Let's Talk Positive Reinforcement Part II

And tonight I get the chance to write my "training" post! See Part I here.
WARNING: I am very prone to stupid errors, so if you note some (Jen, I'm looking at you) or have a question involving continuity etc, please please point it out so I can address and/or change it.

In this installment, I am going to give a brief overview of a few points on positive reinforcement and how it relates to operant and classical conditioning. Secondly, I will discuss why some people choose to use positive reinforcement more heavily than other types of training and conditioning.

Everyone remembers the old story of the dog and the bell and the salivation right? Or you may remember it more commonly as "Pavlov's dog." Classical conditioning is used a little in positive reinforcement training when we discuss the "bridge." Another thing you may all have heard to some degree is "clicker training" right? It's most commonly heard in association with dogs. The whistle that a marine mammal trainer uses when his or her animal completes a behavior is just another form of that same tool. The word "good" can also be used with children or animals, but in animals especially, it can not be as clear. This is called a bridge in that it bridges the behavior desired to the reward promised. It's a little different from Pavlov's bell for a few reasons, but we will get back to the bridge later. (remind me if I forget. ;-)

Positive reinforcement falls under the umbrella of Operant Conditioning. In fact, the different methods of reward and punishment I will be discussing are all part of Skinner's work here. And basically that's what it concerns - increasing or decreasing a desirable or undesirable behavior through different means of reward and punishment. Most positive reinforcement trainers will use a little of all these methods, but will rely mostly on positive reinforcement.

NOTE: A reinforcement technique is used to INCREASE the probability of a desired response occuring again. A punishment technique is used to DECREASE the probability of an undesired response occurring again.

The 4 techniques used are (and some of this will be review)
1. positive reinforcement (adding something to the environment)
2. negative reinforcement (removing something from the environment)
3. positive punishment (again adding a punisher)
4. negative punishment (removing something as a punisher - ex. taking away tv privileges)

All of these techniques can be used in some measure to gain desired behavior. However, it takes skill and knowledge to use any of them well. Many trainers prefer to rely heavily on positive reinforcement, and this is one reason. Positive reinforcement training itself comes with a whole tool box of techniques and ideas that aid in its implementation. It takes a while to perfect. Using the others takes just that much more. Negative reinforcement involves removing something undesired after the completion of a behavior in order to reward that. With a skittish animal, that could simply be your presence. But it is complicated to even explain let alone execute. So that is why many steer more in the positive reinforcement direction.

As for the punishments, there are a few reasons why reinforcement (in my opinion and according to experts who use it) is a more desirable approach if it's possible. (and it usually is).

A subject will only work as hard as it needs to to avoid a particular punishment. For reinforcement (if it is truly a good reinforcer for that individual), a subject will work above and beyond what is asked with often much more spectacular results.
Ex) You may have been punished for staying out past curfew. Let's say your curfew was 9pm. If you came in at 9:05, you were yelled at and couldn't go out for a week. No reinforcement was given. How often did you come in early (say 8:30? 8?) - I'd hazard not often. It was probably usually at best 8:50 unless you were very nervous. Now let's say you were given the keys to Mom and Dad's car for a a day for each minute you were early, and you discovered this every time it happened (whether or not punishment was added for lateness). You would probably have rolled in a bit earlier when you needed those keys right?
If punishment loses its effectiveness, the ante must be upped. And in many animals' cases at least, punishment must be severe enough to cause fear of death to have any sort of lasting effect. Who wants to go there?
Much as in the case with #1, the punisher must often be present for the subject to exhibit the proper behavior. ex) Your dogs are smacked and/or yelled at for getting on the couch. When you're not home, guess where many of them go? You guessed it, right on the couch. You may be able to think of a punisher strong enough that they are so afraid to get on the couch that they won't even when you aren't present. But isn't it much more pleasant to think of a reward wonderful enough?
Punishment has unintended results. Punishment often works, at least in the short term. But how many people have heard of circus animals turning on trainers or beaten dogs becoming unpredictable and thus unadoptable? How many people have seen a child get smacked in public only to see that child turn around and sass the parent, or to see them sulk and stalk away? Yes, these things happen and the moment eases. However, the effects that you can't see continue to color the relationship if it happens often enough. It's also variable depending on the individual, but again, isn't variable joy preferable to variable anger and resentment? Fear does not equal respect. Think back to some of your favorite teachers or other adults in your lives. Were they the ones who ruled the classroom by complete fear and little positive reinforcement? Or were they the ones who built you up? How did you feel about them? Like you would do anything for them? Positive reinforcement used to a knowledgeable person has pretty predictable results. And the "problems" you can have from them rarely include aggression.
Punishment is very tricky to use well. It requires a much more steadied hand, knowledge of behavior and training and calm demeanor than positive reinforcement. And how often is that the case?
When you use punishment, you "model" that behavior for your subject. You are using some sort of aggression, often on someone or something smaller and more helpless than you are even if you preach otherwise. Therefore, it can be viewed as an ok way for them to act as well. Excuse me while I go into a little story to illustrate my point:

A friend of mine was talking about how she recently has started spanking her toddler for "dangerous" behaviors. The way she discussed it, it didn't seem there was much structure or consistency to the application of her punishment (bad in any training scenario.) She was telling us about a recent event in which she spanked her for getting up on a chair. She said that after that, "X got down and scowled at me! She actually scowled! Then she stalked toward me and raised her hand as if to hit me! I told her, NO MA'AM! We don't HIT Mommy! I let her know by my sternest face and voice that in no way was that ever ok. That little stinker has even tried it a couple of times since! Can you believe it?" And yes, actually I really could. And no I'm not saying that a toddler, animal etc will never exhibit aggression with a parent or trainer if they don't see it modeled. I'm just saying that "do as I say, not as I do" is not the most effective platform from which to preach non-violence, and that an aggressive response is more common following an aggressive reaction.

QUESTIONS? I'm sure I've missed something. And if I have, feel free to add to the discussion! I love talking about this stuff, and will feature your questions in various posts as they're asked. Our next installment will involve the actual application of various techniques.

Share On :

Pin It