Monday, October 1, 2012

Shedd Animal Training Seminar: Getting Behavior

So far, we’ve learned that Ken prefers to use positive reinforcement when training behaviors, and that he finds the use of a bridging stimulus to be very helpful. But we haven’t talked about how he gets the behavior he wants to bridge and reinforce, so let’s do that today! The first thing you need to know is that while some trainers will use only one or two methods to the exclusion of all others, Ken uses what works. The method that he chooses will depend on the animal and the behavior.

As an interesting side note, while the concepts were familiar to me, Ken uses some different terminology than I’m used to. Indeed, the words that zoological trainers use differ from things we say in the dog world. As a lover of words, I actually found this fascinating. For example, he uses the word “shaping” differently than we do in the dog world; for Ken, “shaping” simply means training. Each of the techniques he discussed can be broken down into small increments, and he will reward each successive approximation regardless of what technique he’s using… which is shaping. Just not the way we dog people typically think of it.

Targeting is probably the most common technique used in the zoological setting, mostly because it is incredibly useful for teaching husbandry tasks. (Ironically, it’s probably the one I use the least with my dog, although I have found it incredibly helpful when I’m trying to learn a physical skill myself.) Targeting involves teaching the animal to touch an object with a body part, either briefly or for an extended duration. For example, a dolphin may be taught to touch its nose to a colored shape, or a sea otter may be taught to grab a plastic buoy with its paws.

Sometimes called capturing in non-zoological settings, scanning is where the trainer watches for the animal to perform a desired behavior. When the animal does- for example when a dolphin jumps in the air or a beluga whale spits water- the trainer reinforces the animal at the exact moment the behavior happens. As you can imagine, the use of a bridging stimulus is incredibly important here. Scanning is also sometimes called free-shaping, as the trainer will capture a first step toward the behavior and then gradually increase the criteria.

Baiting is also known as luring, and it can be a controversial method. Baiting is done by using food to elicit the desired behavior, and if not done well, can create animals and trainers who are dependent on the use of food to complete the behavior. We actually saw some baiting in action when Ken was working to teach Tanner the sea lion to go in the water on cue. He would give the cue (swim!) and then toss a fish in the water. While it is very useful- especially for novice animals and trainers- Ken said it’s typically not his first choice for a lot of behaviors. Still, he rarely says “never” to a training technique.

Environmental Manipulation
I love environmental manipulation; it can make things so easy sometimes. This is simply the process of arranging things in the environment so that the desired behavior is the only or the most likely option for the animal. It obviously won’t work for every behavior you want, but when it does, it’s genius. What could be better than setting the animal up for success and giving the trainer the opportunity to reinforce the behavior?

This technique, sometimes called molding, is more common in the dog world than zoological settings. Modeling is where you physically manipulate the animal’s body into the position you want (like when a trainer pushes a dog into a sit), and as you can imagine, tends to be either physically impossible or incredibly dangerous with zoo animals. Still, Ken refused to discount it entirely, citing Koko the gorilla as an example. Koko’s trainers taught her sign language by manipulating her fingers and hands into each word, something they could have never targeted, captured, or lured into happening.

This is not a common method of training as it requires a fairly sophisticated animal who understands that it should copy what another animal is doing. Although Ken has done some work around mimicry, it is not widely used. A similar concept is social facilitation, in which an animal learns to do something from another animal. This is typically accidental, such as when a dog learns to bark at passers-by from another dog.

Abstract Learning
Abstract learning is a complex method of learning. This is where two ideas are combined to create a new concept. It happens often with humans, such as when we attend lectures, but is not common in the animal world, although it does happen with cognitive researchers. Alex the Parrot is an excellent example.

This list is certainly not all inclusive. There are other ways of training animals, but these were presented as some of the most common methods. I initially learned through the use of baiting/luring, and appreciate it as a technique. I also do a fair amount of scanning (both capturing and free-shaping). What about you guys? Which technique do you use the most? Are there techniques you’ve used that aren’t listed here? I’d love to hear about your experiences.

No comments: