Rewarding penguins

Perched on a sunny rock in the middle of the penguin exhibit, I watched as the penguin trainer next to me quietly vibrated with happiness. She was overjoyed because a shy young penguin, who had never liked her much despite bonding well with some of the male trainers, had climbed into her lap. Being an animal trainer, of course, she wasn’t going to let the opportunity slide to reinforce this behavior so it would occur more often in the future. She was gently rubbing the penguin on his chest. This startled me: dog trainers insist that the best reward is food. Why was she using physical contact, something that is considered insufficient reward in most other species?

The answer lies just in the practicality of feeding these birds. Unlike sea lions, who go through several buckets packed with fish a day, the little African penguins this trainer works with may eat as few as two herring a day. That’s just two chances to reinforce behavior. I asked why they didn’t cut the fish up into tiny little pieces instead of offering them whole, but watching feeding time made that more clear: feeding a penguin is a careful dance of getting the food down their throat (they often refuse, and what was the favorite kind of fish yesterday may have fallen competely out of favor today) and keeping the trainer’s fingers from getting munched in the process. All the penguin trainers had nets of scars on their hands from interactions with these birds, and were very blasé about being bitten; I was bitten three times in the calves during my short six hours with the birds. (Once was my fault for getting too close to a chick when her mom was nearby, but I can’t quite see how I could have avoided the other two times.) Throwing little pieces of fish is impractical, not to mention difficult in a flock of 27 birds.

So the trainers gradually condition the penguins to like being petted, so they can feel that they are being preened as another bird would do it. Where food is considered a “primary reinforcer” (something the animal requires to live), preening is a “secondary reinforcer” (something the animal has been conditioned to like). In the penguin trainer community, however, there is debate about whether preening could actually be considered a primary reinforcer, as arguably keeping oneself clean of parasites is essential for life.

Either way, an hour spent on a rock surrounded by 27 penguins is an hour well spent.

Olympic level dog training seminar

The nice thing about having weekends free on your ambulatory rotation is that you can do the things normal people do with their weekends, like go to incredibly geeky dog training seminars about the details of learning theory. I went to a seminar on improving the cues you give your dog with the brilliant Kathy Sdao. Some of her main points, below.

• A command is something you give with the expectation of making the animal obey. A cue is something you give to say “you know that behavior that you have learned is a really good thing to do, because you often get paid to do it? Now would be a great time to do it!”
• Oh, for those of you who aren’t dog training geeks, I should back up. “Getting paid” is dog training lingo for the idea that dogs don’t work for us just because they love us. Do you do work because you love your boss? Dogs get paid with food, praise, life rewards (getting a ball thrown), etc.
• You can’t control a behavior. You can only control what happens before and after a behavior, and therefore the animal’s expectation for cues that predict a good or bad time to perform the behavior, and consequences of the behavior.
• You certainly can use cues which are difficult for your dog to distinguish, like down and out, which have the same internal vowel sounds and are very similar to the ear if you don’t have human-level language skills. However, why would you? Choose words that make things easy for your dog by being easy to distinguish from each other. Your dog is the one who is struggling to understand language, something his species does not excel at.
• We know it, but it’s worth repeating: a dog doesn’t refuse to perform a behavior for spite. That’s only something humans do. A dog who doesn’t respond to a command or cue does so for only one of two reasons: a) he doesn’t understand what is being asked of him, or b) he doesn’t feel it is worth his while (he is not being paid sufficiently).
• We think of cues as verbal or gestural. Of course, dogs are more comfortable with gestural cues in general. (One participant found that her dog completely ignored a verbal cue which she had always given paired with a gesture.) You don’t have to make gestures only with your hands! Some people with small dogs find that the dogs respond very well to foot gestures, which are closer to their eye level.
• Cues that we give without meaning to include eye movements, where our attention is, body language (the classic story of the dog refusing to lie down unless the owner bends forward, because that may not be the cue the owner taught, but it is the cue the dog learned). Tone of voice. Time of day. Antecedents like picking up your keys. My dogs have learned that “okay,” spoken while I am on the computer, means I have decided to get up and get myself off the keyboard, even though it is something I say unconsciously to myself and not intentionally to them.
• Dogs live in a sea of information coming from us. Sometimes it is hard for them to pick out the cue we want to give. You think you are just raising your hand to indicate sit, but the dog is taking in tone of voice, where you are looking, if you are bending forward, what your other hand is doing, the position of your feet. It isn’t obvious to the dog that the hand (or word) is what he is supposed to be paying attention to.
• Kathy gave two examples to illustrate that point. The first: you know the feeling you get when you are tuning a radio and you can’t quite get the station, and have to listen to it through static? Dogs live in that world all the time.
• For those who know the invisible gorilla illusion — if you don’t know to look for something, you may not see it. If the dog is paying attention to how far forward you are leaning, he may not even hear the word you are saying to him. After all, words come out of your mouth all the time, and he usually doesn’t understand them. Why should he pay attention to this one and assume it has some importance? Why should his brain even filter is so that he hears it at all?

