Tapering Off Food Rewards

The advice often given by positive trainers to avoid the undesired effect mentioned in my last post ("How to Stop GOOD Behavior") is to taper off reinforcement (rewards).

The advice is that, gradually, you give less and less food. On some repetitions, even though the dog did the right thing, and you want to pay it with food, you just don't. You cheer, you celebrate, you pet, you massage, you allow access to other rewards, but you don't feed.

This is a hard thing for most pet owners to do. Their dog has done something, and they want to pay it. Which is a fantastic breakthrough from the old-school training days of NEVER feeding! Nothing wrong with that inclination.

And, it bears stating: If you can be committed to continuing your reward schedule of 100% -- go for it!! There is absolutely no reason for you to stop. There are too many people who are ashamed of using food to encourage their dog's behavior. Trust me, the general public doesn't give a rat's behind WHY your dog behaves--only that he DOES!

One way to continue to increase good behavior while decreasing food is to escalate criteria. This can be difficult for most owners to recognize at first. They are so excited that the dog complies at all that they can't imagine more!

Escalating criteria automatically tapers off the rewards, because your dog will not meet the new criteria 100% of the time.

What does "escalating criteria" mean?

"Criteria" means the precise behavior (usually multiple behaviors for beginng dogs) that earns the reward. The criteria can be thought of as the dog's "definition" of the behavior.

"Escalating criteria" is redefining the behavior, making it harder for the dog. It means you expect a teensy bit more than what you used to expect. You are making a new and more specific "right answer".

For example, you reward faster reponses -- ones within 2 seconds instead of 5. Stays are now with distraction. Or more tempting distraction. Or you only reward the down position if both rear feet are on the floor (Sphynx position), not if the hips roll over to the side. You reward ONLY recalls where the dog sprints as a response to your call, instead of any response at all.

Choosing correctly escalated criteria is what "separates the the men from the boys". Make it too hard, and the frustration is punishing. Make it too easy, and the boredom is punishing. Just to make it even more complicated--there is no one "recipe" of the right criteria for every dog!

However, choosing criteria that the dog gets right about 75% of the time is a good place to start. This is, however, just a suggested rule of thumb! No magic. If you and your dog have the patience and drive to attempt something the dog only gets right 10% of the time, it doesn't mean it's wrong. However, choosing criteria that keep a high success rate (high rate of reward) will keep your and your dog's interest and enthusiasm high.

Once you have chosen your new criteria, you STOP food-rewarding anything that does not meet that new, exact definition. It is important to note that this failure to provide an anticipated reward could be called "negative punishment" by your dog. This negative punishment is what conveys the idea that the lesser responses no longer consitute the "Right Answer".

Should you ignore the NOW undesired, but still very good answers? NO! Acknowledge that those formerly right answers are STILL right-ish. Provide encouragement and appreciation-- pet, praise, get excited. Use your dog's lower-value rewards.