This week’s Science Dog essay is an excerpt from Chapter 8 of “Dog Smart: Evidence-based Training with The Science Dog“.
I introduced the previous chapter with a story about starting each orientation class at AutumnGold with a version of the training game. While I emphasized that our students are usually impressed by these demonstrations and immediately catch on to the power of positive reinforcement, I would be remiss to not mention that we do see the occasional “frownie-face” in the audience during these demonstrations.
What I am referring to is the human version of this:
That face, human form, tells us that the student expressing it is not convinced and is usually taking umbrage with the use of food treats to train dogs. Mr. or Ms. Frownie-face invariably raises a hand to utter some version of the following:
“I don’t want to use food with my dog to train him because I want him to work for me out of love [or respect, or because I am alpha, or because I am King Tut, ruler of the world]”
Okay, maybe I made that last bit up. But you get the picture.
While we get the frownie-face and the resistance that accompanies it less frequently than in the past (thank you positive trainers!), we still see it now and again. So, in this chapter we explore evidence for staying, as much as possible, within the positive reinforcement (+R) quadrant of Skinner’s four consequences. I also will provide a means for communicating this information to the doubting Joes, Josephines and Frownie-faces of the world when you encounter them as clients, in classes, or as your neighbors.
Training in the +R Quadrant: I don’t think it is an outrageous claim to assert that the vast majority of people do not want to harm their dogs, either physically or emotionally, in order to train them. Unfortunately, a substantial number of dog owners continue to think that using punishment is the only effective and reliable way to train dogs. These beliefs may arise from continued reliance upon “dog-as-wolf” myths that tell owners they must establish dominance over their dogs, or upon the view that using positive reinforcers in training is synonymous with bribing. (These beliefs are false, as Joe finds out at the end of this chapter). For now though, let’s look at what we know to be true about the aversive control of behavior, commonly referred to as “correction-based” training, versus training methods that focus primarily on positive reinforcement, commonly referred to as “reward-based” training.
Correction-Based Training: Aversive training methods, even if “balanced” with positive reinforcement, rely upon a dog’s natural desire to avoid pain and discomfort. The dog pulls forward into his leash; a collar jerk occurs; the dog moves back into a loose-lead heel position to avoid the discomfort. If a consequence is not sufficiently unpleasant, the dog has no reason to change his behavior to avoid it and learning does not occur. Therefore, by its very definition, a training approach that relies partially or fully on aversive consequences involves causing some level of discomfort or pain to the dog.
In addition to the discomfort that this approach relies upon, there are emotional costs. The basic emotions associated with pain and discomfort in dogs (as in humans) are fear and anxiety. Although proponents of correction-based methods argue that anxiety and fear can be minimized by using the mildest intensity of an aversive that is necessary, there is no evidence that such a level exists. Rather, all of the studies that have examined the use of aversives to control behavior in dogs have reported signs of stress and/or fear as direct results of these training methods (see following section in this chapter for details).
A third problem with reliance upon aversives in dog training is that the exact nature of a dog’s response is not always predictable. Although some dogs move away from an aversive stimulus if there is an escape route available (for example, a dog stops pulling into a corrective collar), others may freeze in place, panic, attempt to run away, or become aggressive. As a result, the risk is that the response of the dog is not always what was intended by the trainer. This is a common problem because applying an aversive only provides the dog with information about what NOT to do, but does not provide information about what TO do. Essentially, the dog is forced to learn through the process of elimination. Negative reinforcement relies on the dog’s ability to select the desired behavior that will allow her to escape or avoid the aversive. Because a variety of behaviors are often equally successful in avoiding an unpleasant consequence – for example, running away or showing aggression – the behavior that is elicited each time a correction is applied may not be the behavior that the trainer was expecting to see.
Finally, because stress is often introduced with the use of negative reinforcement and punishment, the use of correction-based training as a humane approach to training is questionable. In addition to the potential for intentional or unintentional abuse, aversives that are associated with the owner have the potential for damaging the relationship between the dog and his owner. The overuse of aversives or using corrections that are too harsh can cause generalized fear and avoidance as the dog may learn that one behavior that will allow him to avoid discomfort and fear is to simply avoid being near his owner.
No one wants this. Why take the chance when there are better ways? (In the remainder of this chapter, we explore these better ways along with the evidence that supports their use).