We played a fun game with the participating dogs called “Prove It.” The handler asserts something like “my cue for sit is the word ‘sit.’“ The challenger says, “Yeah? What if you say it with your eyes closed? With your hand over your mouth? Looking at the ceiling? Whispering? Standing on a chair? Will your dog still do it then?” I worked with a handler whose dog did standard commands flawlessly, promptly, and enthusiastically. When her handler told her to sit while looking at the ceiling, the dog stared at her hopefully, trying to figure out what she was supposed to do. (During this game, we took lots of breaks to give the dogs easy tasks and reward them.)

Fun day learning about how to communicate with your dog, and gave me good insights into my own training methods. Kathy Sdao is an excellent teacher. If you are a beginning trainer, her classes may seem a little arcane to you, but I highly recommend her if you are a learning theory geek, or someone who teaches other people to train dogs.

Letting go of dominance theory

Recently I took my dog Jack along on a visit to a friend who was dog sitting. My friend’s resident dog has always gotten along with Jack pretty well, but this guest dog, Ally, had the habit of rushing at Jack when he entered the room, barking at him and generally behaving in a manner that alarmed all the humans involved. Ally was also hesitant around strange humans, in this case myself and my boyfriend. She would come up to us only gingerly, and was easy to scare away with sudden movements or loud noises.

My friend theorized that Ally’s issue with Jack was that she wanted to be the alpha dog, because dogs see the world in terms of a pack structure, like wolves. He felt the way to fix the problem was to let Ally know who was really boss, so whenever Ally rushed at Jack he would yell at her. The idea that a problem in dog training can be solved by asserting dominance is known as dominance theory; use of dominance theory has been publicized recently by Cesar Millan, the Dog Whisperer. A recent article in the Boston Globe addresses the controversy about Millan’s approach, and has this quote from Karen Pryor, who popularized clicker training, which is a very different approach:

...while his critics dispute Millan’s claim that domestic dogs are pack animals and should be treated as such, Pryor proposes that Millan’s hard-line message speaks to the real pack animal in the room. “We’re the ones who care very deeply about who’s boss and we don’t want to stop believing that humans are superior,’’ Pryor says. “We’re primates that have gone strongly in the direction of hierarchies. Dogs? They don’t care about that at all.’’

This quote really tickled me, even though in my opinion it goes a little too far. No offense to Pryor; I think her book Don’t Shoot the Dog should be required reading for any pet owner. But I think saying dogs don’t care about hierarchies at all may be overstating the case. In my house, it’s clear which dog is in charge, though the other two don’t seem to much care who comes second and who comes third. When I sat in on sessions with a veterinary behaviorist last year, I saw dogs diagnosed with dominance aggression, and I agreed with the diagnosis.

However, I do object to the blanket application of dominance theory to all problems in dog training. Wild dogs don’t run in packs the way wolves do; even if they did, to say that dominance issues are the answer to every canine behavioral problem is silly. Dogs are more complicated than that. In Ally’s case, I don’t think she was trying to show Jack that she was in charge; I think she was scared of him, and trying to deal with the situation in the best way she could come up with.

And who developed the idea that yelling is what makes a good leader? Are managers who regularly yell at their subordinates considered good bosses? My favorite bosses were always people who helped me solve my problems, not people who got angry at me when I couldn’t find a solution on my own.

The American Veterinary Society of Animal Behavior has a position statement on dominance theory, in which they provide a good framework for approaching its use in training. I do believe that overuse, or misuse, of dominance theory will gradually fade from the way we as a society manage our dogs, but for now, it seems to be hard for us as humans to let go of it.