My last entry on the seminar was quite awhile ago, but I really did want to say more about it! Since this blog is about reactive dogs, I wanted to share what Pat said about them.
Pat says that almost all aggression and/or reactivity is based on some sort of stress, whether it’s because someone is threatening the dog’s territory or resource, or because the dog is scared or unsure. Since it’s based in emotions, most people choose to treat it classically, but in recent years, some people are adding operant components to the treatment. Let’s analyze the two, shall we?
Classical Conditioning
Since aggression is based on stress, it makes sense to treat that underlying emotion. If, for example, the dog displays aggressive behaviors when he sees a big scary dog, we need to teach him that bog scary dogs actually aren’t scary. In fact, they’re awesome! We do this by pairing something awesome (usually high value treats) with the sight of the big scary dog.
In order to do this effectively, we must keep the dog calm enough that he can see the trigger and eat the tasty treat. If he displays aggressive behaviors, we need to reduce the stimulus (usually by moving further away) and increasing the value of the treat. Pat recommends working with a single stimulus, at the same intensity, for 20 minutes or longer before switching to a new trigger or a new level of intensity. This helps the dog learn that seeing a big scary dog is actually a predictor of good things.
Sometimes, detractors will say that this is foolish, as we are simply bribing our dogs. Done correctly, this is not the case. We aren’t trying to prevent a reaction or distract the dog. We aren’t holding the treat out as a lure. Instead, we are allowing the dog to form a new association.
Operant Conditioning
Over the past ten years or so, various trainers have been coming up with ways to treat aggression in an operant, yet still “positive” manner. Pat talked about one of the first of these methods: CAT, short for Constructional Aggression Treatment, and showed a video of her doing CAT with a dog named Juni. I’d seen the video before, at the last seminar I saw with Pat, and I have to admit: I didn’t like the procedure then.
To do CAT, you stand with your dog (or you tether him), and bring a trigger to his threshold. You don’t go over the threshold, you simply bring him to the point of noticing, but not reacting. When the dog offers a socially appropriate behavior, or when the dog reduces the intensity of his aggressive behavior, the trigger retreats. Basically, you’re shaping calmer behavior, with the ultimate goal of “switch over,” which means the dog begins to offer affiliative behavior.
The reason I objected to CAT before was because it upset me when the dog did go over-threshold (and anyone working with a reactive dog knows that despite your best efforts, it happens). I didn’t like that the handler ignored her dog while he was upset, either. I also thought it looked like it could be flooding. On seeing the video a second time, though, I concluded that it probably wasn’t. To truly be considered flooding, the dog must be exposed to the stimulus at full intensity, and that exposure must remain until learned helplessness occurs. In other words, the dog would have to shut down emotionally. However, this is not what happens in CAT; it’s a highly regulated level of intensity, and the dog’s behavior controls whether the stimulus remains or not.
CAT is not without its critics, and there are even offshoots of the procedure attempting to make it more positive. Still, I’m no longer opposed to it like it was. Although I don’t think it should be the starting place for treating aggression, it is a useful tool for some dogs.
That, and… during the seminar, I found myself wondering: is this what I did with Maisy? I know that my plan wasn’t exactly CAT, as I did counter classical conditioning along with it, but when I did misjudge a threshold and she went over, I simply ignored it, and waited for a moderate duration (ten to twenty seconds) of calm, appropriate behavior before I rewarded her with attention or treats.
It worked, and it worked quite quickly. It did cause some stress and confusion in the short run, so I suppose that it was part of that 1% of the time where I’m not strictly “positive,” but Maisy quickly reached “switch over,” and has been pretty amazing ever since then. She seems far less stressed and much more interested in social interactions.
Anyway, I really enjoyed hearing Pat speak about aggression again. It’s amazing how much a year’s worth of reading and experience and, well, growth, allows you to take new things away from familiar ground.
Showing posts with label pat miller. Show all posts
Showing posts with label pat miller. Show all posts
Tuesday, May 25, 2010
Friday, April 30, 2010
Pat Miller Seminar: Functional Analysis
Last weekend, I had the good fortune to attend a single-day seminar with Pat Miller in Wisconsin. (No, I don’t spend all my time going to seminars, but yes, I’ve been lucky in the canine education department lately.) I had seen Pat previously, and the seminar was basically a scaled-down version of what I had seen a year ago. Even so, there were some very thought provoking moments, and of course, I want to share them with you!
Pat spent some time talking about operant conditioning, which I always enjoy thinking about. Of course, since I do spend a fair amount of time reading about operant conditioning, it was mostly review. (I’ll assume this is the case for most of my readers, but if anyone’s interested, I’d be glad to write an Operant Conditioning 101 post.)
