Saturday, August 6, 2011

What you need to know about classical conditioning

Classical conditioning is a basic concept in training so let's look at it for a few minutes.

Below is a rather humorous video showing a college psychology student demonstrating the use of classical conditioning on his roomate. He is using it DIFFERENTLY than we are, he is using it to establish that a sound means that the roomate is about to be shot with an air pellet. Warped, I know. But the demo is clear.

Classical conditioning happens in our world (and our dog's world) whether we like it or not. It answers the question, "What is predictable in my world?". When we pick up the leash, it means we're going for a walk. When we open the container where we keep the dog food, it means its supper time. These are all associations the dog makes through repetition. We present a neutral or meaningless signal (the click) prior to the delivery of the treat, and after enough reps of this the dog comes to expect the treat as soon as he hears the click. See how the sequence is important? First the click and then the treat.

You know how your dog learns to anticipate things? You pick up your car keys and he runs to the front door? This is classical conditioning. He's come to associate your keys with the door opening.

With clicker training, we click, and then we deliver a reward.  The "conditioned response" on the part of the dog is the EXPECTATION of the reward arriving, and the happy feeling that goes with that.  The response we conditioned there was one of anticipation of the reward.  (In Pavlov's dogs, the anticipation of that reward also resulted in an automatic salivating response, so in their case the salivating also became part of the conditioned response).

This is useful in that we can click at the moment of good behavior, and instantly elicit that feeling of "reward coming!" in the dog.  THEN we give the reward.  This allows us to tie that happy feeling with the exact moment of good behavior, which then allows us to work at a distance from our dog.  In other words, it buys us a moment or two before we have to actually pay up!  But it still cemented that FEELING of the reward with the exact moment of good behavior.  This kind of laser-beam timing helps us shape behaviors very specifically.

Different behaviors or "conditioned responses" take different amounts of time (and repetitions) to condition.  Some happen quickly with just a few reps, and some take hundreds of reps.  It helps to create a nondistracting environment when you intentional create a classically conditioned response-- that way the signal and the event that happens next are easily identified and connected in the subject's mind. 

Imagine if you were at a noisy dance club, and suddenly you felt a drop of water on your forehead.  What happened right before that?  Was it a particular note of music?  A particular dance move nearby?  A flashing light in the club?  Who knows?  There's too much going on.  But if you were in a quiet empty dance club and a light flashed and then you felt the drop of water, and that happened 5 times in a row, you'd understand the connection much faster.  You have to be able to identify those two events in connection with each other.  If the environment is too distracting, it's hard to isolate the events in your mind enough to identify them as being connected.

Think of this when you're trying to give your dog a cue.  Pay close attention to your body language-- beginner trainers often have (unintentionally!) sloppy body language and they THINK they are giving a clear cue but their cue is inconsistent.  (This is where it really helps to have a trained eye watching you and giving you feedback-- or videotape yourself on your digital camera and play it back to see what you're REALLY doing). 

It can be a little stressful for an animal (or a human) when they suspect there is a pattern but they are having trouble identifying what it is.  It is a relief and a comfort when we make classical conditioning easy by making the first and the second event very clear and very easy to connect. 

One very cool application of classical conditioning happened with a young girl who was gravely ill.  The last resort for her was a very toxic drug, and they wanted to see if they could "train" her immune system to respond as though the drug were present, without having to actually keep GIVING the drug.  They paired the drug with a strong smell and strong taste each time they gave it.  The smell and the taste were very distinct,  and were not commonly found in the child's life up to that point.  After a few reps, they gave the smell and taste again, but secretly withheld the drug.  Her immune system responded again (favorably) as though they had administered the drug.  This showed that our immune systems can be conditioned to respond favorably (or unfavorably) to external stimuli.  VERY interesting implications. (I learned of this case on PBS). 

This week, pay attention to classical conditioning and where you see it in action in your life and the lives of those around you.