These are my notes from the ORCA Great Minds conference at UNT in March 2009. The Great Minds conference brought half a dozen top trainers together to talk about animal training, clicker training, operant conditioning and the future of training. This is the fourth post of my thoughts and notes and the first page of notes about Steve White. To read more about the conference in general, please see this post.
After hearing Steve Martin talk about the art of training, we were in for another treat–Steve White. Two great lectures in one morning, and both by Steves!
Steve White’s lecture was titled “Skinner and the World Collide!” He addressed the history of clicker training and operant condition training and then talked at length about how training in our Messy World sometimes ends up working differently from the controlled environment of the Skinner box. Here are some of my thoughts on some of the things Steve White discussed.
Life is not a lab.
What works in the lab doesn’t always work in the real world. Or, put in a way that anyone who’s ever trained a dog could probably understand, what works in the kitchen or living room at home might not work at the dog park tomorrow. It’s the classic owner chasing after the badly behaved pooch—“But he never does this at home!”
As trainers, we usually don’t have the benefit of a controlled lab environment without distractions. Instead, we have our messy world, composed of an infinite variety of environments and highly variable stimuli. The animal must learn to generalize because the environment is constantly changing.
Because or world is so messy, there are quite a few times when the animal does not respond to the trainer’s cue. What drives behavior? The animal’s perception of their environment. Perhaps the animal did not respond because they did not perceive the cue. Look! There goes a squirrel, and the dog completely misses the trainer’s recall signal.
This ties into the other point Steve White made about cues. Often our animals don’t understand our cues. This is the trainer’s fault, as they often ask something of the animal that they have not adequately prepared the animal for. Imagine asking a kid to drive a stick shift car when he’s never driven one before. You’re probably setting him up to fail. With our animals, we need to make sure we are always setting the animal up for success.
This is one reason why I’ve become increasingly uncomfortable for using pressure and especially escalating pressure to teach cues to horses. Many natural horse trainers teach their horses to back by wiggling the lead rope. They wiggle it very softly and the well trained horse backs quickly to the end of the lead.
However, when they first start teaching the behavior, the horse has no idea what the little wiggle means. So, they give the little wiggle, then a slightly larger one, then they’re swinging the rope, until finally, if the horse hasn’t backed yet, they’re swinging so hard that the snap on the lead rope is popping the horse in the chin. The horse finally gets uncomfortable enough that he backs and with a frightened look in his eye that says–“Hey, what did I ever do to deserve that?”
The horse learns very quickly that the little wiggle of the finger means he should back, and a competent trainer only has to use the extreme pressure a few times. However, when my horse backs up, I want him backing up because he wants to, not because he knows the discomfort he’ll feel if he doesn’t. Too many times, I see pressure/release trainers start training a behavior by putting the animal in situations where the animal does not understand how it can be successful.
Reinforcers and self-reinforcing behaviors
As well, there are reinforcers everywhere! On the ground, in the air, on the other side of the door, in the animal’s head. The trainer can run into trouble when she doesn’t anticipate or notice things that are very reinforcing for the animal, such as smells on the ground, birds in the trees, other people nearby or any other number of things.
Steve talked about the power of intrinsics and how behavior is often self-reinforcing. Many behaviors are genetically hardwired or can be classically conditioned to be self-reinforcing. The good trainer is able to recognize behaviors that are self-reinforcing to the animal and either use them as reinforcers or as rewards in behavior chains. For example, some of the “trick” behaviors we teach get reinforced enough that the animal no longer even wants a reinforcer, the getting to perform the behavior becomes reinforcing in itself. Some horses love playing fetch, just as a dog would. Many horses also really seem to enjoy standing on pedestals. I’ve known many horses who would go out of their way to go stand on a pedestal.
One of the most interesting things Steve said was that the key to reliability is to get behaviors to be self-reinforcing. I think there is so much truth in this idea. If the animal does not enjoy performing the behavior, then the trainer will always have to be managing and checking on the animal to make sure the behavior gets accomplished. If the behavior becomes self-reinforcing, then the task becomes easier for the animal and the trainer. Both have to exert less energy and both end up happier in the end.
Self-reinforcing behaviors can also be a nightmare for the trainer if the behavior is something the trainer does not want the animal doing. Think of things dogs love to do that most owners hate–barking, digging, pups chewing on the furniture, and so on.
There’s a popular saying in the clicker world that not clicking a behavior will extinguish the behavior. Steve made the point that this is completely untrue. In fact, this is a dangerous assumption to make because it denies environmental reinforcers.
Behaviors become extinguished when they are no longer being reinforcered. If we remove the clicker, we remove the reinforcer we were providing. However, if the environment is still reinforcing the behavior, the behavior will continue, rather than be extinguished.
Be sure to check back tomorrow and later in the week, as I’ll be blogging about the rest of the conference speakers. If you don’t regularly read my blog, but find this interesting, I encourage you to subscribe via the RSS feed or by e-mail subscriptions (both located at the top right. Thanks for reading!