Punishment & Reinforcement

I was reading Upsi's latest post on Discipline and it reminded me of my learning theory course in undergraduate school. Since I've kept all my college notebooks, I pulled out the one for that course and looked at my old operant conditioning grid. I've made a graphic of it for this post. I think it will help explain one of the main theories of behavior modification.

So these are the four contingencies of operant conditioning. In order for them to work however, this requires the person or animal to be able to see the contingency (relationship) between their behavior and the reinforcement/punishment/etc. If they can't see this, they can't learn, and you can't change their behavior with these methods. However, these methods have their own problems and pitfalls.

Positive Reinforcement: If the reward isn't meaningful to the person, then it won't work. It also best delivered as soon as possible for young children who don't yet understand the idea of delayed gratification. One should also be careful not to accidentally reward behavior unknowingly. By yelling at a child who acts out a parent may think they are punishing the child, but if the child is seeking attention, any attention, even bad attention, can be reinforcing!

Extinction: If you take away a pleasant stimulus, you are not likely to extinguish the target behavior unless you are also teaching a desired, alternative behavior. This technique can also prompt aggression and typically there is a spike in the undesired behavior before extinction begins.

Punishment: Type I punishment, aka positive punishment, may prompt aggression or avoidance toward the punishing person. It also does not teach the behavior you DO want. Many parents punish but fail to reinforce desired behaviors or teach desired alternative behaviors. Furthermore, it does not extinguish the behavior, it simply suppresses it in the presence of the punishing person. Punishment should never be an outlet for anger and should never been done in anger, which is another mistake parents, like my NM, make. It's far too easy for an angry parent go beyond mere punishment and into physical abuse!

Type II punishment, aka response cost, is also often used in anger, without a contract set ahead of time between the parent and the child. Without a clear agreement being made between the two parties, there is no clear contingency. When the relationship between the behavior and punishment is not clear, learning is difficult or impossible. Type II punishment shares all of the other pitfalls that Type I has. My NM liked to use this one, too, especially as I got older. It also requires taking away something that is meaningful to the person. If losing the privilege/object doesn't hurt, then the punishment is ineffective.

Negative Reinforcement: Difficult to use outside laboratory as it requires presenting an aversive stimulus for constant periods until the desired behavior occurs. My old textbook gives an example of a man wearing a device that beeps a warning and that makes an irritating noise when he slouches. It shuts off when he straightens his posture and helps him learn to sit up straight all the time, lessening his back pain.

This is of course just a brief overview. It's important to note some people use different terms for the things I have described and these are merely the ones I was taught. I think others use the term "response cost" differently than my professor did. But in sum, it is far better to use techniques that don't require the use of aversive stimuli. Not only are they more effective, they don't inflict harm on anyone.