Sunday, February 28, 2010

How to Train a Cheetah


A rabbit-coursing Cheetah in the UK, circa 1937.

I am always a bit amused by the folks who talk about scientific dog training.

What they are trying to suggest is that there was no good dog training before B.F. Skinner (or Karen Pryor) and that they have discovered a secret science.

Now I am a big fan of B.F. Skinner. A great man.

I am an even bigger fan of Keller Breland and Marian Breland, who studied with Skinner, and with Bob Bailey who married Marian Breland after Keller died and who went on to helpe popularize reward-based animal training here in the U.S.  Karen Pryor will tell you most of what she knows she learned from Bob Bailey, who will tell you most of what he knows originated with Marian and Keller Breland who, in turn, learned a great deal from Skinner.

That said, B.F. Skinner did not invent operant conditioning, any more than Newton invented gravity.

Operant conditioning predates Skinner by many millenia, and not just in training dogs.

Consider this text from an 1865 publication entitled "Dog Breaking: the Most Expeditious, Certain and Easy Method".



Hey, that's operant conditioning!

Well yes it is. And in a book called "Dog Breaking" no less.

Hmmmm.... What's that about?

Well, it has more to do with language than technique. Language, you see has always been a problem in dog training.

The term "breaking," as used here, is simply a term moved over from the world of horses. It does not mean you are beating a dog into submission. Far from it, if you actually take the time to read the text.

But of course, when a rival publication came out in 1882, the term "breaking" was the new bogey man (every new product needs to first demonize the old).

Breaking? That must mean beating the dog to death right?

Who wants something that is broken? No one!

The new publication was entitled, "Practical Dog Training: or, Training vs. Breaking" and it wanted everyone to know that you could actually train a dog with a bit of food and a check cord.

And never mind if that was the the exact same message of the earlier text!

The "dog training wars" had started, and YES, it was mostly nonsense right from the start.

Of course, when B.F. Skinner's disciples came along, they too promised a new "scientific" system to replace the "old coercive" system of dog training that they said had preceded them.

They had just learned, they said, that animals could be taught to come for a piece of food, and that they could be taught to avoid things that made them uncomfortable.

Really? Tell us more!

Of course when Skinner showed up, he did not clarify things much in terms of language, did he?

In fact, he made an even bigger muddle of the language by using words like "punishment".

Punishment? Who wants to punish a dog? No one!

In fact what was being meted out most of the time was not "punishment" as we commonly use that term, but a simple instruction that says "we need less of that."

What Skinner called "reinforcement" was simply instruction that "we could use a little more of that."

Language continues to obscure, of course.

When Cesar Millan talks about "dominance," he is not talking about beating a dog as some assume; he is saying the human needs to be setting the agenda. The human needs to decide what food is served, when it is served, who gets it, and in what quantities. The human needs to decide when a walk starts, where it goes, and when it ends. The human needs to decide the rules of the house, not the dog.

Of course words like "dominance" tend to set off a lot of folks, just as words like "punishment" do.

Most of the time these folks have not actually bothered to read B.F. Skinner or Cesar Millan, much less page through old dog training books by Barbara Woodhouse or William Koehler, or the even older texts on "dog breaking."

It's so much easier to demonize stuff if you have not actually read it!

OK, enough about that. Point made (if a point is to be made).

I promised a lesson in how to train a Cheetah. Where is that?

The paragraphs below come from that earlier-named 1865 text on "Dog Breaking," and once again we find operant conditioning at work:

The education of the cheeta is no less progressive than that of the dog; and whatever patience the latter may require from his instructor, the former demands far greater; not so much from want of docility, as from the nearly total absence of all the feelings of attachment so conspicuous in the canine race.

The cubs when they are very young are stolen from the rocky fastnesses where they are usually bred. They are immediately hooded, and allowed no other exercise than what they can take when they are led about by their keeper. While he is feeding them, he invariably shouts in a peculiar key. In a month or so their eager looks, animated gestures, and possibly cheerful purr-ing, testify that they comprehend its import as fully as a hungry young ensign does "the roast beef of old England." They are then slightly chained, each to a separate bandy (bullock-cart), and habituated to its motion. They are always fed during the drive. They thus learn to expect a good meal in the course of their airing.

After a time the keeper, instead of feeding a promising pupil while he is a prisoner, goes to a little distance from the bandy and utters the singular cries now so joyfully heard, upon which an attendant slipping off the chain and hood the liberated cheeta runs to his trainer to be fed. By degrees this is done at increased distances. He is always conducted back to the carriage by the keeper's dragging at the lump of meat of which the animal retains a firm hold.

The next step is to for the man to again commence feeding near the cart, but without making any noise -- the removal of the hood being the only thing to tell the spotted beast to begin to look about him for his dinner. The last step is the substitution of a kid or wounded antelope ... His education is now completed; but for many months he is never unhooded at a herd unless the driver has managed to get the cart with a very favoring distance."


If you are a hawker or falconer, this all might sound pretty familiar.

Indeed, falconers had been training and flying wild birds for a millenia before B.F. Skinner trained a pigeon, or Bob Bailey noted that "you cannot choke a chicken."

But don't tell that to the "modern scientific" trainers. They still have something "new" to sell!
.

5 comments:

Sighthound Scott said...

It's funny how personal and emotional semantics can be for some folks.

I recall an otherwise amiable conversation I overheard between two owners (at a dog park? I don't remember where for sure) that went south quickly as soon as one mentioned the word "command." Apparently, according to the other, "cue" is the only correct term if you have a brain and any hope of a fulfilling relationship with your dog.

Come to think of it, she'd probably also have an issue with my use of the word "owner" above in describing her. :0)

Doug said...

Yeah - those methods do look very familiar. I agree that it is bizarre how up in arms people get over semantics.

Mina said...

As part of my studies I've got a book called 'Dog Breaking for the Gun' on my reading list. I haven't had a chance to open it yet, but this post gives me hope it's not as bad as it sounds!

smartdogs said...

Great post - (and I hate to be a weenie term freak) but... your example of associating gunshots with feeding is Pavolvian or classical conditioning, not operant.

Cheers,
Janeen

PBurns said...

It's actually classical AND operant. Classsical is old Pavlov -- a bell or buzzer comes to generate a biological response (saliva, etc.) in expectation of food. Operant conditioning is a shaped behavior that is not biological in basis -- like teaching the t dog to crouch so as not to get hit with stray shot.

Now, here's a question: What's a "Grizzly dinner bell" where a .50 cal shot brings in grizzlies who expect to be fed on the just-shot elk and/or its gut pile? I would argue it's operant, but it's also classical too isn't it?

P.