Generalization and Discrimination – Two Important Concepts

These two concepts really shed some light as to how dogs learn. They are important for us to understand because it will help us feel less frustrated when things with our dog are not going our way and to maximize learning.

Dogs do not generalize:
Here is an example of what generalization means: If I see my friend Candy at the grocery store, I still “remember” that Candy is the friend I met when I joined the book club a few months ago. Different contexts, but I can still “piece together” that Candy is the same person I met at the book club and now she is at the grocery store. We can do this so easily that we think dogs should too!

Here is where we get into trouble when we ignorantly believe dogs operate the same way. I teach a dog to sit or lie down (or whatever the case might be) in the kitchen.

The dog is performing the behavior quite well, so I surmise that he has really learned sit, etc. Now, I go to take the dog for a walk and ask him to sit before crossing the street but he does not! Well, what happened? Short answer: dogs do not generalize.

From the perspective of the dog, sit happens in the kitchen, not when going for a walk unless of course, I also teach it when going for a walk… Ahh… it all starts to make sense, right?

So what do we need to do in order to make sure that the dog sits when going for a walk, saying hi to a stranger at the arts fair, etc? Well, we must practice in those contexts until the dog has learned that sit actually can take place in the kitchen, the sidewalk, and when saying hello to people.

Lack of generalization, by the way, is one of the many reasons why dogs “work” for the trainer but not for the owner… the dog must generalize the new learned behavior with the owner.
So when my client tells in frustration (or perhaps embarrassed?) that her dog should do “x”, “y” or “z” behavior because she has seen the dog do so at home I tell them that the dog is under-trained under the new circumstances. There might be, of course, a million other reasons why the dog is not performing, but generalization is often one of them.

 

Discrimination:
Dogs are MASTER discriminators! They are so good at this that often professional trainers must find ways to “beat the dog” at this game … Think about it. You must become a very good observer in order to very quickly assess if someone or something is friend or foe because your life depends on it. That is reason #1.

Reason #2 has to do with domestication. There is a lot to say about this. But suffice to say, that with the process of domestication, came along the fact that dogs are always observing us (okay, they do take long naps – so when they are not napping…). They pay close attention so that they do not miss their dinner, or the walk, or a chewy or snuggle time on the couch with you. In other words, they are looking out for #1.

Try this experiment at home: When you are around and your dog is sort of relaxing or even kind of snoozing do something that you know your dog has associated with either getting a treat, go for a walk, etc. Say, for example, that you keep your dog’s leash in the closet by the front door, so now, when your pup is minding his own business open the closet door. Did your dog open half-mast his eyes to ascertain if the leash is now in your hand? Did he come into the room just seconds after he heard the closet door opening?

The same happens when we use poor mechanics in training and we reach for the treat before the dog has finished the behavior we are planning on rewarding. It does not take too many of these bad moves for the dog to now stare at the treat bag. The dog has learned probably with one trial where the good stuff is kept (of course he can also smell it) but he has learned that we can produce the good stuff by reaching into the treat bag and as a result of this, we now have a dog whose eyes are glued to the treat bag…

The good news is that now the place of discrimination in learning we can use it to our advantage. Because of discrimination, dogs can learn that this response and not that one gets them what they want.

If we become keen observers of what the environment in general, and our behavior means for the dog, we can much better manipulate consequences for them which, by the way, it is consequences- and not the cue (or command) that produces behavior in the first place and keeps it going strong once learned.

The same goes for generalization. By incorporating generalization in our dog’s learning we are “proofing” the behavior, which means they can actually perform where it matters.