Monday, January 24, 2011

Reply to Patricia McConnell

This is a reply to Patricia McConnell's critique of one of my articles.

Patricia McConnell with One of Her Border Collies
Thank you, Dr. McConnell, for mentioning my blog article here. Oscar Wilde said that the only thing worse than people talking about you, is people not talking about you. Still, I stand by what I wrote. And I’d have to say that some of the people commenting here have actually proved my point for me.

It might help to know that the rationale behind my “Unified Dog Theory” series at is a, perhaps vain, attempt to help end the divisiveness in the training world, but also based on a very real need to help educate dog owners and trainers that there aren’t just two philosophical or methodological choices — dominance and positive reinforcement — there’s a third form of dog training, one that’s used to train working dogs: police dogs, herding dogs, search-and-rescue dogs, etc. It’s been modified for use with pet dogs by former police dog trainer, Kevin Behan, who learned his trade from his father, Jack Behan, a famous figure of the 1950s and 60s (and probably the first famous dog trainer in America). As Kevin grew as a trainer he disagreed with his father’s reliance on dominance, and set out on his own to find a training method for police work that wasn’t abusive, and formulated a training program in his 1st book, Natural Dog Training.

So there are really 3 common forms of dog training: dominance, +R, and drive training (in general) and Natural Dog Training (in particular).

One of your readers mentioned Sigmund Freud, and seemed to be scratching her head over why his name comes up in my articles. There’s actually a sound, scientific reason for it. That’s because the basic principles of drive training are more consonant with Freud’s pleasure principle than with Skinner’s experimental outgrowth of the pleasure principle, positive reinforcement. One of the ways Freud defined pleasure was as the release of internal tension. This is a physiological phenomenon, measurable in real time. On the other hand, positive reinforcement is only measurable after the fact through statistical analyses. Theres no real physiological basis for it. It isn’t an actual object, marker or event; it’s a function of statistics. (Plus, most behavioral scientists will tell you that there’s no real way to determine if a behavior was learned through positive or negative reinforcement.)

Personally, I’ll go with actual physiological realities (Freud) over statistical probabilities (Skinner) any day.

Another reason for writing this series, is to bring the wolf model — the real one, not the one that’s been proven invalid by modern research — back into play.

Formal obedience training got its start (by Max von Stephanitz in the early 1900s), in part, as a way of imitating the predatory motor patterns of wild wolves. This bit of history is missing from both the dominance and the +R views of training. And it’s very important, in terms of this current discussion, because juvenile wolves don’t begin hunting until they’re well into adolescence. That’s why I wrote in my article: 

This is the model that has been set in place by Nature, and has worked for millions of years. Why change it now? Why force puppies to pay attention and ‘learn,’ when Nature is telling them to jump around, bite, play, get distracted, and amuse the heck out of their owners?

The other problem is that it’s long been believed that a dog (or puppy) has to be calm in order to learn; dogs can’t learn when they’re highly stimulated. I’ve found that the exact opposite is true. I think it’s best to teach obedience skills as part of an active, high-energy game, where you stimulate the dog’s urge to bite, focus it on a toy, and teach him that he gets to win the toy by obeying your commands. It’s not a good idea to do this with puppies because once they get wound up it’s for them to wind down.

The more actively the dog’s whole organism is involved — his emotions, his kinetic energy, his instincts, and his brain — the better and faster he’ll learn. This is something, that frankly, you can’t do with young puppies because they only have 3 play settings: Off, Play Hard, and Play Way Too Hard.

It’s time we re-think the whole idea of puppy obedience classes, and perhaps set them up more as owner orientation classes, where the owners can watch their puppies play while the instructor explains a few simple training techniques for teaching their pup’s basic manners, but does so through the spoken and written word, without using the pup to demonstrate the process. That way the owners can learn two important things: how to teach their pup manners, at home, on their own time, and how much fun it is to watch puppies play together.

