorndike tells it like it is.

Again, Ian’s orientation as a trainer of pet dogs is obvious. The average dog owner doesn’t care about precision, and doesn’t have the consistency or patience needed to sort through the various quadrants and schedules, so I understand why Ian thinks we should avoid discussing learning theory with clients. We need to quit worrying about the science and terminology and just train. We should help them, not confuse them. It’s hard to argue with that.

Still, as a dog geek, I struggle with this idea. Personally, I enjoy understanding the science behind what I’m doing. Ian said it himself: learning theory is valid. I like thinking about what I’m going to do. I love planning my sessions. I also think it’s fun to take data and evaluate what I’ve done with the goal of doing better next time.

As a competitor, I want precision. I enjoy the challenge of being consistent enough to get amazing results. I strive to be as clear as possible in my criteria. In many ways, I do try to train like a computer, and I don’t think that limits me. I enjoy pairing clicks with not only treats but also heart-felt praise when my dog does something exceptional. I see no reason to have to choose between computer or human. That’s sort of the beauty of being human, after all: I can think outside of the box and combine the best of both approaches.

I know I’m not the normal dog owner. I spent Halloween weekend at Ian’s seminar, and I’m spending hours writing up my notes for this blog, after all. I would ask all of you if you’re normal dog owners, but I suspect I know the answer to that. You are reading this, after all.

Instead, go ahead and analyze what Ian said. Tell me how it makes sense, and then how it confuses you. Tell me how you use, or don’t use, learning theory with your dog. I know you’ll have lots to say!

17 comments:

Sam said...

Working in a pigeon lab where we HAVE used DR schedules, I can ABSOLUTELY say that a computer can differentially reinforce lower rates of response, faster rates of response, etc.

There is differential reinforcement of OTHER behavior (DRO) - called omission training - that Dunbar might have been referring to, where we reinforce everything EXCEPT one behavior (ie. reinforce a dog for lying down, for chewing a bone, for sitting, etc, but no reinforcement while barking). I can see how that might be a little more difficult for a computer to do, but still possible, in some instances.

I also question the idea that only differential reinforcement schedules can improve behavior.. it has been proven in the lab that variable interval and ESPECIALLY variable ratio schedules lead to a higher, more consistent response rate. So, if we're talking about improving /a/ behavior, it's certainly possible to do using those schedules. If we're talking about changing the desired behavior? Yeah, that's when DR schedules come in.

I am really big in to learning theory, as you can tell!! I can see how the average dog owner might not want to hear all of those technical things, but I think that it'd do them a lot of good if they were familiar with them. ESPECIALLY where behavior issues like fear or aggression come in - which affect pet dogs just as much as competition dogs!

Does Dr. Dunbar advocate for only non-physical punishment? I don't know much about him, and am curious.

Kristen said...

I've recently been surprised at the number of instructors who do a lot about the quandrants/learning theory to pet training students. I DEFINITELY think that most of the time they don't need it, and typically don't want it.

And it's not just that....it's a HARD concept. I've been in too many college classes that cover it...and the instructors/professors get it wrong or make it even more complicated most of the time. If THESE people can't do it, if the students trying to study it for tests have a hard time, how can our training students who want behavior change be expected to understand?

My 4-H'ers have to learn it some years. I'm able to get even the young 'uns (10-12 years!) to know it well... but I'm really upset it's in their book.

As far as trainers...yes, they should be very good at this sort of thing. That said, if they know it and can't apply it well...it's not all that helpful. And based off of what I see most of the time... not enough actually understand it well.

Excellent post!

Katherine said...

I had a hard time with this in his seminars. On the one hand, he was telling us to use scientific terms for behavior with our clients but on the other he was telling us to throw learning theory out the window and not focus on it anymore.

Last Saturday the group of trainers I work with had dinner together and discussed some of the other trainers in the area that will resort to shock collars, etc. when absolutely necessary. Like if a dog is chasing after a FedEx truck and nothing else will stop him. It brings you to a matter of ethics, what is more important, sticking to positive only training or preventing the dog from being run over when that fails?

In talking with Ian it seems that he doesn't like shock collars but if they are used properly, you should only ever have to use them ONE time for the behavior. If you have to continually use them you are doing it wrong and shouldn't be using a shock collar at all.

His seminar was interesting and made me think alot about how I organize my training classes, how I talk to my clients and how I deal with my own dogs. One of the things that really bothered me was the fact that when training recalls he only rewards the fastest recalls out of like 10 trials. With some dogs they would give up after three times in a row of not being rewarded because they can't win the game.

I took away what I could from his conference and put the rest on a "think about it" shelf.

Crystal said...

SAM- I'm afraid my understanding of reinforcement schedules isn't great, so forgive me if I get the terminology wrong. I believe Ian was discussing straight up differential reinforcement, though- reinforcing only BETTER responses. Straighter, faster, closer. I know that variable reinforcement is the best way to strengthen a behavior against extinction, but just randomly reinforcing won't get a sit straighter, will it?

I am quite jealous of your experience in the pigeon lab. Could you speak to whether or not the animals you've worked with are "simple" or have "limited interests"? I have no experience, but that comment just really rubbed me the wrong way.

I have a couple of posts in the queue about Ian's philosophy on punishment, but I'll give you the sneak peak: He believes you can punish animals without using pain or fear, and he accomplishes that primarily through verbal feedback.


KRISTEN- Yeah, I don't think we need to waste time in pet dog classes teaching the quadrants. People don't really need the details. I think it's possible to give the basic ideas without using all the terminology. That said, professionals should have a decent grasp of it, and how to use it. I think Ian was more concerned about the implementation. It was his impression that people argue learning theory on the internet ad nauseum, but don't spend time actually TRAINING. I know that I definitely have more book knowledge than training chops/experience at this point in my life. I need to work on changing that...


KATIE- Ian didn't tell us to use scientific terms with students. Could you expand on that a bit more? But, yeah, I sometimes felt like he was contradicting himself during the seminar. I'm not sure if I just didn't get what he was trying to say, or if I was missing the subtlety of context or what.

He did say that while he doesn't like shock collars/collar corrections, that if you use them, it should be once or twice, tops. Punishment should stop behavior, after all. I have a whole post coming up on this. :)

Could you talk a little bit about what he inspired you to do differently with your dogs?

Ninso said...

Hi Crystal! I linked over from Sarah and Layla's blog. I think you and I met once at a Rally trial. Thanks for the notes on Dr. Dunbar's seminar. It's really interesting to read people's recaps. I only have what I'm hearing from other people, but I think it's important to beware of extremes in dog training (in anything for that matter). Some might take learning theory to an extreme, but that doesn't mean it should be thrown out altogether. True, most dogs will learn just about anything by any method, if enough time is put into it. They are smart like that. But I don't think that means we stop trying to improve the training process and make it faster, better, easier on our canine friends. If you're going to train a dog to do anything more than the very basics, you have to learn something about how to do it. I'm having trouble understanding why Dr. Dunbar thinks it is preferable for people to learn his methods over learning theory. If you have a dense owner, either one is probably going to difficult to teach them.

Crystal said...

Hi, Ninso! I went over to your blog, and your dogs do look familiar, so I bet we have met.

I agree that it's wise to avoid extremes in pretty much anything, although I do have a certain bias in my methods. :) That said, I can understand criticizing people who say "I only use positive reinforcement and negative punishment," and then shun anything that falls in the "wrong" quadrant, even if it would otherwise fall in line with their methods (presumably dog-friendly). While it's good to understand what you're doing, and the pros and cons of that decision, I think it's a mistake to get hung up on it.

Laura, Lance, and Vito said...

