"The Power of Play, Part I."
Does play enable dogs to learn things instantly and permanently?
Before I studied formally to become a dog trainer, I had a black-and-white English field setter named Charley. One or two people might remember him as Charlie (sic) “the bubble-eating dog” on David Letterman’s NBC show in late 1989 and early 1990.
Charley and I were in Central Park one autumn evening hanging out with a group of dog owners, when a new dog—a fox terrier also named Charlie (note spelling)—showed up.
After the preliminaries were over, this new Charlie got my Charley, to start chasing him. And in the course of their game, New Charlie did a maneuver where he faked left and went left, which totally fooled my dog, causing him to go right and tumble sideways into a tree. He yelped in pain. But before I could run over to check on him, he stood up, shook himself off, and ran after the other Charlie again, barking and chasing him even harder than before.
Huh, I thought. That was a very negative experience, and yet, I supposed, because perhaps his adrenalin and endorphins were set at a pretty high level, it only made him want to play even harder, not give up and go home.
Then, during the next lap a truly amazing thing happened: the new Charlie tried the same maneuver that had fooled my Charley, but this time he wasn’t fooled. He stayed right on New Charlie’s tail. I was stunned. I had always thought that learning required trial and error and/or repetition, but in this case my Charley had learned the other Charlie’s “strategy” through just one experience.
Still, I thought, it certainly was a singular experience. Maybe that’s all Charley needed. But then another strange thing happened: this time the new Charlie faked right and went left, the exact opposite of the previous move. But my Charley still stayed on his tail. Then the other Charlie faked left and went left, but my Charley still wasn’t fooled.
I wasn’t sure how this could have happened, but thought there may’ve been something my dog saw during that first fake which registered in some part of his brain; there was some ineffable change in the other Charlie’s kinetic energy, motion, or emotion that signaled fake as opposed to real, though my dog didn’t recognize its significance until he crashed into that tree.
They played like this for a while, and then New Charlie and his owner left, and a little while later my Charley began playing with another dog.
And here’s where it gets really good. During a chase with this new dog, Charley immediately incorporated the fake-left-go-left maneuver he’d just learned. He’d never done it before that day, yet he effortlessly incorporated it into his play repertoire, including, on subsequent laps, all possible variations. In other words he’d “generalized” everything in just one, head-over-tail experience.
It was instant yet permanent learning.
That’s when I first began to realize that there must be something about the physical and emotional dynamics of play that increases a dog’s ability to process information, something that decreases the amount of time it takes him to learn a new behavior, and which also increases his ability to remember a new skill to the point that he’s able to immediately make it a part of his own behavioral vocabulary, seemingly forever.
I’ve seen this in other dogs since then, and I’m still not sure how it happens. But it seems to me that for dogs rough-and-tumble play involves instincts and emotions which simply require a great deal of attention to detail, not to mention a great deal of attention to a lot of details all at once (a gestalt, if you will). Because of this, free play probably creates more dendritic connections in one single burst of learning than free-shaping or force-based training does over a period of weeks or even months. In other words, free-shaping creates connections one branch at a time. Force-based training creates similar connections but, since they’re based on fear, a lot of unwanted, survival-based connections also get made. But play-based learning seems to create whole dendritic forests in a single burst of joy and excitement.
Do I know, scientifically, that that’s the case? No. But there is some scientific evidence possibly pointing in that direction. Social play in rats, for example, activates certain brain growth factors such as BDNF (or brain-derived neurotrophic factor). These chemicals stimulate neurons and synapses for a) more growth, b) more differentiation, and c) better long-term survival (the learning “sticks” better and longer). (Gordon, et al., 2003). It’s also known that animals who’ve had fewer opportunities for free play as youngsters are more likely to exhibit aggressive behaviors as adults. And this isn’t just due to lack of socialization, it’s that play is vitally important for healthy, lasting growth of the neural connections between the aggression centers in the brain like the amygdalla, and the impulse control center in the frontal lobes (Potegal & Einon, 1989).
Still, it seems to me that no matter how quickly these brain growth factors work, they don’t do so instantaneously. They still take a certain amount of linear, chronological time to get their effects. And what I observed years ago in Central Park didn’t take any time at all. It happened instantly, and it included the ability to “generalize,” a kind of cognitive ability animals don’t ordinarily exhibit, and which is thought to be a higher function of the neo-cortex.
Is this kind of learning simply the result of mirror neurons, which may very well be part of a dog’s cognitive architecture (whereas the highly-developed neo-cortex isn’t)? That would make sense in terms of my Charley being able to imitate the other Charlie’s moves, but it wouldn’t account for his ability to learn all possible variations in one burst of learning, and then repeat them, including, again, all possible variations, with other dogs. Perhaps he didn’t learn them at all that day. Perhaps he was born “knowing” them; all of the other dog’s moves were encoded somewhere in my Charley’s DNA and that first experience he had with New Charlie acted as a trigger to de-code them.
Or maybe it’s not encoded into the DNA at all. Maybe it’s part of a morphic field or network consciousness. Either or any of these possibilities would solve one of the biggest problems cognitive scientists have with what play in animals is, not to mention how and why it works.