A blog about problems in the field of psychology and attempts to fix them.

Friday, February 28, 2014

How to Explain Behaviorism: Operant and Classical Conditioning

There are many explanations that can be used to help people understand the Behaviorist Point of View. Some are very factual, others argue towards practical concerns, and still others are highly philosophical. This is the first in a series of posts trying to show these styles of explanation in compact and easy-to-digest form. Feedback is welcome. Because of a guest lecture I must give soon, the first post will focus on outlining operant and classical conditioning. The order is not meant to imply that this should be the first thing you tell someone about behaviorism, nor to imply that it is the most convincing line of explanation.
 -----------
How to Explain Behaviorism, version 1: 
Operant and Classical Conditioning

Operant and classical conditioning are two different ways in which organisms come to reflect the order of the environment around them. They are not perfect processes and they certainly cannot explain facet of human and non-human behavior. That said, they are surprisingly reliable processes, and they can explain much, much, more about human and non-human behavior than anyone would have thought before extensive study of those processes began.


It is probably best to think about operant and classical conditioning as offering two different types of developmental stories. They are not stories about what a behavior is, now, but rather stories about how that behavior got to be that way. Classical conditioning stories are about things happening around the animal, no matter what the animal does. Operant conditioning stories involve consequences of the animal's action, i.e., what happens when the animal operates upon the world as an active agent. There is some debate about whether we need two types of stories. There are good reasons to go either way, including some recent genetic evidence that they can be disentangled. None of that really matters here; all that matters is that you understand the two types of stories and their consequences for future behavior.

Note below that "stimulus" can refer to any object, event, or situation that an organism could potentially respond to. Note also that "response" can be anything the organism does. For now a "response" could be an overt action (such as jumping up and down), a covert action (such as tensing your leg without moving it), or even a thinking or feeling, so long as we conceiving of those as active, rather than passive.

Operant Conditioning

Operant conditioning stories involve an animal doing something that changes the world in a way that produces, crudely speaking, a good or a bad outcome. When an organism does something that is followed by a good outcome, that behavior will become more likely in the future. When an organism does something that is followed by a bad outcome, that behavior will become less likely in the future. The action and outcome could coincide because of natural laws or social conventions, because someone purposely set it up that way, or it could be that the events followed due to random chance in this animals life history. For example, in pretty-much any animal's it is good to stop touching overly-hot objects (natural law), in some worlds telling a parent you love them results in good outcomes (social convention), and in some worlds tapping a baseball bat five times on the left corner of the mound is followed by a home run (random chance).

Operant conditioning stories require that the outcome be reinforcing or punishing to the particular animal in question. (There are ways to specify that so it does not involve circular reasoning, but we don't need to go that deep.) For example, candy might reinforce one person, but not another; some might find a graphic kill-sequence in a violent video game punishing, while others find it reinforcing; etc.

Over time, the story goes, if a certain type of outcome consistently follows a particular behavior, this will affect the rate of future behaviors.

Example Traditional Story: A cat is put in a "puzzle box". It performs a wide range of behaviors, because cats don't like to be in cages. Eventually one of its flailing limbs pulls a lever that opens the cage door. This happens many times, and each time the lever gets pulled a little bit quicker (there is no "ah ha!" moment).

Tradition vs. Necessity: Traditionally operant conditioning stories start with a relatively "random" behavior, but they could start with any behavior. Traditionally the story then introduces an arbitrary consequence, but in real-life situations we usually care about socially-mediated consequences. Traditionally many cycles for the consequence to make big changes in the frequency of future behavior, but sometimes the changes can be quite quick and others it can take a very long time. In the traditional story the consequence always follows the behavior, but there are many cool affects that we know about when it does not the consequence is intermittent (i.e., the "schedule of reinforcement"). Traditionally the consequence has to be immediately following the behavior, though there are some exceptions, you probably want to stick with the traditional version here.

Enhanced Traditional Story: Often operant conditioning stories are enhanced by adding a "discriminative stimuli", which indicates that a particular contingency (a particular connection between action and outcome) is in effect. For example, an experimenter working with rats might have a light that, when on, means that lever pressing will result in food. Similarly, a special education instructor might have a picture of a hat that, when held up, means that saying "hat" will result in an M&M.

Other Classical Operant Stuff: You can do amazing things with discriminative stimuli. You can train people to respond to very specific stimuli, or to very general "categories" of stimuli. For example, we can get pigeons to discriminate Monet's paintings from Picasso's. Also, by drawing out the "schedule" of reinforcement, you can also train animals to respond for many, many times without getting reinforced. For example, we can get people to pull slot machine levers scores of times without a win.

After Conditioning: After the events of an Operant Conditioning story, a behavior either has an increased or decreased rate of occurrence. Often there is a big increase or decrease specifically when a particular stimulus is present. So, if you know the world that a person has lived in before, you know something about why they respond now in certain ways in the presence of certain objects, events, or situations.