With all that said, I don’t think there’s anything wrong with using food or positive reinforcement in training pet dogs. Behan himself writes about the value of these things in his 1st book, something which caused a little conflict with some of the K-9 trainers he worked with over the years. But I also think it’s important to take a look at what Ian Dunbar has been writing about on his blog recently. 

“Laboratory study has revealed a variety of reinforcement schedules. Puppy training has revealed that most of these are notorious[ly] ineffective, or impossible to administer … Wake up! Puppy training has taught us that most of this stuff doesn’t work too well.”

Dunbar also says that one of the worst things you can do is reward a dog every time he obeys. Yet Bob Bailey, perhaps the most knowledgeable animal trainer we have at least when it comes to operant conditioning says that a dog should be rewarded every single time he obeys.

Like me, Bailey has also been somewhat critical of the +R movement, particularly the tendency some within it have to isolate themselves from the dog training community at large. Last year, after someone gave him a link to one of my blog articles, Bailey sent out a newsletter in response. In it he wrote, “The‘clicker training community’ has insulated itself from much of the public and from trainers not embracing the ‘purist’ methods … How many have heard me say, ‘Beware of he or she who claims [to have] The Truth?’”

Some people have closed their minds, but believe it or not, Cesar Millan sometimes gets things right, and some well-known figures in the +R world sometimes get things wrong. The truth is, dog training is as much of an art as it is a science. In fact, it may be more of an art than a science. We all gravitate towards what feels right to us. 

My “Unified Dog Theory” is an attempt to give people who are unaware of the scientific principles behind drive training, or don’t know how to utilize it with their dogs, some simple tools to start them on their way.

Hopefully, it will start the ball rolling toward bringing all dog trainers together rather than setting up opposing camps and sniping at one another. (And I’ve been as guilty of that in the past as anyone.)

Thanks again for mentioning my article.

Lee Charles Kelley
Life Is an Adventure — Where Will Your Dog Take You?

Wednesday, January 12, 2011

Unified Dog Theory IX: Language, Representational Thinking, and Object Relations in Dogs


There are three interesting studies in the news right now.

The first is about Chaser, a border collie who's been rigorously trained to remember the names of all 1022 of her toys, and can differentiate between the verbs "to nose," "to paw," and "to take." She learned to do this in a series of 4-hour sessions, taking place over a period of three years, using a ball as a reward for making a successful match between a word and a toy, or a word and an action.

Some in the media, and even in the scientific community, have touted Chaser's abilities as evidence that she has a rudimentary capacity to understand human language, and that her linguistic abilities are on a par with those of a 3-year old child.

Most animals -- dogs and humans included -- process salient features of their environments by making internal representations of them. This is pretty simple stuff. We look at an object but our eyes don't really see it, they just provide raw sensory data, which is translated by the visual circuitry in our brains into a representation, or mental image, of that object. Dogs are clearly able to do this, otherwise they wouldn't be able to dream or catch Frisbees (or, perhaps, dream of catching Frisbees).

However, when get into the realm of what words "mean," we're entering the territory of representations of representations. This means that for Chaser's abilities to rise to the level of a 3-year old child's linguistic skills, she would have to be able to understand words in a more abstract fashion, not just as verbal cues. (She would also have to be able to use words herself, do so in novel and inventive ways, and differentiate between nouns, adjectives, pronouns, verbs, etc., and use them with a fair amount of grammatical precision.)

Chaser's basic ability to respond to verbal cues, relative to her prey drive (toys are prey objects, after all), isn't unusual. Dogs, like wolves, are social predators, so they're constantly reading us for social cues. They do this so casually and so often that many times we're unaware of what's going on. So, to me, what's most interesting about Chaser isn't that she can respond to auditory cues, but the staggering number of them she apparently has stored in her memory banks. (Tellingly, her owners can't remember the names of all her toys; they have to rely on a written list!)

Meanwhile, another study, this one from UCSD, suggests that babies differentiate between words that relate to pictures of objects and those that don't in much the same way that adults do.

The researchers used MEG (which measures magnetic fields in the brain) and fMRI to estimate brain activity in 12 to 18-month old infants when they were shown pictures of familiar objects then heard words that were either a match or a mismatch to the name of the object. (Interestingly one of the tests involved showing the infants a picture of a ball followed by the word ball, versus a picture of a ball followed by the word dog.)

The brains of the infants lit up in certain areas when the word matched the picture. And the same parts of the brain lit up when human adults were given the same tests. Plus, these parts of the brain weren't those normally associated with language, such as the Broca's and Wernicke's areas. In fact, a much older part of the brain -- the cerebellum, which controls attention, and motor skills -- also lit up.

The third study may seem totally unrelated, but I think it can tell us something about how Chaser learned the names of so many toys, and also why matching words to objects seems to involve the cerebellum in human infants and adults.

Researchers at Princeton tested two groups of students for their ability to retain printed information based on how difficult the font (or typeface) was to read. One group was given a homework assignment printed with either a Comic Sans or Bodoni/Italic font, printed at 60% grayscale, while a second group was given the same assignment printed with the easy-to-read Arial font, printed in pure black.

The first group did substantially better on retention.

The researchers write, "It is not the difficulty, per se, that leads to [learning] improvements but rather the fact that the intervention engages processes that support learning."

What might those processes be?

The researchers say that "pinning down the precise mechanism [is] quite challenging." But in a recent blog article here, I wrote about dopamine circuits in the brain, and how they're not really "reward circuits," as they're often referred to, but attentional pathways. And that what they seem to reward is paying attention to changing patterns in the environment.

From that article: We're now discovering that the real purpose of dopamine is to help motivate us to gather new information about the outside world quickly and efficiently. In fact dopamine is released during negative experiences as well as positive ones. (The puppy who gets his nose scratched by the cat doesn't need further lessons to reinforce the "no-chasing-the-cat" rule; he learns that instantaneously, with a single swipe of the cat's paw.) This adds further importance to the idea that learning is not as much about pairing behaviors with their consequences as it is about paying close attention to salient changes in our environment: the bigger the changes, the more dopamine is released.

Could pattern recognition explain Chaser's amazing abilities?

I think so. Chaser is a border collie. (So was Ricoh, who knew the names of over 200 objects.) And border collies aren't bred for their linguistic abilities but for their herding skills. In other words, language is not a prerequisite for herding sheep but pattern recognition (which is an evolutionary pre-cursor to logic and language) is.

I said earlier that in the study done on how infants discriminate between word/object pairs that match and those that don't, one of the areas in the brain that lit up was the cerebellum, which is operational during motor control as well as during attentional tasks. And if dopamine is designed to make us pay attention to changing patterns in our environment, then it makes sense that the cerebellum -- which controls attention -- might also be involved in the cognitive process of finding words and objects that match up, and ignoring those that don't. This would also explain why the college students who had to expend more energy recognizing the patterns in their homework printouts retained the information better than those whose homework was easier to read.

I also said that dog owners are often unaware of how easily and how often dogs read our social cues (i.e., pick up on our patterns).

Years ago, when my dog Freddie and I would go on long walks in Central Park during off-leash hours, and he would do his business (#2), if I got distracted I would sometimes be unable to find where he'd left his "present," which would leave me perplexed and frustrated.

Fred, meanwhile, would usually be off sniffing around, or looking for other dogs to meet or squirrels to stalk. Sometimes, as I stopped and looked around, baggie in hand, I would say, "Freddie, where did you do your business?" And for some reason, one I can't exactly figure out, he would almost invariably stop what he was doing, come back over to the spot, and sniff his own poop, something he never did on his own.

I didn't train Fred to do this. He learned it on his own. There were no external rewards (though I did thank him). And perhaps more importantly, he didn't look at me or reference me; he never gave any outward signal that he "understood" my words. In fact, if anything it seemed to me that this was all his "idea," not mine. And yet he almost always went back to his spot and sniffed it.

So, is Fred's behavior an example of a capacity to understand human language, a facility for pattern recognition, or something else entirely?

I'll let you decide.

"Changing the World, One Dog at a Time"
Join Me on Facebook!
Follow Me on Twitter!