Sunday, July 14, 2013

Great Book for Understanding Your Dog's Noodle

93% of a Dog's Brains are in her Nose
Most dog books are about picking a dog or training your dog or fixing dog problems.

This one is not one of those. It is entirely about how dogs understand their world ("dog cognition") and the experiments that reveal their nature.

You see many dog books by PhDs (sometimes in a subject having nothing to do with dogs). I've read several and they tend to follow a pattern. Credentials. Some observations with experiment to back them up. Then, very quickly, they fall into conventional wisdom, anecdotes, opinions, and other "facts" that are justified not by science and experiment, but purely by the authority of the author.

Almost everything said about dogs in "Genius of Dogs" is based on rigorous, repeatable and often ingenious experiments. If there's only a single experiment backing up an observation, the authors tell you it's possibly questionable till someone else can repeat it. That's honesty! That's respect for your reader! Dog books don't usually admit to their fallibility.

The "Genius" in "Genius of Dogs" refers to those mental abilities dogs have that are truly exceptional in the animal kingdom. The two that really stood out:
1. Dogs are brilliant (compared to even primates) at recognizing human direction (like pointing and facial expression). What's truly amazing is that this does not seem to be a learned thing. As young puppies, relatively isolated from human contact, they still pick up on gestures like pointing. Wolf cubs and chimps can't without a lot of training.

2. Dogs aren't unusually gifted at solving problems like mazes or getting around obstacles or opening latches. In fact, they seem to be not all that bright in some experiments. But if you SHOW them how to solve a problem, they will pick it up almost immediately. They might not know how to go around a fence to get some treats, but once they see another dog or a human go around, they get it. I've seen this brilliance at copying behavior with my own dog. I spent a day and a half laboriously teaching him to go through a dog door with treats and luring and what not. He slowly got it. But with another dog who couldn't figure out the dog door either, when he saw my dog, Pogo, jump through a couple times, he figured it out instantly. 30 seconds of training vs a day and a half! Learning by example is a dog thing. Learning by conditioning (clicker/reward/positive or Alpha/do what I say cause I say so and I'm the boss both work, but aren't playing to the dog's mental strengths).

The book is essentially a list of experiments and results strung together with some personal anecdotes from the (apparently) main author, Michael Hare. The anecdotes are mostly benign and sort of amateurishly told compared to the meat of the book. But they are harmless and tend to relieve what might otherwise be a moderately dense book. (It's not a text book, but it is a book filled with explanation and evidence).

There is a small section on training at the end. Mostly it's critical of current training methods for following classical conditioning/behaviorist/operant clicker methods. The main point is that dogs are not black boxes and their best way of learning is not from simple action/reward/punishment, but by using their strengths (learning by copying and by reading human gesture). That's not completely fair because all clicker training (and alpha training too) emphasize the human connection with the dog. The dog must give the trainer his/her attention. The dog is not put in black box isolation from other stimulus. Still, the authors' point that training could better take advantage of dog Genius is well taken and could lead to some remarkable new training techniques if some human genius could work out a system...

No comments:

Post a Comment