In this video I describe the how conditioning to be used to train more complex behaviors. This can be accomplished with chaining, which involves linking together previously conditioned behaviors, and shaping, which involves reinforcing successive approximations of a desired behavior. Next I provide examples of behaviors which cannot be conditioned due to biological constraints on learning described by Keller and Marian Breland.
Don’t forget to subscribe to the channel to see future videos! Have questions or topics you’d like to see covered in a future video? Let me know by commenting or sending me an email!
Check out my psychology guide: Master Introductory Psychology, a low-priced alternative to a traditional textbook: http://amzn.to/2eTqm5s
Chicken Obstacle Course
https://www.youtube.com/watch?v=-nI27Zwi9VU
B.F. Skinner Foundation – Pigeon Ping Pong
https://www.youtube.com/watch?v=vGazyH6fQQ4
Rat Basketball at Wofford College
https://www.youtube.com/watch?v=jAQSEO25fa4
Marian & Keller Breland – The Misbehavior of Organisms
http://psychclassics.yorku.ca/Breland/misbehavior.htm
Video Transcript:
Hi, I’m Michael Corayer and this is Psych Exam Review.
In the previous videos we’ve seen the principles of operant conditioning and this shows how we can use reinforcement in order to increase the frequency of certain behaviors. The examples we’ve looked at so far have involved simple behaviors, things like pressing a lever in order to get a food pellet or pecking at a disc in order to get a food pellet.
This reinforces these behaviors and so they increase in their frequency. You might be wondering, “how do we get more complex behaviors to emerge? Pressing a lever is fairly simple but what if we want to do something little more complicated?” One way that we can get more complicated behavior through reinforcement is call chaining. Chaining is really just linking together multiple simple behaviors.
The way this would be done is that each individual simple behavior would be conditioned and then we’d only provide enforcement when multiple behaviors have been performed in a sequence. They’re “chained” together, each individual behavior is like a link in a chain. For instance if we had a pigeon and we wanted to teach it a more complex behavior, we would start with each step. So first we might have the pigeon press on lever.
We would condition this behavior, so we initially reward the pigeon each time it presses the lever. Then we want to condition another simple behavior, let’s say pecking at a disc.
So we have a colored disk somewhere in the operant box there and when the pigeon pecks at it, it gets a reward. Then what we would do is we would link these behaviors together to make a chain. Now the pigeon would only get the reinforcement if it presses the lever and then pecks the disc and when it does both of those things together, then it gets the reinforcement.
Then if we wanted we could add a third behavior, or a fourth or a fifth. We can keep adding links in this chain and I’ll post a link in the video description box where you can see a chicken who’s been conditioned this way in order to complete an obstacle course.
So the obstacle course has multiple behaviors that need to be completed in sequence and the chicken would be conditioned to perform each of these individual behaviors and then it only gets the reinforcement after the first two behaviors have been completed, then three, then four behaviors. Eventually the chicken can complete the entire obstacle course and then get the reward. This is one way we can get more complex behavior to emerge using operant conditioning.
What about behaviors that are hard to teach in the first place? Pecking a disc or pressing a lever, even in sequence, are still simple behaviors. If we want to get more complex behavior, we’re going to need another technique, and this is called shaping. Shaping refers to the idea that we have to shape the desired behavior. You don’t have to do much to get a pigeon to press a lever, it might put its foot there and accidentally press the lever and that’s the behavior that you want so you can reinforce it. But for more complex behaviors, they’re not just going to occur on their own.
If I want to teach a pigeon to play ping pong, which Skinner did, I can’t just sit around and wait for the pigeons to start playing ping-pong. It’s never going to happen, so I’m never going to be able to reinforce it. Instead I have to shape their behavior. I need to encourage them to do things that get closer and closer to what I want.
So shaping refers to using successive approximations of the desired behavior. So that means we reward things that get closer and closer to what we actually want until we finally get the behavior that we desire. We reward successive approximations of the behavior.
