Understanding behavioral momentum and its applications to establish effective behavioral change

"Behavioral momentum" is understood as two parallel concepts in science and animal training. One concept deals with reinforced behavior, as related to the rate of reinforcement, developing resistance to change in the presence of interference (T. Nevin). The second concept deals with achieving low-probability behavior through meticulous pairing with high-probability behavior (F. Mace). We will explore these concepts as they are studied in the laboratory and applied in the field. Understanding these concepts will provide us with better skills to request and shape for increased compliance and challenging behaviors in birds and other species.

Phung T. Luu Behavior and Training Solutions, LLC Animal Behavior & Conservation Connections, LLC

Hello. Sorry for the technical difficulties. We have a group of brains to solve this problem. Hope you had a good break. We will try to go through this quickly. What I would like to talk about is behavioral momentum and it's probably a concept that we've used at one point or another. I have used it in training but never realized I was using it until I was talking with my professor. I got interested in learning more about it from a scientific view.

Before I go into presenting the concept, let me give a background on what I do so that you have a baseline on why I make the decisions I do. Everyone trains differently in this room. There are probably 10 or 20 ways to train something depending on who you talk to. I manage 2 programs. One is an educational program that deals with preservations. We provide educational programs and highlight some conservation efforts in our programs. Free flight is the key to how it influences the work I do and the approaches I take to make decisions.

The other is through consultation and animal training programs at again zoological facilities or animal facilities around the country. Just as some short clips of what we've done at different parks around the world. This slide, it's kind of cool, I didn't just put it up there for the coolness of it, to highlight the range of animals or birds in this case that we work with. The most important components are safety that you as a trainer aren't hurt by the animals that you're working with, but also because we work with free flight birds that they agree to fly away. You look at them wrong and they want to fly away. The eagles can be quite dangerous animals as well.

So hopefully with some of that background you will be able to gain a little bit of insight into why I make some of the decisions that I do through this talk this morning.

So the purpose of sharing behavior momentum concept, this is just a review for some of you but an introduction for others. For me it was a program that I used already but I didn't put that much thought into it, until it was pointed out that I was using it. In making decisions, in what we do, I often to refer to this formula. Share/review additional tool. PATH = Philosophy measured against resource. When we are teaching, that's really the key components about how we make decisions, hopefully that will tie back to the explanation of why I'm sharing the concept of behavioral momentum with you today.

And that really ties into efficiency in what we do, especially being entrepreneurial when it's my own budget or my own money. You have to be very efficient because you're not using someone else's dollar. Time is one of the major resources that we have to account for. So efficiency to produce results in the shortest amount of time and the least amount of resources as possible.

So I am going to go through the next few slides about the researchers about behavioral momentum and then we will bounce back to some of the fun stuff about applying it. There are two lead researchers. While they both do work in behavioral momentum, Mace diverges from the true definition.

According to T. Nevin, "Persistence of behavior increased in the face of challenges or disruptors such as distraction or discontinuation of reinforcement." We will probably revisit that with more examples shortly. That theory is foundation ... Law of effect by Thorndike (1911). You will probably hear more of that today. That is that, a behavior that is followed by pleasant consequence is likely to be repeated. Behavior that is not followed by pleasant consequences is not likely to be repeated. The durability of the behavior depends on the consequences of the behavior in the past. Or more simply put, behavior is the function of its consequence.

It also has its roots in Newton's second law of motion that states, when an external is applied to an object in motion, the change in velocity is related directly to the magnitude of force, and inversely to the object's inertail mass. I don't expect you to understand it that well, but we will simplify it as we go along.

My root is as a trainer. A lot of time we go back and try to tie in or find some research foundation to back what we do in the field. The other component of what I do is that I teach. I can't just say it's because what I say so, so I have to therefore have to provide background information as a support.

In behavioral terms, "When a disruptor is applied to ongoing behavior, the decrease in response rate is related directly to the magnitude of the disruptor, and is related inversely to the equivalent of behavioral mass." When a behavior has accumulated large enough mass, the behavior has then momentum or forward momentum in the presence of obstacles. disrupter?

There have been a few experiments. The researchers conveniently took data from college basketball games. That's a lovely strategy for research. So they observed about 30 something basketball games, right about this time of year, the March Madness, gotta go do some research. And, "Behavioral momentum in college basketball." (Mace, Lalli, Shea & Navin). The decisions of coaches reinforced some of the concepts of behavioral momentum, the decisions that we make in order to try to win games.

They found two main results. One is that if scoring momentum is interrupted, then the likelihood of scoring is decreased. If you watch any sport games, when the opposing team goes on a run, which the other coach needs to call time out, or fake an injury or something to stop the momentum. They found that supports the theory of behavioral momentum. And secondly, if faced with low scoring run, the team recovery is likely if they had positive scoring within 3 minutes previously. If they were doing really well, and then they went with low scoring, they are likely to recover quickly and go back to a high score run because of that momentum.