Classical Conditioning

Classical conditioning stories involve (at least) two things that coincide "out there" in an animal's world. Those things could coincide because they are causally related due to natural laws or social conventions, or it could be that the events occur at random in relation to each other and this animal just happens to be the animal that experiences them together. For example, in pretty-much any animal's world lightning is followed by thunder (natural law), in some worlds hearing "say cheese" might be followed by a camera flash (social convention), and in some worlds eating lamb dinners could coincide with hearing bad news from loved ones (random chance).

Classical conditioning stories also require that the organism already have a developed response to one of the two events. For example, thunder could make you flinch, a bright flash could make you wince, and bad news from loved ones could make you cry.

Over time, the story goes, if two things are repeatedly paired together out there in the world, the organism will come to respond to one as they already respond to the other.

Example Traditional Story: When Mary was a child her father liked to take many pictures of her. He he always said "Say Cheese" before he took the picture, and he always used a flash. Every time the flash hit Mary, she winced slightly. Now, whenever she hears "Say cheese" she winces.

Tradition vs. Necessity: Traditionally classical conditioning stories start with a response that seems unlearned (an Unconditioned Response to an Unconditioned Stimulus), but they could start with any response the animal already has. Traditionally the story then introduces something the animal has no existing response to (a Neutral Stimulus), but it usually still works for stimuli that already elicit some response. Traditionally the neutral stimulus comes to evoke the response associated with unconditioned stimulus after several pairings (thus becoming a Conditioned Stimulus), but sometimes only a single pairings is required, and sometimes neutral stimuli fail to convert to conditioned stimuli even after many, many pairings. Traditionally the stimuli have to be very close together in time, but sometimes you can create conditioned stimuli when the pairings are far apart.

In many cases, where the traditional story does not hold, there has been a lot of research into the exceptions, and we have very good understandings of why such exceptions should exist. For example, after a single event many animals will learn to avoid novel tastes that were associated with becoming sick quite a bit later. This makes a lot of evolutionary sense; poison food present a big risk, and one dose not normally experience the full effects until quite a bit after ingestion. On the other hand, when dealing with fairly arbitrary pairings of stimuli, as we get all the time in our modern world, the structure of the traditional story holds. For example, why should anyone ever have become excited by hearing a computerized voice say "You've got mail!"? Because of several pairings, that's why.

Other Classical Conditioning Stuff: You can do amazing things here with generalization and discrimination training, and there are many other interesting phenomenon that scientists have discovered.

After Conditioning: After the events of a Classical Conditioning story, the presence of a conditioned stimulus elicits a conditioned response. So, if you know the world that a person has lived in before, you know something about why they respond to certain things in certain ways now.

A Bit of Light Theory

Philosophical behaviorism can be very deep. In this context, all I will say is that most behaviorists believe we can explain a great deal about human behavior using the types of stories above. That is, the preferred style of answer to a run of the mill "Why did he do that?!?" question will begin with "Well, in the past history of that person, doing that behavior resulted in...."

Because these explanations are all about the way the world around the person works, and the person's past history in that world, you don't need to include traditional "mental" explanations. That doesn't mean that traditional "mental" stuff doesn't exist, but it does suggest that we can explain an awful lot about human behavior before we would need to start talking about them.



5 comments:

  1. With no time yet to read the article but I saw that you stated that is the animal or the person the reinforced one's. However what is reinforced is their behavior.

    I'm looking forward to have the time to read the entire article. Surely as always it will be very interesting. ;-)

    ReplyDelete
  2. Good catch. As you will notice the explanation tries to be both accurate, and not technical. It is a difficult balance.

    ReplyDelete
  3. Ah, I can't believe it's taken me long to see this post!

    Minor point: in the "operant conditioning" section you have a bolded section titled "Other Classical Conditioning Stuff" and I assume you mean "operant" not "classical".

    More importantly: I thought this was a great introduction to operant and classical conditioning, and a really good way of trying to introduce behaviorism in a non-controversial or confrontational way.

    My only real quibble is that I would disagree that it's an accurate explanation of behaviorism for the simple reason that the philosophy of behaviorism does not depend on the validity of conditioning. So if (somehow) we learn tomorrow that classical and operant conditioning are false, the philosophy of behaviorism would still be valid.

    It might be more accurate to view it as an explanation of behavioral psychology or behavior analysis. But that's just me being pedantic.

    ReplyDelete
  4. Mike, I wrote a long reply, but the computer ate it. Here is the summary:

    You are right about conditioning methods and the theory of behaviorism being logically separate, and that is why I didn't want to do this one first. On the other hand, understanding the science of operant and classical conditioning did help convince many that behaviorism was plausible.

    ReplyDelete
    Replies
    1. That's a shame about your post being eaten, I did the same thing the other day where I comprehensively corrected someone's misunderstanding of an aspect of philosophy of mind in an essay-length response on my phone. I then attempt to click on part of it to include an apostrophe that I'd missed, the screen jumped up and the part that I clicked on was now the "Cancel" button..

      But otherwise I agree with your response. It is certainly a useful way of getting people to be less skeptical of behaviorism and see it as a serious philosophy with real contributions to the world. It's just that I've had the opposite argument with people who reject findings within behavioral psychology (or the field itself) and they then treat that as a rejection of behaviorism, so I can see the advantages and disadvantages.

      Delete