Clean Run Magazine - November 2008 - 35

Stress-free Learning Psychologists once thought that in order to learn a discrimination animals had to link the correct choice (the S+) with reinforcement and they had to link the incorrect choice (the S-) with a lack of reinforcement. This last part can be pretty stressful because animals don’t like things that signal no reward. They often get really upset when they see the S-. Pigeons turn away and flap their wings angrily. Rats attack a cagemate. If the S- is a thing, rats will kick shavings over it. Dogs whine and move as far away from it as they can. By all indications, animals come to view the S- as aversive and so they find trial-and-error learning stressful. H e r b Te r r a c e , a t C o l u m b i a University, decided he could teach a discrimination without the stress. He called his procedure errorless learning. Think back to the bear/cow discrimination. Suppose you want your dog to choose the bear. First start off with a very large bear. Every time you present the bear you make it really exciting and impossible to ignore so your dog responds to it and you reinforce him. Just as often, present the cow but the cow is teeny tiny, and you don’t move it or make it exciting. Hopefully, he won’t even notice the cow and so he doesn’t choose it—he waits for the more exciting bear. Progressively, over many repetitions, you make the cow larger and slightly more exciting and the bear smaller and slightly less exciting. But because you’ve established a strong habit of choosing the bear and not the cow, the dog continues to respond to the bear. He might notice the cow occasionally because you are gradually fading it in but he’s unlikely to choose it. Little by little, you end up with the same-sized bear and cow. Throughout the process, your dog has made few, if any errors, so he’s unlikely to experience stress. He also has no “baggage” associated with the cow; he doesn’t dislike it—he just doesn’t choose it. There is no published data on dogs trained by errorless learning. But a pigeon learning to choose between a red and a green circle via the traditional trial-and-error method will make between 2,000 and 5,000 errors before becoming proficient. C o nt r a s t t h at w i t h e r r o r l e s s learning—the pigeon learns more quickly and makes, on average, about 25 mistakes. The major limitation of errorless learning is that it is only appropriate if you never want your dog to respond to the one you’ve designated as incorrect. Through the fadingin process, the dog learns to pay attention to the bear and ignore the cow. With standard trial-and-error learning, he learns to avoid the cow because he associates it with nonreward, but he still pays attention to it. If you now change over from the bear to the cow, the dog will figure out the switch reasonably quickly. Errorless trained animals almost never master a swap. So errorless learning is only useful in agility for those situations where you never want your dog to make one of the choices, like entering the weaves incorrectly or lying down underneath the table rather than on it. Learning about the Choices Some trainers advocate introducing the dog to one choice, for instance, teaching the left turn, and, when the dog reaches proficiency, switching to © CLEAN RUN Dog Tested, Human Approved Agility Flooring © WWW.MATTALLISONPHOTOGRAPHY.COM RB’s canine rubber mats are used by top trainers to provide a durable, resilient surface for dog play, agility practice or competition. Anti-fatigue qualities reduce stress on joints, back, and muscles while providing a non-slip surface. Fewer Injuries • Added Traction Joint Protection • Easy to Clean The ﬂooring choice for the 2005 and 2006 AKC Agility Nationals 800-525-5530 • www.rbrubber.com Up to 97% post-consumer recycled content National Distribution Centers November 08 | Clean Run 35

http://WWW.MATTALLISONPHOTOGRAPHY.COM http://www.rbrubber.com http://www.rbrubber.com

Clean Run Magazine - November 2008

Table of Contents for the Digital Edition of Clean Run Magazine - November 2008