As a quick summary of that idea, there's an effective strategy to build and maintain behaviors in new and adverse conditions. Some of you have had a chance to talk with Bob Bailey about his work in the field. A lot of what they do is based on this behavioral momentum concept, of really building up behaviors, and this behavior maintains this mass, so that once introduced into new environments the behavior is maintained. Whether you're training dogs, which are exposed to new environments where you want a target behavior to be maintained, this is where it applies. You have to train the animal so that they have such good reinforcement facility, so that when we take them to another facility, they have the momentum behind those behaviors.

There are also more complex reinforcement schedules.. there are plenty of PhD people in this room to search out for those answers.

We are going to show a real quick video here. Thank you to Steve White for sharing this video with me. This is from a trainer called Randy Hare. I saw this last year and I thought it was really applicable in terms of behavioral momentum and maintaining a target behavior in the presence of distraction. I guess it does have sound, but we don't really need it. To explain what it is, it's scent detection, in these boxes, and the reinforcer is the .. but during, when the dog is working, there are so many distracters, they are dumping tennis balls on top of the dog, and sometimes chunking them at him, but the animal has been so well reinforced for finding that scent that those distractions are irrelevant. There's another one too. These tennis balls are not causing him to lose his mission. alphak9.com video. The dog can be picked up no problem, he is still very focused on his target behavior. That's a really nice video about behavioral momentum.

According to F. Mace, he holds a similar definition. "The tendency for behavior to persist following a change in environmental condition." The difference with Mace is that.. more on high probability, to create momentum. And this is actually where more of my focus is. Using high probability behaviors or commands to elicit low probability behavior.

There are similar studies out there that show this momentum-like behavior to be advantageous. (Catado, Kolko, Neef and Egel, 1986). Their work is mostly about vocal prompts, physical guidance, to help the learner engage in the desired behavior. Error reduction. However, there is limitation to those techniques where there may be full contact, like in physical guidance, a lot of full contact, and in some cases that is not desired to try to move an elephant physically, or try to move a tiger physically for example. So the concept of behavioral momentum would be much more useful in situations like that where safety is much more of a concern where you don't necessarily want full physical contact with the animal. They did several experiments. Behavior is behavior that exists across a range of organisms whether people, animals, rats, cats, chickens, they did experiments with some individuals with mental retardation that I will share with you here now.

One of their subjects was Bart, a 36 year old man with a severe mental retardation with an IQ of 42. Bart was a large man. 200 pounds, 6 foot 1 inch. Not a guy that you can do physical guidance or manipulation with. He lived at a behavior and analytic university group housing for about 6 months, and when he first started there, they were successful with obtaining compliance and behavior, with just general positive reinforcement training. However, at 2 months they saw challenges more aggression from him, non-compliance, so the procedure, they took some baseline of Bart's behavior, and they measured random.. they identified 20 low-probability behavior and they measure that on a fixed time schedule. High probability command sequence were delivered in series of 3-4 rquests at 10-s intervals preceding presentation of a low-probability command. So go wash your hands, etc. High probability commands were give me 5, give me a hug, and he would do those regularly.

They also mixed up when they asked those, so there was no pattern attached to those. There were different classes. They were offering two different sessions at .. two 15-min sessions per day. The results that they saw, the baseline compliance of just simply asking for low-probability behavior, when they say Bert, Albert do this, clean your plate, it was about 68% of the time. Or Albert, don't put your feet on the table, that was 53.57% of the time. There was some psychotropic intervention during these trials, so Bert was put on Howadol... the result for high probability command sequence through .. Bert, don't put your feet on the table, they saw an increase of 87.5% .... Results: high probability command sequence prior to low-probability with a do command, which was "albert please take your.." and then compliance went up to 90.5% of the time, and for "don't" it was 44% without the pairing of the high probability. They did several of the tests and they went back and forth to go to the reverse to see if they can do consistent results. So they went back to that, so we're seeing some pretty consistent results. So I'm going to skip these.

Here's one where they pair a high probability and a low probability for do and don't commands. There was an increase to 93% for "do" and 90% for "don't". That's 30-40% increase in compliance from no procedure at all. So we can see that this is a valuable procedure.

But the strategy of using this procedure requires careful arrangement to setup the environment correctly so that the desired behavior can be triggered. And it's also generalized to a continuous reinforcement schedule. The high probability command has to have a pairing with that so that the behavior is maintained. The positive reinforcement was appraise like "Good job Bert" and "Thank you" and "You did great". So that's how it's maintained with the high probability command.

The relevancy to training is that it provides an additional tool to modify training behaviors. In this context, it's to be able to maintain behaviors in adverse conditions. However, there are consequences to these effects.

Consequential effecto f behavioral momentum. The advantages of behavioral momentum: - reduced frustration - reduced aggression - increased compliance - Safer alternative when working with high risk animals - high risk defined as physical injury to trainer or escape behaviors. - increased efficiency

I am going to tyr to mess with this program.

http://youtube.com/watch?v=pY5rl-DiEjU

I don't know who this is. I apologize if you're here. It's off youtube, so I think it's fair game. So what you're seeing is a black voucher on a trainer's hand during a presentation. So we will see some undesired behaviors in a moment. What he's supposed to be doing is applying.. to another trainer. There's another trainer that is off-screen right now. With vulchers it's not a target behavior, it's an undesired behavior.

