Friday, October 5, 2012

Shedd Animal Training Seminar: What to Do When Your Animal is Wrong

Ken is unabashedly in favor of positive reinforcement when training. I am too, and in fact, I think most people see the value in using positive reinforcement for training. But people often struggle with what they should do when the animal they are training is wrong, or does something they don’t like. Some people will argue that the only way to deal with incorrect responses is through the use of punishment- after all, it decreases behavior, right?

Ken Ramirez
 Well... yeah. Punishment, aversives, negative reinforcement… it all works. Ken acknowledges this freely. However, he adds that using these approaches can be fraught with pitfalls and often have unintended consequences, especially when done by novice trainers.

To a certain degree, Ken agrees with the advice “reward what you like and ignore what you don’t.” He also acknowledges that there are behaviors that simply cannot be ignored- like aggression. Thankfully, there are many ways to deal with problem behavior. That said, if you’ve come across a behavior that can’t be ignored and you can’t figure out several possible solutions, the situation probably requires skills that are more advanced than you have. And that’s fine! Animal training is something that even the experts continue to learn about, and consulting with others is standard practice.

Ken’s go-to method for dealing with incorrect responses is something called the Least Reinforcing Stimulus or Scenario (LRS). An LRS is designed to deal with unwanted behavior without causing frustration for the animal. It is deceptively simple: to implement an LRS, you provide a completely neutral response for a short period of time, followed by an immediate opportunity for the animal to earn reinforcement. Of course, an LRS is a bit more complicated than that, so let’s take a look at it in more depth.

First, like everything in training, timing is important. The LRS must be implemented immediately so that the animal understands which behavior was incorrect. A poorly timed LRS will likely frustrate the animal, and what you’re looking for is a calm acceptance from the animal.

Next, the LRS should be brief. In general, when an LRS is used, it will probably last from 3 to 5 seconds. Of course, this length may vary. It needs to be long enough that the animal notices an interruption in reinforcement, but not so long that you upset the animal. For trainers and animals that have a relatively fast rhythm going in training, the LRS can be quite short. For slower trainers and animals, the LRS will need to be longer. We saw this playing out in the daily sessions we watched; the sea lions were quite quick, and their LRS might only be a second or two. On the other hand, the lizard we saw was much slower, and his LRS needed to be longer as a result. That said, do not be tempted to extend the length of the LRS beyond what is strictly needed. The LRS should be the same length despite how “bad” the mistake was. It is not a time-out procedure.

One of the most important features of the LRS is that it should be a completely neutral response. This means that you should refrain from having an emotional response (no scowling or grumbling, for example). However, you do not need to freeze. An LRS works because it is an interruption in the flow of reinforcement in which you simply don’t respond to the animal. If you’re looking at the animal, continue looking at it. If you’re looking away, continue to look away. The only exception to this is if what you’re doing is reinforcing.

Finally, you need to provide the animal with an immediate opportunity to earn reinforcement after the LRS. Because you don’t want to cause frustration, the best way to do this is by offering the animal a different, easy behavior. Ken tends to use targeting because that’s such a strong behavior for the animals and trainers at Shedd alike, but any behavior that the animal knows well will work. Ken will then do a few other behaviors before asking for the behavior that the animal failed at earlier.

Astute readers will have realized that the LRS is really meant for an animal with prior training (because there needs to be at least one fallback behavior), and with whom you have a relationship (so that you know how long the LRS should be). It will also only work when the animal is participating in the process. You can’t use an LRS with an animal that’s disengaged from the training process because it won’t notice that you’ve interrupted the reinforcement process.

The LRS, while simple on the surface, really does work. It’s been proven both practically and scientifically. It is, technically, the first step to extinction, so it can take a bit of time. Many positive reinforcement trainers do the LRS quite naturally; I know I’ve done them, but not because I knew that’s what I was doing. Having some knowledge about them definitely helps me understand how to implement them better.

What about you? Do you use an LRS, or something similar? This is an admittedly new concept to me, so I’d love to hear about others’ experience with it.


Anonymous said...

I really, really love yoour blog. Tomorrow is my Rudy's second "growl class" for reactive dogs with a local positive reinforcement behaviourm and I googled "reactive dog blog" on a whim ... and have been reading your story off and on all day! So nice to know a) there are others who get how I feel and b) that there IS light at the end of the tunnel! So thank you for that!

I do practise LRS, though I didn't know that is what it is called! It's just something I started to do naturally that seemed to be the only thing that got Rudy out of his reaction quickly. So he explodes at a biker, I keep staring at him, and he suddenly turns on a dime (like he knows!) and sits and stars at me. And then I ask him to do something (usually lie down and then speak) and then I click, treat, and move on.

Clearly I'm no savant because I need professional help, but it's neat to know this is an actual "thing"!


Anonymous said...

Holy, holy typos. Embarrassed I clicked submit before a typo check! J

Crystal Thompson said...

Hi Jane! There is TOTALLY light at the end of the tunnel. It looks differently for everyone, but improvement is always possible. I'm so glad you're doing something to help Rudy.

Also... don't put down your own abilities because you need professional help. Even the big names (like Ken Ramirez) consult with other trainers, and I find that just talking over my current progress and struggles with another trainer really helps clarify things. Plus, having another set of eyes on your dog is so helpful!

Good luck and keep me updated! :)

Anonymous said...

So, it sounds like you use an LRS when you are already in a training scenario (because you are interrupting reinforcement, right? So it assumes you are actively reinforcing at the time?)? Or am I not getting it?

Our main problem with dealing with some of Shanoa's undesirable behaviors is that they occur at random times when we are not training/reinforcing. I struggle with options to discourage those behaviors that do not use positive punishment. Right now, for example, if she has a barking outburst in the home, she's removed unemotionally to the bathroom for a few minutes until she can calm herself. I guess it's kind of a positive punishment (right?) but it's not aversive to her and seems to help her calm herself.

Not even sure if I'm making any sense!


Crystal Thompson said...

Nicky, you've got it! An LRS is an interruption in reinforcement, which assumes that you're in the midst of a training session.

I don't tend to worry about which quandrant I'm in. I worry more about whether or not I'm hurting or scaring my dog in the process of training. If I am, I try to avoid doing that. If I'm not- and if it's working- I worry less about it.

I think it gets more complicated with dogs with anxiety, because there is a physical issue (brain chemistry) causing the problem. I'm not convinced that punishment works when there's a physical issue; it's like suggesting you could prevent a dog from having low blood sugar if you hit them hard enough. Of course, there's lots we can do to help manage the situation and prevent low blood sugar (or anxiety attacks) from happening, but once that physical state has been triggered, I think you just have to deal with making it better.

I'd argue that what you're doing isn't actually a punishment (since the behavior isn't decreasing), but rather, just a method to help her calm down. Does that make sense?

Anonymous said...

So is an LRS technically negative punishment? Since we are removing attention from the behavior?