Dog training: What is positive-reinforcement (R+) dog training?

So, I've already talked about some different types of traditional punitive training techniques in my previous post.  I'll move right along with the more modern, science-based techniques I mentioned: positive-reinforcement (abbreviated as R+) training and clicker-training.  A lot of people seem to think these methods are a little out of reach or difficult to understand, but I hope to clear them up a bit right here, right now.

Positive-reinforcement is ultimately very simple once you get the idea.  It's all based on Pavlov's theory of classical conditioning and B.F. Skinner's operant conditioning.  If you aren't familiar with the famous Pavlov's dog experiment, I'll summarize it quickly for you.  Ivan Pavlov, a researcher, conducted an experiment by ringing a bell prior to each feeding the dog in the experiment was given.  Over time, he observed that when the bell rang, the dog began to salivate, ultimately anticipating the oncoming food before it was ever presented.  The bell had become a condition stimulus that the dog associated with food.

This basic concept, which is really what positive-reinforcement training is all about, is that if you condition a dog by creating an association between a conditioned stimulus (CS) (i.e. the bell in Pavlov's experiment) and a reward (i.e. food/treats), the previously neutral stimulus (the bell) will now elicit a conditioned response (i.e. salivating/expecting food).  It's science!  B.F. Skinner furthered this idea by discovering that you can shape behavior once you are able to condition animals.  

Have I lost you?  I hope not.  Here's where it all comes together.  Small devices that make a "click" noise when depressed are convenient and can be used in place of a bell.  That's clicker-training.  You can also replace the click with the word "yes" or many other stimuli, for that matter--it's a matter of preference, though many trainers prefer a clicker because of the consistency and recognizable sound.  First, you create an association between the CS of your choice and the receipt of treats (also referred to as "charging" the clicker in clicker-training, by clicking once and immediately giving the dog a treat, and repeating this several times in several different charging sessions).  Once the dog makes the association that the CR means that treats are coming, you can communicate to the dog very quickly and efficiently when they've done something you want.

One misunderstanding many people have about reward-based training such as this is that the dog must become dependent on treats in order to behave.  This is incorrect, because whatever reward you are using to teach the behavior, be it treats; praise; toys; or other things the dog finds rewarding; they are only used during the beginning stages when first teaching a particular behavior.  They are phased out over time, and in fact, if you were to continue rewarding the dog each time they successfully perform the behavior, the training would not be as effective.  Once the dog understands the behavior we are asking of it, we use something called an intermittent reinforcement schedule, which, in layman's terms, basically just means changing it up and not giving the same reward each time.  Think of it in terms of gambling: the idea that you might win something makes it that much more appealing to do.  Your dog is essentially becoming a gambling addict (in a good way!).

You can train very complex behaviors with this type of training.  For an example of the amazing things that are possible with clicker training, please check out the video by one of my training heroes, Emily Larlham (and her dog, Splash), below.

Splash the Border Collie!

The benefits of positive-reinforcement-based dog training are many.  In this type of training, you're teaching the dog the think for themselves—they're getting more from the training than they would with punishment-based methods.  Dogs tend to really enjoy this type of training because they're engaged and having fun.  If you're using treats with a dog that loves food, then the dog will enjoy food in return for doing things you ask.  If you have a dog who loves to play with toys, then the dog can learn that doing things you like brings play sessions with toys and they'll be more inclined to work with you.  In the case of learning loose-leash walking, moving forward is the reward, so if you reward a dog by moving forward only when the leash is loose, they quickly learn to stop pulling!  It's a beautiful, effective method that is humane, fun, and lasts longer than simply punishing a dog.  The process of learning the behaviors you teach them is so much more participatory, rather than just telling them they did something wrong, you're being more proactive and productive by telling them what to do instead.

Edit 10/2013: A nice video example of the difference the dog experiences between these two training methods is linked below.

http://www.youtube.com/watch?v=yQAayRtXkwE