"Black vulcher flying & fun antics last chance forever bird" (youtube video pY5rl-DiEjU)

Ultimately I just wanted to show that without that momentum of what he's supposed to be doing, he engaged in an undesired behavior. They need to make sure they go back up and build up enough reinforcement history and build up higher value behaviors to do something else.

Establishing.. consistent reinforcement schedule. Continuous reinforcement schedule. There are two behaviors that we are working on. Establishing high-probability behavior with a Condor. There's three behaviors; go away, which you just saw, and there's going to be a come back to your station, and then there's a stay on the station as well. Working with a condor, that's big, and potentially risky and dangerous. We need to make sure the animal has a really good repetoire of what we're doing and we need a high rate of compliance on high probability behavior. So we're building this behavior and lots of reinforcement. So if I'm working with this animal further, if I work on additional behavior, this set of behavior would be really useful. Suppose I train for scale next, and scale is a brand new behavior, I can go right next to this behavior like "go to station" because there's lots of reinforcement on this, without creating frustration, and frustration could lead to aggression and I don't want to be in an enclosure with a frustrated condor.

Here's another video that shows the side effect and outcome of high probability behavior and how it maintains the behavior and how it can influence other behaviors in that package as well. Outcome of high-probability behavior to maintain and request additional behaviors (WTH). This is a white zawl hawk, found in Texas with an eye injury, so he was not releasable through rehab. So he came to our program something education. I don't know if you know .. it might look quite simple and easy, the procedure that we're using, but there's a lot of reinforcement history that went into training that behavior. There's no reinforcement in between. That's the point I'm making. The momentum of the step up behavior was quite high. The momentum is quite high. So the compliance rate is maintained. So beyond the step up behavior, there's some additional behavior we're asking of this animal; step on the gloves, stay on the glove when i move around, and be able to put equipment on gently, that's another one as well. So this allows us to request more behaviors from this animal without seeing undesired behaviors.

I am going to show you a series of videos with.. and this is, this video was taken for demonstration purposes, so I don't just sit around and do this. So, to set you up, there are 3 behaviors. The rouse behavior.. they shake their feathers. The other is a foot wave, and the one that we're training that has low probability is a turn around behavior. So we're going to show you all 3 of those and in a package as well.

With those two behaviors, the rouse has been in the bird's repertoire for some time. We just play around with them, give them cues so that they have something to do. The wave was taught 2 days prior to filming. So the behavior was recent.. and then the turnaround, ... macaw. Well, no turning back now. But I can show you the final product. The low probability turn around was instilled in training, so we prompt the bird to spin around away from us, and it does the complete around, we continue to prompt it, using the behavioral momentum we are making some shortcuts. Instead of following through these.. in this demonstration, we injected these other behaviors in, we're taking some shortcuts. So this is the.. end product. High probability to low probability. There is still some prompting in that, off to the right of the screen. So there, there was the prompt. He hesitated, a little bit of help. There's a lot of distraction happening, by the way. This is still quite fresh, this is probably the second session or so. There was a request there, she didn't do it, so now we're inserting the two behaviors that she knows well. Then asking for the turn. There is a lot of offering of behaviors. Because the last trial she hesitated, so we went ahead and paired the hgih probability behaviors in to keep the momentum of learning so that it doesn't just stop.

So it's a one-case scenario to show you, but I can guarantee you that we can demonstrate this over and over again with other animals with this kind of scenario. Here's another fun one with a donkey. Now, what this.. this video was shot with the purpose of demonstrating this concept. There's a lot of distractions, and bird activity. There's some people.. so the dog has lots of distractions. And so this dog knows high-5, that's a high probability behavior, low 5 is a high probability, and the new behavior is jump on to this stool. So let's watch it really quick. We're demonstrating the high probability concept, but the momentum effect of behaviors existing in the context of lots of distractions. He's never been up on the stool before. Does he know it? No, he just simply taps it, it's not a behavior that he knows. There's a fine line.. how much do you ask, to get the next approximation and lose the others because you have asked too much? Soon enough he is going to get inventive and we capture his desire to be inventive. As you can see it's a very low probability for him to get on there. A few reps, and he didn't quite get it. So we are going to insert the high probability behavior there, and then bring it right back. We can probably chain that session to demonstrate another topic, which is time is a .. reinforcer.. we are showing the momentum of previously established behavior in order to give effect to low probability behaviors.

There are some challenges to remember when you're using this kind of pairing because it can easily turn into a superstitious behavior, where behaviors can become moot.. when he's hesitating to get on to the stool, and I ask him to get off, and offer him a high probability behavior, then that becomes learned. I was careful in this behavior to change up the two, it's not always high-probability low-probability but sometimes the opposite order, so that the behavior of jumping up on to the stool is not chained to the other high-probability behavior. The concept of strengthening behavior or reinforcement bonding is useful to maintain or improve behaviors in adverse conditions, or it can be used to strengthen weak behaviors, by pairing it with previously highly reinforced behavior where it's useful to train new or more difficult behaviors quickly to avoid frustration.

reinforcement log book?