Avoiding Treat Dependency
Posted on January 14, 2013 by Leah Roberts
One of the most common complaints reward-based trainers hear is, “My dog will only do it if I have a treat in my hand.” Detractors of dog-friendly methods use this as evidence that “treat training doesn’t work in the real world.” But the fact is that treat dependency is a result of an error in execution of the method (operator error), and there are very simple fixes.
1. DON’T REACH FOR THE TREAT UNTIL AFTER THE DOG HAS PERFORMED THE BEHAVIOR AND YOU HAVE MARKED IT.
There are three parts of training a behavior. A – the antecdent, which is either a cue, prompt or lure; B – the behavior, such as sitting, and C – the consequence, which is the delivery of the food. “Marking” the behavior means letting the dog know the instant he performs it that he met the criteria for earning the food and that it is about to be delivered. The behavior can be marked by a clicker or a verbal “yes” or “good” (or “schmuck” or whatever – it doesn’t matter, as long as it’s the same word used consistently).
The procedure looks like this:
A – Say “sit” or use an already-trained hand signal (cue); or lure the dog into a sit with a piece of food; or prompt the dog into the sit by the use of hand targeting or a target stick.
B – The dog sits, and you mark it.
C – You reach for the food and deliver it to him.
Dogs pick up on visual cues very easily, usually more easily than verbal cues. So even the act of reaching for a treat before the behavior is performed can become a signal to your dog that there is food available. Being smart critters, they can quickly learn to distinguish between the times when you signal the possibility of food and times when you don’t. And that’s when you get a dog who will â€œonly do it if I have food in my hand.â€
If you are luring (using food held to his nose to manipulate him into position), make sure that you “fade the lure” after no more than 3 repetitions. Hold your empty hand in the same position as when you were holding the treat, but don’t actually reach for it until after you have marked the behavior. You can then slowly shape that hand movement (a prompt) into a hand signal (a cue).
To teach a dog a behavior like “sit,” I prefer to use the method of “capturing” instead of luring. This method initially employs only B and C, and the cue (A) is added later.
Step 1. Simply wait for the dog to sit on his own. When he does, mark it and toss a treat to him so that he has to get up to take the treat. (This resets him for another repetition more quickly).
Step 2. Repeat Step 1 several times, until the dog learns what behavior is earning him the treat and begins to offer it on purpose. Either he will look at you expectantly as he lowers himself into a sit, or he will sit and then swing his head to look at you as if to say, “Oh, hey, I did that thing – are you going to give me a yummy?” At this point, begin to give the cue (either say “sit” or give a hand signal) just as he is lowering himself into position. Continue to mark and toss the treat.
Step 3. After several repetitions of Step 2, begin to cue him (either saying “sit” or giving the hand signal) before he starts to sit on his own. Give him a moment to think about it – you probably won’t get it instantly. He has to process the information. When he sits, mark it and then reach for the treat. At this point, begin to deliver the treat to his mouth instead of tossing it. In fact if he stands up before you give him the treat, wait for him to sit again before giving it to him. This will begin the process of teaching him to hold the sit.
Done in a quiet place with no distractions, this whole procedure can take a matter of minutes. Now you have your dog sitting on cue, and the food has never been presented before the behavior has been performed and marked, avoiding treat dependency.
2. HAVE TREATS ON YOU EVEN WHEN YOU’RE NOT TRAINING.
Humans have 5 million scent receptors in their noses. Depending on the breed, dogs have a minimum of 125 million of these scent receptors. If you have treats anywhere on or near you, you can bet that your dog knows it!
The savvy dog can also become treat dependent if he can discern when treats are in the area and when they aren’t. If you have treats in your pocket when you ask him to sit, he can easily figure out that when you don’t have treats in your pocket and ask him to sit, he’s not going to get anything for it – and he may or may not perform the behavior.
To avoid this, simply keep the treats around, even when training isn’t happening. If you wear a bait bag when you train, wear it when you do the dishes or vacuum the house. Or keep some treats in your pocket, or the treat container sitting out. This way your dog will learn that the sight and smell of food doesn’t necessarily mean he’s going to get anything. The only reliable signal that a treat is available becomes his performance of the behavior.
Obviously once the dog performs the behavior reliably in all circumstances, treats can be phased out. But for a beginner learner, to get that behavior reliable, there should always be a “paycheck.”
One of the biggest mistakes new trainers make is phasing out treats too quickly. If the dog isn’t performing the behavior fluently, it’s not time yet. Once he is doing it reliably, fade them on a “variable schedule.” This means that sometimes he gets the treat on the second performance, sometimes on the fifth, etc. Mix it up so that he doesn’t learn to count (yes, they can do that!) and never knows when the treat is coming. (Note: If the behavior begins to degrade, you are rushing the variable schedule by asking for too many repetitions before reinforcing. Reduce the amount of responses required to earn the reinforcement, keeping it variable, and begin to stretch the schedule out more slowly).
You can also begin to use “life rewards” at this point. For example, substitute a “good boy!” and a scratch behind the ears for the treat, if this floats your dog’s boat. You can also ask for a sit before clasping his leash on, or before you open the door to allow access to the yard. Anything the dog desires can be used as a reward for performing the cued behavior.
Using these two tips, you can train using food without ever creating a dog who “only does it when I have a treat.” And reward-based training works for ANY behavior you want to train, be it a simple sit, fetching the newspaper, or running an agility course.
Share this:: Twitter