The Problem with Punishment

Anytime punishment is applied in dog training, you can see some unwanted side effects.  These unwanted side effects are often called behavior fallout in the dog training and canine behavior world.

Positive punishment is not positive in the way that most people think of it.  It is application of something that the dog doesn’t like to stop a behavior.  (To learn more about this concept, see our blog post How do dogs learn- Operant Conditioning.)  This type of punishment can be mild, like saying no in a sharp tone or very strong such a high level of shock.

One possible example of behavior fallout is a poor association with whatever the dog is looking at, at the time of correction.  For example, a dog and their family live in town and have an electric fence.  They begin to go to the boundary to greet passerbys and as a result receive a shock.  The dog learns, through classical conditioning, that people walking by cause the shock.  (To understand more about classical conditioning, please see our blog post Pavlov is at work with your puppy.)  Some dogs can learn this in one trial, others learn it in several trials and still some never make that association.  It always depends on the dog.

Using a mild aversive can have the same behavior fallout but it is less likely to occur or produce a strong reaction.  In this example, let’s say that when a dog jumps up on counters you yell “no”.  If a dog has a strained relationship with a person, this could cause a sensitive dog to run and hide or be apprehensive about approaching the person who issued the punishment.

 It should be noted here that a dog that is sensitive to sound may find yelling “no” highly aversive and a dog who is looking for attention and accepts any interaction as wonderful may find this rewarding.  Rewards and aversives are always defined by the dog and not the human.

Punishment has to be consistently applied to be effective.  In the example above, with the dog jumping on the counters, if you only catch the dog doing it 4 out of 6 times (for example), your dog may learn to play the lottery.  This means, he’ll give it a shot because he may get yelled at or he may not.  The possibility of rewards may be greater than the possibility of punishment.

More about punishment tomorrow.

Pavlov is at work with your puppy

Classical conditioning is defined as conditioning in which the conditioned stimulus is paired with and precedes the unconditioned stimulus until the conditioned stimulus alone is sufficient to elicit the response.  This phrase was coined by Ivan Pavlov and is always at work in dog training. 

It is incredibly important to realize that while dogs do learn by operant conditioning, classical conditioning is working too.  Dogs can associate objects or people as good or bad.  For instance, if everytime your dog is around children your dog gets kicked in the face the dog learns to associate children (neutral stimulus)= kick in the face (bad thing).  Eventually, this dog may see the situation like this: children are a bad thing.  The children are no longer a neutral stimulus and worse they are a predictor of bad things.

If everytime your dog is around children and they drop food on the floor (as children often do) then dog learns that children=yummy food on the floor (good thing).  Overtime dogs learn that children are a good thing.

Let’s look at another example.  When Rufus the terrier was a puppy, he saw a man a couple times with a beard that harassed him.  This bearded man poked him, made him bark and laughed when he tried to run away.  Now when Rufus sees a bearded man- he thinks that they are going to harass him.  The bearded man (which was a neutral stimulus) has become a predictor of bad things to come (a conditioned stimulus)

If Rufus’s experience were different where everytime he met bearded man he got a really yummy cookie, the bearded man (which was a neutral stimulus) would be a predictor of good things (a conditioned stimulus).

For some dogs it takes many bad experiences to react to a situation and for others it only takes one.  When you are socializing a puppy, you are working to make positive associations with new and novel people, objects, etc. 

Classical conditioning does not work alone.  Operant conditioning is always at working the background too which makes it sometimes confusing as what to reward and what not to reward.

Dogs learn through Operant Conditioning

B.F. Skinner coined the term operant conditioning and used it to describe how consequences affect an organism’s behavior.  Skinner believed that an organism operates his environment and figures out what works best for them.  Dog trainers have used operant conditioning for years to change behavior in dogs.  Operant conditioning is always working in the background with your dog whether you are actively training or not.  A dog trainer or behavior consultant should have an excellent understanding of these concepts while the average dog owner needs only a basic understanding of these.

There are three main components of Operant Conditioning:

1.       Reinforcement-  Something that makes the behavior occur again.

2.      Punishment-  Something that causes a behavior to become less frequent.

3.      Extinction-  No consequence to behavior at which point the behavior becomes less frequent.

The first two components can either be positive (adding something) or negative (taking something away).  The four quadrants of operant conditioning are as follows:

Positive Reinforcement is adding something the dog likes to increase the frequency of behavior.  An example of positive reinforcement is giving a dog a treat for sitting. 

Negative Punishment is taking something away that the dog likes in order to decrease the frequency of behavior.  An example of negative punishment is not giving a dog a treat until he does sit.

Positive Punishment is adding an aversive (something the dog doesn’t like) to decrease the frequency of behavior.  For instance, a dog is off leash and starts to chase a squirrel.  A person applies a shock from an electronic collar to the dog. The dog stops chasing the squirrel.

Negative Reinforcement is taking away aversive stimuli (something the dog doesn’t like) to increase the frequency of behavior.  You generally need to apply positive punishment to get the effect of negative reinforcement.  In the above example a dog started to chase a squirrel and a shock was given.  In order to use negative reinforcement, the shock would stop when the dog ran back to you.

It’s important to remember that using punishment in dog training can have behavioral fallout.  If you are training positively and you withhold too many rewards too early, your dog can experience a high level of frustration which is not conducive to learning.  If you are using positive punishment, your dog can associate the wrong thing with the aversive you applied.  Classical Conditioning is at work along with operant conditioning.