Ahh yes, the moment when Ian slammed on all the dog trainers in the room by telling us we we analyze too much and don't train enough :) I completely agreed with him that pet classes have zero reason to go into learning theory and even I don't really care about what quadrant I'm using. I do think it's fun to over analyze everything else about training though and enjoy my yahoo groups where things are discussed!

I think the most relevant thing I got out of this lecture was to really think about what I am rewarding. Differential reinforcement is what we do when consciously shaping a behavior, but I need to remember to keep that up once the behavior is learned. A lot of times I get lazy and unless I am specifically working on straight fronts or finishes, I tend to reward every single one unless it was absolutely horrid. It certainly maintains his willingness to do what I ask, but it isn't helping me to get better scores in the ring!

Sam said...

Crystal -

My professor, who is a behavioral psychologist, often describes the pigeons as "emotional." I've found he's right. They may not be as sophisticated in temperament as dogs are, but they are definitely individualistic. The bird that I've primarily been working with is a workaholic, able to bang out several hundred pecks in one MINUTE, sometimes up to 4 pecks a second. Yet, there are other birds who take sort of a casual approach to lab.. a hundred pecks per minute or so.. not terribly motivated by the food.. etc. There is yet another pigeon who has anxiety/sensitivity issues! Despite the fact that she's been treated well and raised in the laboratory, this bird has flat-out refused to work for certain people. Definitely not "simple," if you ask me!

Some of the pigeons are "interested" in food. Some are not. Some are reinforced by the presence of certain people. Etc. etc. Same for dogs. Some are "interested" in smelling poop. Others, cheese. Still others, a scratch on the butt. It's all Premack. It all means that there is nothing inherently special about a reinforcer.. just that the reinforcer is a behavior that the animal will engage in if given the chance. I don't think dogs and lab animals like pigeons or rats are all that different in that aspect. Dogs just have the added variable of human companionship and bonding.

I still feel like a computer is capable of a differential reinforcement schedule, at least in the rudimentary form. You can't tell a computer to differentially reinforce a "better" response, but you can tell a computer to differentially reinforce fast responses, and maybe even set other criteria (positioning, like standing/sitting straight) with more sophisticated programs. I'm not really sure why Dr. Dunbar feels that he needs to bring that up, though - shouldn't that make us better at implementing various schedules of reinforcement? Shouldn't we use it to our advantage?

Verbal feedback as a punisher sounds interesting. Sounds like some form of a no-reward marker, right? BTW, you're totally making me want to write a post on punishment - I learned a lot of interesting things in the lab about the implementation of physical/aversive punishment!

Crystal said...

LAURA- I agree with you completely. I like to think about learning theory, but I do find it gets in my way sometimes when I'm training. It's much better if I simply train, and then analyze what I did later. It was hard not to feel defensive about my involvement in yahoogroups, though...

I also agree that he encouraged me to think about using differential reinforcement more. I've been doing that for straight fronts and finishes, although I have found that she needs to get at least one treat to start or she gives up and her performance gets worse. My dog is interesting. :)

Crystal said...

SAM- Thanks for your in-depth comment. I DO hope you'll post about what you've learned about punishment. That may be the closest I get to a lab. :)

I suspected that pigeons are complex creatures, and it really bugged me that Ian dismissed them as "simple." I have no particular affinity for pigeons, but I just didn't think that was a fair criticism.

Ian's point was that since humans are better than computers at judging whether or not a behavior is "better," (ie, better at using differential reinforcement), we shouldn't try to emulate "computer training," ie, learning theory. Personally, I think it's a pretty big leap from "we can do some things better than computers" to "we shouldn't use learning theory because it was developed by computers."

Anonymous said...

I attended the same seminar in Houston, with the addition that I'm trained in Education Psychology, which is applied Learning Theory. What Ian said was: Transcend Learning theory without abandoning the science. We observe and measure the facts, and apply what works.

