The rate of reinforcement is the number of times you reinforce during a given period of time. Our program starts with a very high rate of reinforcement meaning we reinforce quickly and often.
The rate of reinforcement can and should be reduced, but there is a process that must be followed in order to keep the dog motivated and interested in the work. Here are some principles that should help with shifting your dog from working for food to working for the love of the work.
- Fade luring as soon as the dog can offer a behavior remotely close to the desire behavior..
- Fade reward placement as soon as the dog can finish a behavior reasonably well without reward placement..
- Get the food off your body as soon as the dog no longer needs reward placement, is working enthusiastically, is consistently correct at the behavior, and no longer needs rewarding for every correct attempt at the behavior. When getting food off your body, you’ll move from giving small pieces of food fed from your hand to larger amounts called jack-pots given in a bowl.
- Never remove reinforcements all together.
- Reduce reinforcements in such a way that the dog continues to think that a reinforcement could come at any time, especially when the dog puts forth great effort.
- Instead of gradually reducing reinforcements in a linear fashion, move your reinforcement schedule from (1) consistent and predictable to (2) selectively random and then finally to (3) unpredictably random. The following sections will show you how.
Step 1 – Continuous Reinforcement
Use continuous reinforcement when a dog is first learning a behavior. Continuous reinforcement means that you are reinforcing every correct response. Reinforce every behavior until your puppy has met the following five milestones with the behavior.
- Luring is no longer being used.
- Reward placement is no longer being used.
- The dog will work for jack-pot rewards with the food off of your body.
- The behavior is solid with required criteria.
- The behavior is solid in the environment you are training.
- The behavior can be chained either with itself or with another behavior. Chaining a behavior means to perform two behaviors in a row without reinforcement in between. For instance, if I ask my dog to down and then to sit without reinforcement in between, I would have chained two behaviors together. The two behaviors can be the same behavior performed twice or it can be performing two different behaviors one after another.
Step 2 – Selectively Random Reinforcement
When you have met the above five milestones, you can begin what we call selectively random reinforcement. Selectively random reinforcement is a reinforcement schedule that is neither consistent, nor truly random. Every correctly done behavior is no longer reinforced. Instead, you are moving toward random reinforcement. However, instead of using a truly random pattern for determining when to reinforce, there are very specific guidelines to making your reinforcement schedule more effective than arbitrarily random reinforcement.
Guidelines for Selectively Random Reinforcement
- Instead of reinforcing every correct behavior, reinforce only the dog’s best efforts. Effort must be reinforced while at the same time mistakes are not. Therefore, the dog must be set up in such a way that when a dog is putting forth his best efforts, there is a very high probability that he will be correct.
- When chaining together multiple behaviors, have the most difficult behaviors at the end of the chain so that the difficult behaviors are rewarded more often than the easier behaviors. When you chain behaviors together, only the last behavior is actually being reinforced.
- Begin using jack-pot rewards when you begin chaining multiple more complicated behaviors together between reinforcement. These are larger rewards that are given at least several yards away from where you are training. With a jack-pot reward, the dog hears the marker word and then runs to a designated place where the trainer gives him a small bowl of food in a crate or on the floor or ground. Because the dog is randomly being asked to work longer sequences and because the food is no longer close by, he needs a larger reward than a single treat or a few pieces of food to keep him motivated.
- When the dog has struggled with a particular behavior, reinforce it every time they get it right until they are getting it right consistently. If a dog makes a mistake, make sure that during his next try at that behavior, he is correct and gets reinforced for it. Make the exercise easier if necessary.
- Most importantly, be careful not to train over threshold. If your dog is failing, it is usually because he is not being reinforced enough. When mistakes are made, go back to continuous reinforcements (step #1) until your dog has met all five milestones again for moving to selectively random reinforcement.
- When your dog is almost perfect with the behavior, move to truly arbitrarily random rewards. At this point in training, it will be difficult to ascertain best efforts or most difficult exercises because your dog rarely gets it wrong. Therefore, reward randomly. Keep your dog guessing. Keep him thinking that the reward could come at any time as long as he is putting forth good effort and doing the exercise correctly.
Step 3 – Truly Arbitrarily Random Reinforcement
Once behaviors are truly learned to perfection, you may be wondering why you’d need to practice them at all. Why not spend training time learning something new or practicing those behaviors that aren’t up to your standards and need work? There are actually several reasons for practicing these already learned behaviors. We’ll discuss two of them.
First of all, if a skill is never practiced, the ability to perform that skill will begin to lessen.
Secondly, many of the simplest and most elementary skills are often performed in the context of performing other skills. These simple skills will get sloppy if not occasionally reinforced.
An example is a good set up in heel position. If the perfectly learned set up is not randomly reinforced, the dog will quit trying to set up in good heel position. He’ll do a sloppy set-up thinking that the reinforcement will never come prior to starting the meat of an exercise. Remember, the dog must think that the reinforcement could come at any second. That includes the very first second! Therefore, every single moment of any work needs to be reinforced at least on an occasional random basis.