What I found really helpful was Pat’s discussion of “Functional Analysis.” This is where we look at a behavior systematically using the mnemonic of “ABC” to remind us to look at the antecedent of the behavior, the behavior itself, and the consequence following the behavior, which either reinforces or punishes the behavior. The consequence is defined not by what we think should happen, but rather what actually happens.
In other words, if we do something as punishment, but the behavior doesn’t reduce in frequency or intensity? It wasn’t a punishment. If the behavior continues, it was simply aversive, and the dog has learned nothing. If, on the other hand, the behavior increases, our action was actually a reinforcer.
Although we can often correctly predict how the consequence will be perceived by our dog, we will sometimes be wrong. For example, Maisy thinks that getting squirted in the face with water is superawesomefun. If I were to use a squirt bottle as a punisher and didn’t pay attention to her response, I would be in for a pretty big surprise when the behavior I thought I was punishing actually increased. Since the dog’s subsequent behavior determines which of the principles we utilized, Suzanne Clothier is probably right: some of us trainers think too much.
Pat’s discussion on functional analysis yielded something else that was pretty cool: we actually have two opportunities to change behavior. You can manipulate the consequence so that behavior either increases or decreases, but you also have an opportunity to change behavior before it even happens. If we change or prevent the antecedent, we might be able to change the behavior. Doing it this way does seem a little iffier to me, but it’s definitely possible!
How? Well, management is one way. Although management has a high risk of failure at some point, it doesn’t always fail. And management is a good, useful, and sometimes necessary thing, not to mention often easier. I’ve written before about things I do to help manage Maisy’s reactivity.
For behaviors rooted in emotions, like reactivity, we can also prevent the behavior by changing the dog’s feelings about an antecedent. For example, if a dog always barks and lunges at a bicycle, we can change the dog’s feelings about the trigger through classical conditioning. (Again, I assume my readers are familiar with classical conditioning, but if you want a post on it? Just let me know!)
This discussion of functional analysis took topics that I already understood, and helped me think about them in a more sophisticated way. Has it changed the way I train? No, not really, but it has helped me understand why I do what I do a bit better, and I’m quite grateful for that. So, thank you, Pat, for the great discussion!
Pat spent some time talking about operant conditioning, which I always enjoy thinking about. Of course, since I do spend a fair amount of time reading about operant conditioning, it was mostly review. (I’ll assume this is the case for most of my readers, but if anyone’s interested, I’d be glad to write an Operant Conditioning 101 post.)
What I found really helpful was Pat’s discussion of “Functional Analysis.” This is where we look at a behavior systematically using the mnemonic of “ABC” to remind us to look at the antecedent of the behavior, the behavior itself, and the consequence following the behavior, which either reinforces or punishes the behavior. The consequence is defined not by what we think should happen, but rather what actually happens.
In other words, if we do something as punishment, but the behavior doesn’t reduce in frequency or intensity? It wasn’t a punishment. If the behavior continues, it was simply aversive, and the dog has learned nothing. If, on the other hand, the behavior increases, our action was actually a reinforcer.
Although we can often correctly predict how the consequence will be perceived by our dog, we will sometimes be wrong. For example, Maisy thinks that getting squirted in the face with water is superawesomefun. If I were to use a squirt bottle as a punisher and didn’t pay attention to her response, I would be in for a pretty big surprise when the behavior I thought I was punishing actually increased. Since the dog’s subsequent behavior determines which of the principles we utilized, Suzanne Clothier is probably right: some of us trainers think too much.
Pat’s discussion on functional analysis yielded something else that was pretty cool: we actually have two opportunities to change behavior. You can manipulate the consequence so that behavior either increases or decreases, but you also have an opportunity to change behavior before it even happens. If we change or prevent the antecedent, we might be able to change the behavior. Doing it this way does seem a little iffier to me, but it’s definitely possible!
How? Well, management is one way. Although management has a high risk of failure at some point, it doesn’t always fail. And management is a good, useful, and sometimes necessary thing, not to mention often easier. I’ve written before about things I do to help manage Maisy’s reactivity.
For behaviors rooted in emotions, like reactivity, we can also prevent the behavior by changing the dog’s feelings about an antecedent. For example, if a dog always barks and lunges at a bicycle, we can change the dog’s feelings about the trigger through classical conditioning. (Again, I assume my readers are familiar with classical conditioning, but if you want a post on it? Just let me know!)
This discussion of functional analysis took topics that I already understood, and helped me think about them in a more sophisticated way. Has it changed the way I train? No, not really, but it has helped me understand why I do what I do a bit better, and I’m quite grateful for that. So, thank you, Pat, for the great discussion!
Subscribe to:
Posts (Atom)