Wednesday, November 16, 2005

An update on the animal experimentation

You may recall that in an effort to stave off canine senility and entertain myself, I embarked on a series of fun psychological experiments/games with my dogs. I'm sure you've all been waiting with bated breath to hear about them.

Dogs are thought to have the cognitive abilities of a 2- to 4-year old human, and one of the important things that happens in early human development is the idea of object permanance - the concept that an object exists even when you can't see it. It may seem obvious, but it's actually a pretty amazing leap. Think about very small babies playing peek-a-boo: the reason they are so delighted to see the stuffed bunny reappear is because for the moment that it was hidden behind your back, it genuinely ceased to exist for them. Object permanence gives us the concept of people and things as independent, abstract beings.

So my exercise is a variant on 3-card monty, or the shell game. I set a piece of kibble on the floor, and cover it with one of two cups that I place face-down on the floor. When I tell them to "get it!" they have to nudge the correct cup with their nose to get the treat. Wrong cup, no treat. Dogs supposedly see blue clearest of all colors, whereas red looks like black. I thus have a blue and a red cup.

Not surprisingly, Nelly is better at this game than Toby, although it did take her a while to figure out what the hell I wanted from her. She now has almost 100% accuracy, so I'm thinking of increasing the complexity - perhaps a third cup, I don't know. They both get less accurate as we go on, which I suspect has to do with the fact that both cups start to smell like treats, and they rely on their noses first and foremost.

One of the fascinating things that Stanley Coren does in his books is to make real what it's like to have a brain that works in a fundamentally different way. He does this particularly well in the book How Dogs Think, with this analogy: Suppose you walk into a room and see a table, covered with a tablecloth, with a pen and a book and a bunch of other things on it. Your brain takes in all of that quickly and effortlessly - the layering of items one upon another, the variety of items, etc. However, the smells in the room are probably only an amorphous blob that makes little impression, unless there's something that really stands out like the smell of fresh bread, or ammonia.

For dogs, it's basically the reverse. In addition to having more complex noses, they have entire lobes of the brain devoted to smell that don't exist in humans. So they pick up each smell distinctly, with an awareness of placement and layering that's analogous to our sense of sight. Coren notes that smugglers will naively try to "mask" smells from detection dogs by, say, putting a bag of drugs in the gas tank of the car. A drug-sniffing dog will simply smell a car with drugs in the gas tank, much as we see the items on a table, clear as day. Their vision, on the other hand, will be more of a blurr. Other than being highly evolved for motion detection, vision is a dog's secondary go-to sense.

So one of the hardest things to do is train a dog to choose something on the basis of color. They can learn it - it's not that they don't see color, because they do - but it takes hundreds of repetitions for them to figure out what you want. "Oh, it's the green that you care about!" Suppose you laid out three balls for a human, and they got a reward for picking the "right" one - they just have to figure out the pattern you're looking for. Now imagine the correct answer is the ball that smells like banana, rather than the one that smells like apple or orange. Think how long it would take you to figure that out, and you've got an idea what it's like to train a dog to choose colors.

Anyway, I do think the dogs are enjoying the game, so I'll have to think of new ways to keep it interesting for all of us.

No comments: