Operant And Classical Conditioning

OPERANT CONDITIONING

Operant Conditioning has some potentially confusing expressions - especially when viewed in the light of "regular" language. So this page hopefully will help illustrate what they mean.

 First thing to answer is: "What exactly is operant conditioning?" It is delivery consequences based on the actions, decisions, and/or choices the operant (in this case, a dog) performs. The consequence is best delivered right as the behavior is either performed (for wanted behaviors) or just before it begins when the dog is thinking about performing the behavior (for unwanted behaviors). 

When applying these consequences, we can add things to the dog's environment (or to the dog himself) or take them away from the dog's environment (or the dog himself). Adding is termed "Positive" while taking away is termed "Negative". Notice how the connotations of these words that exists in "regular" language is no where in the definition.

In addition, we apply consequences with the goal of either increasing a behavior's frequency or decreasing a behavior's frequency. If the goal is increasing a behavior, that's called "Reinforcement". If the goal is to decrease a behavior, that's called "Punishment". However, the caveat to this is that it depends on the dog's point of view as the dog is the operant and the operant is the focus of this method. What this means is that you have to recognize how your dog considers a consequence.

For example, Wally jumps on me and I pushed him down. In most humans' minds, this would be punishment, especially if I pushed him down hard. However, he could see it as continuing the game, especially if he likes physical play. What I thought was punishment was actually reinforcement! He will be more likely to jump on me in hopes of getting another push and another game.

On the other hand, if I turn my back to him and ignore him until he gets off me, that is more effective as I'm denying him what he wants, which is my attention. For a dog like Wally, that tends to make the point. For another dog, however, it might be ineffective. The dog might take it to mean "oh hey, I can jump and nothing will happen to me!"

The dog's point of view is what counts, not what we intended to do.

So now what? We have "Positive" and "Negative" and "Reinforcement" and "Punishment". Well we can combine them as follows:

(Examples in italics inside parenthesis. Remember these are just examples.)

Positive Reinforcement: Something is added to the dog/dog's environment that the dog likes and wants to have happen as a result of a behavior. (I say "abres" and Wally opens the door. I praise and give him a treat. Wally will be more likely to open the door upon hearing "arbes" next time.)

Negative Reinforcement: Something the dog doesn't like is removed from his enviornment (bad thing goes away) as a result of his behavior. (A trainer says "sit" and the dog gets his ear pinched until he sits, then the pinch stops. The dog will sit more often upon hearing "sit" because it keeps the pinch away.)

Positive Punishment: Something is added to the dog which he does not like as a result of his behavior. This is sometimes called "applying an aversive" or "using an aversive on the dog". (If Wally could reach the counter and put his paws on the counter and I sounded an air horn right as his paws touched the counter, Wally would be less likely to put his paws on the counter.)

Negative Punishment: Something the dog would want to do or is enjoying is removed from the dog as a result of his behavior. (Wally wants to greet me and jumps up on me uninvited. I turn away until his paws are on the floor. Wally will be less likely to greet me by jumping uninvited.)

Extinction: This isn't a consequence, but a process or sometimes considered a result. This the dog giving up on a behavior in a situation because it never works out for him. It can be an effective way to have the dog "delete" a behavior, but it must be consistent. Otherwise, you're just making the dog want to keep trying for longer before giving up. Extinction is almost always resisted by the dog, and often results in what's called the "Extinction Burst" where the behavior gets taken to the next level or twenty in a desperate effort to make it work. There IS a good use for extinction bursts (shaping in particular even attempts to trigger extinction bursts as it can create new behaviors).

CLASSICAL CONDITIONING

As good and wonderful and powerful as operant condition is, it can't do everything. Classical conditioning has a very powerful impact on training.

So what is classical conditioning? It is the association of one stimulus (event) with another such that the first stimulus predicts the second. The famous "Pavlov's Dog" is a classic example of this principle. At first the bell was rung, then the dog had food presented to him. This is repeated constantly and eventually, the dog would anticipate the food just on hearing the bell. The association is complete.

This is a very powerful tool! If you've ever heard of, or performed, what's called "charging the clicker", that is classical conditioning. At first, you just click the clicker and give a treat...over and over and over again. You know you've burned it in the brain when the click has the dog immediately looking for the treat (and preferably with a quick movement, indicating a triggered desire). Congrats, you just developed a reward marker - a VERY useful thing in training.

Yes, markers are created via classical conditioning - reward, no-reward, even punishment markers are possible thanks to classical conditioning.

Classical conditioning is also a capable way to change emotional responses. In fact, it's pretty much the best way to modify emotional responses. The reason is that you are "teaching" a new emotional response to the trigger. Say a dog is scared of other dogs. If you give treats while the dog is aware of, but not overwhelmed by the other dog, over time, the fearful dog can change his expectation. This can be done with scary objects as well.

As you can see, classical condition has many uses that only it can achieve. It can also enhance the usage of operant conditioning, thanks to its ability to develop markers. Both operant and classical condition are always at work and strengthening each other. Every time I give Wally a click and treat, I'm reaffirming that the click should predict a reward as well as whatever behavior that caused me to do the click in the first place. I'm also building positive emotion in the interaction with me, the process of training, and the behavior itself will become a marker/association with a reward eventually, trigger the same emotions as getting the actual reward. Remember the bell? The dog never got food - his mind and body just reacted as if he did. That's the power of classical conditioning, and combining it with operant condition makes trainers a potent force in a dog's life.