How would this work with teaching pigeons to play ping pong? What we would do is we would start by getting the pigeon interested in the ping pong ball. So every time the pigeon gets close to the ping-pong ball, it gets a reward. This is going to increase the frequency of the pigeon getting close to the ping-pong ball. Then when it touches the ping-pong ball with its beak, then that will get a reward. Then touching isn’t quite enough, then it only gets the reward when it pushes the ping-pong ball with its beak.
Now it has to actually push the ping-pong ball away. If it does that, then it gets a reward. Once that’s been established, that’s not enough, we go to the next approximation, which is to push the ball a particular distance away. This will encourage the pigeon to hit the ball harder with the beak until eventually it only gets the reward if it hits the ball off of the edge of the table.
Then we take two pigeons who’ve been conditioned this way, we put them on either end of the table facing each other, and each one will keep trying to push the ball off the other side in order to get the reward. As a result we’ll have these two pigeons who are playing ping-pong with one another. We couldn’t get this behavior initially, but through this process of shaping, by gradually reinforcing approximations of the behavior, getting closer and closer to what we want , then we can get this complex behavior.
I’ll post another link in the video description here of pigeons playing ping-pong so you can see this. I’ll also post one of some rats playing basketball. So the shaping process would be used in the same way in that video. Each rat would be rewarded first for touching the ball then only for picking up, then only for carrying it, then only for bringing it near a hoop, then finally only for picking it up, carrying it over to the hoop and placing it inside.
So that would also rely on shaping to get this behavior. Now we can see that we can get very complex ehaviors to emerge from this but there’s certain limitations. There are some behaviors that we just can’t get to occur no matter what we do. This is a limitation of this operant conditioning. This was demonstrated by Keller and Marian Breland.
The Brelands were animal trainers and they were using the principles of operant conditioning to get some complex behaviors in the animals they trained but they found that there were some behaviors that they just couldn’t get to occur.
One problem that they had was they wanted a raccoon to pick up these little coins and bring them over to a piggy bank and place them inside. This would be a cute performance. The problem they had was the raccoons started treating the coins as if they were food once the coins, they had been reinforced for picking up these coins and this caused the raccoons to associate the coins with food and as a result they started treating the coins like food. They started rubbing them with their paws and “cleaning” them the way they do with food. The Brelands couldn’t get his behavior to stop. They couldn’t seem to condition it out of the animal.
They had a similar problem when they were trying to condition pigs to push some coins along the ground only a few feet. Seems like a fairly easy behavior to condition. We can get pigeons to play ping-pong we should be able to get a pig to push a disc a certain distance. The probably they had is once the pigs started associating these discs with food, then they started treating them like food, they started “rooting” them, digging them into the ground with their snout which is something they would do when searching for food.
This behavior that arose couldn’t be conditioned away. What the Brelands realized was the problem they were having is that the behaviors they were trying to condition were too similar to the animals’ natural instincts.
So they referred to this misbehavior as instinctive drift. Or you’ll also see this called instinctual drift. The idea here is that if a behavior that we’re trying to condition is too similar to an instinctual behavior, then the instinct will win out, the conditioning won’t be able to occur. The instinct will disrupt the conditioning process.
The animals have certain predispositions for certain behaviors and those predispositions are very powerful and conditioning is not strong enough to overcome them.
I’ll post a link in the video description box where you can read the Brelands’ paper on this called “The Misbehavior of Organisms” where they describe this example with the raccoon and the pig and a few other animals that they trained where they ran up against this instinct and they couldn’t get certain behaviors to be conditioned.
This shows us that there are biological limits on the conditioning process. It’s not the case that everything is the result of learning and experience.
Just like we saw with the learned taste aversion, where we had a biological predisposition for certain associations, here we have some biological predispositions for certain behaviors and if we try to pit conditioning against those behaviors conditioning isn’t going to be able to win, the instinct is going to take over.
I hope you found this helpful, if so please like the video and subscribe to the channel for more.
Thanks for watching!
