kitchen table math, the sequel: Animals in Translation

Monday, July 30, 2007

Animals in Translation

on hyper-specificity:

Animals and autistic people are splitters. They see the differences between things more than the similarities. In practice this means animals don’t generalize very well. You have to be really careful about the way you teach animals because of this. I had an interesting talk recently with a lady who keeps wolf hybrids for pets (something I don’t recommend). She said if you’re going to have a wolf hybrid as a pet, you have to socialize it between 4 to 13 weeks of age that all men are OK, not just the owner. Otherwise they’ll think the owner is OK, all other men are the enemy. You have to do the same thing with women, children, toddlers and babies, and you have to socialize the animal to different members of each category separately. It’s not just children who are OK, it’s toddlers, and it’s not just toddlers, it’s babies. And it’s not just the neighbor’s toddler who’s OK; it’s all toddlers, and so on and so on.

You have to do the same thing service dogs’ behavior when they’re learning to take a blind person across the street. The dogs don’t generalize from one intersection to another, so you can’t just train them on a couple of intersections and expect them to apply what they’ve learned to a brand new intersection. You have to train him on dozens of different kinds of intersections: corners where there’s a light hanging in the middle of the intersection and crosswalk lines painted on the pavement, corners where there’s a light hanging in the middle of the intersection and no crosswalk lines, corners where the traffic lights are on poles, and so on. And you can’t just train him in one city. You have to train him in different cities.

This is why dog trainers always make people train their own dogs. You can’t send a puppy away to obedience school, because he’ll only learn to obey the trainer, not you. Dogs also need some training from every member of the household, because if only one person trains the dog, that’s the only person the dog is going to obey. And you have to be careful not to fall into “pattern training.” Pattern training happens when you always train the dog in the same place at the same time using the same commands in the same order. If you pattern train a dog, he’ll learn the commands beautifully, but he won’t be able to perform them any place other than the spot you trained him in, or in any sequence other than the one you use during training. He’s learned the pattern, and he can’t generalize the individual commands to other times, settings or people.

People who teach autistic children deal with exactly the same challenge. A behaviorist told me a story about an autistic boy he’d been teaching to butter toast. The behaviorist and the parents had been working really hard with the boy, and finally he got it. He could butter toast. Everyone was thrilled, but the joy didn’t last too long, because when somebody gave the boy some peanut butter to spread on his toast, he didn’t have a clue! His brand new bread-buttering skill was specific to butter, and it didn’t generalize to peanut butter. They had to start all over again and teach him how to spread peanut butter on toast. This happens all the time with autistic people, and with animals, too.

That animals and autistic people are hyper-specific was a pretty good insight if I do say so myself.

The problem is that neither Temple nor I understood, while we were writing the book, how difficult it is for non-autistic people to generalize what they've learned from one setting to another.

Transfer of learning.

The $64,000 question.


and see:
Animals fear details that people do not notice

.................................

Apparently there is a lumpers versus splitters situation unfolding in the world of ornithology.

.................................

No comments: