Reinforcement schedules can make or break any training procedure. If the correct ones are used and at the proper time, the dog learns rapidly and maintains performance. If the incorrect one is used, your dog will only get bored, and his performance can deteriorate.
While there are many reinforcement schedules, we will look at only the most important for us in training dogs. The two broad areas are the interval and ratio schedules. A fixed interval schedule is based on time, and the animal is rewarded after a certain period of time after having been correct. In a variable interval schedule, the dog is rewarded after varying periods of time. In training this is not applicable in most situations except possibly the stay exercises where the reward to the dog initially would come after fixed intervals of time and later at varying intervals to build reliability.
The most important for us to consider is the fixed and variable ratio schedules. These are based on a reward coming after a fixed or varying number of correct responses. If our purpose is to teach a dog a given response, for example, sitting, he is rewarded every time he sits. While this is the quickest way to teach a dog a behavior when the reward is withdrawn a dog trained this way rapidly becomes unreliable or, that is, the response extinguishes rapidly. Since we cannot always reward the dog each and every time he does as we ask, we must use a reward schedule which maintains reliability. This schedule is a variable ratio schedule. The dog is rewarded, for example, after every third, fifth, or ninth time. Since we are using no regular pattern, the dog does not know whether or not he will be rewarded for obeying. However, since he is rewarded sometimes his behavior extinguishes slowly when the reward is withdrawn.
How then do we switch from a fixed interval of rewarding the dog every time to a variable reward schedule? First we must establish when the dog has “learned” the behavior we are after. This is really up to the individual. Let’s say we are going to say the dog has learned to sit when he does it four times in a row, being rewarded each time. We now skip one, but reward the next two. The dog probably thinks you just forgot that time, but still responds. Next you might reward on the second, fourth and fifth response. By gradually extending the number of responses the dog must do before rewarding and varying that number, we gradually get away from a reward for each correct response.
I have purposefully avoided mentioning the type of reward since this is up to the owner. Each dog is different and what is a meaningful reward to one dog may not work for another. Therefore, each handler must find out what really turns his dog on.
©1980 W.H. Morrison, III