There is little argument that what Ian Dunbar does is very effective and very rapid. He is still working on theoretical explanations. For example he talks about binary reward, but in fact, he gives levels of reward, rather than binary. And, he talks about non-aversive punishment, which a term that might fit into Learning Theory, but a more accurate term might be advisement feedback.

On a completely different tack, 'all' dog trainers talk about Conditioning, which is a term from Behaviorism. However, when I teach a dog to sit, achieving near 100% compliance with a happy dog, I believe that the dog is Not giving me a simple Stimulus-Response. Instead, I take the cognitive approach and I believe that the dog is going through a Stimulus-Decision-Response.

From a practical standpoint, this may not matter, but from a theory perspective, this means that my Reward is a marker for the correct behavior, as opposed to a reward. If this is true, then it provides a different theoretical explanation for why a dog will sit hundreds or thousands of times with no reward, once it has learned the cue.

-Hank Simon

Anonymous said...

BTW, yes Ian Dunbar's verbal feedback is a type of no reward marker (as Bob Bailey recommends). However, Ian's 'instructive reprimand' provides more information in the form of "no, do this, I love you." And he recommend gently insisting and repeating the instructive reprimand. Theory says this is wrong, which Ian acknowledges. But it works - quickly - you can't argue with the facts. The theory is wrong ... at least with dogs.

- Hank Simon

Crystal said...

Hi, Hank! Thanks for your comments. (And thanks for signing them. It's always nice to have a name to a comment.) I haven't studied learning theory formally, although I've done a lot of reading outside the classroom. This means that, naturally, I am still missing many of the finer points. :)

One of the things I really need to read is the Breland's "Misbehavior of Organisms." I think that might help me better understand some of what Ian said, because like you point out, the stuff he's doing is working, even if we don't understand why.

That said, I really struggled with Ian's presentation. Some of it just doesn't make sense based on my personal experiences with my dog. Doesn't mean he's wrong, of course, it just makes it harder for me to understand him.

And then, like you said, he's still working on the theoretical explanations. As a result, some of what he said seemed contradictory on the surface (such as "feedback is binary" and "you want your feedback to express value"), which makes it REALLY hard for me summarize what he says accurately and in a way that makes sense for those who weren't there. I hope I'm doing okay, but I know I'm missing some (many) of the nuances to his message.

Anonymous said...

Crystal, I like and greatly respect Ian, and I have near absolute confidence in what he does. However, his theory is still a work in progress, and bottomline, he always says to let theory guide, but follow the facts. He quotes Thorndike - the original behaviorist - that feedback is binary: reward for good, punish for bad - effectively the hot/cold game. But, Thorndike and the older behaviorists believed that everyone - people and animals - were blackboxes with no cognitive responses. We know that is wrong... 100 years later. I'm not sure what Bob Bailey thinks and does (he trained under Skinner).

But, in fact, Ian 'expands' his binary feedback, beyond hot and cold, to many shading of warm. And, I agree, it is contradictory to listen to, but not in how he implements. I also wish he'd stop saying 'punishment' because that is such a very charged word...

- Hank Simon

Crystal said...

Hank, I can get behind "let theory guide, but follow facts." I wish he would have said that in Minneapolis- or, if he did, I wish I would have picked up on it! I know that, in 3 days, there is stuff I missed. :)

Have you seen him actually implementing these ideas with animals? I would have loved some demos with real dogs. I think it would have given me a clearer understanding of what he said, especially since some of what he recommended are things that haven't worked with my dog in the past. I guess I'll have to go search the Dog Star Daily website and see if he has any.

Anonymous said...

Some of those old Sirius CDs, things he was selling during the workshop (and available online) and maybe some on Youtube, have videos of him training. But, he doesn't do dogs at his current 3-day workshops. He might do some at his place in California. I've never seen him train (except people)...

- Hank Simon

Houstontx Dogtrainers said...

Hi, just wanted to tell you, I enjoyed this blog post. It was funny. Keep on posting! Such a lovely blog you have shared here with us. Really nice.
________________________
Dog Training Houston