- Location
- Andorra and Spain
WARNING! You are in the "behaviour geeks" section!!
Just thought I'd put that in to avoid people's brains getting inadvertently fried
Here's the thing, someone said something along these lines in another group earlier:
"If you are trying to teach a conflicting behaviour, you are inadvertently punishing the behaviour you don't like as you reinforce the one you do".
I see where the person is coming from: we define punishment as something that reduces the frequency/likelihood of a behaviour happening in the future. So, if a behaviour is happening less, it must be being punished, right?
Hmmm, I disagree. Because the definition of punishment is actually a consequence of a behaviour that makes that behaviour less likely to happen again in the future. So, if there's no behaviour, there can be no consequence, and ergo no punishment.
Thinking about a real life example, and the one that was given was a dog jumping up to greet. They wanted to reinforce the polite greeting so that became the norm, and the jumping up behaviour was made less frequent.
I say that, if you manage the antecedent arrangement (environment, arousal etc) so that the dog is unlikely to jump up, and heavily reinforce the conflicting behaviour, then you are not punishing the jumping up because it's simply not happening. It will eventually die out because you are building a very strong reinforcement history for the conflicting behaviour you do want.
Then I got on to thinking about matching law. Because, in the real world, the chances are that we won't be managing the environment 100% and, at times, the dog will still jump up. For many dogs, jumping up at someone is inherently reinforcing, it doesn't need a treat or particular attention to make it more so. Sometimes turning your back or ignoring the dog works as a punisher, but for many dogs, just the jumping itself is reinforcing. Let's say we don't do anything differently when the dog jumps up, so they are obviously getting reinforced, otherwise they wouldn't do it.
In that case, we have the jumping behaviour and the incompatible behaviour (let's say a sit - although I don't like that in the real world as an alternative greeting behaviour). Let's say that the reinforcement for jumping is of the same value as the reinforcement given for sitting. So, if we were to arrange the environment in such a way that we could ensure the dog sat ten times more frequently than he jumped up (and was reinforced each time), then the matching law would say that the dog was ten times more likely to sit than jump up in the future. Therefore, we can reduce the frequency of a behaviour happening without using punishment at all.
I looked a little more into this, and came across the Humane Hierarchy, which you might have seen before:
So what I'm talking about here is DRA (Differential Reinforcement of Alternative behaviour), sometimes referred to as DRI (I = Incompatible). So you can see that it's on the hierarchy, but isn't considered as humane as "pure" R+, presumably because you're dealing with conflicting behaviours and we know that conflict creates stress.
But, looking higher up the hierarchy, there we find Antecedents, which is the environment and everything in it. So if we manipulate the environment so that the behaviour doesn't happen, AND employ R+, and preferably deal with the dog's arousal state which is driving the behaviour, which also comes under antecedents, then we're as high up that hierarchy as we can possibly be - assuming we've already ruled out the behaviour being driven by anything on that top layer.
All this going to say that we can absolutely make a behaviour less likely to happen without punishing it in any form whatsoever.
Just thought I'd put that in to avoid people's brains getting inadvertently fried
Here's the thing, someone said something along these lines in another group earlier:
"If you are trying to teach a conflicting behaviour, you are inadvertently punishing the behaviour you don't like as you reinforce the one you do".
I see where the person is coming from: we define punishment as something that reduces the frequency/likelihood of a behaviour happening in the future. So, if a behaviour is happening less, it must be being punished, right?
Hmmm, I disagree. Because the definition of punishment is actually a consequence of a behaviour that makes that behaviour less likely to happen again in the future. So, if there's no behaviour, there can be no consequence, and ergo no punishment.
Thinking about a real life example, and the one that was given was a dog jumping up to greet. They wanted to reinforce the polite greeting so that became the norm, and the jumping up behaviour was made less frequent.
I say that, if you manage the antecedent arrangement (environment, arousal etc) so that the dog is unlikely to jump up, and heavily reinforce the conflicting behaviour, then you are not punishing the jumping up because it's simply not happening. It will eventually die out because you are building a very strong reinforcement history for the conflicting behaviour you do want.
Then I got on to thinking about matching law. Because, in the real world, the chances are that we won't be managing the environment 100% and, at times, the dog will still jump up. For many dogs, jumping up at someone is inherently reinforcing, it doesn't need a treat or particular attention to make it more so. Sometimes turning your back or ignoring the dog works as a punisher, but for many dogs, just the jumping itself is reinforcing. Let's say we don't do anything differently when the dog jumps up, so they are obviously getting reinforced, otherwise they wouldn't do it.
In that case, we have the jumping behaviour and the incompatible behaviour (let's say a sit - although I don't like that in the real world as an alternative greeting behaviour). Let's say that the reinforcement for jumping is of the same value as the reinforcement given for sitting. So, if we were to arrange the environment in such a way that we could ensure the dog sat ten times more frequently than he jumped up (and was reinforced each time), then the matching law would say that the dog was ten times more likely to sit than jump up in the future. Therefore, we can reduce the frequency of a behaviour happening without using punishment at all.
I looked a little more into this, and came across the Humane Hierarchy, which you might have seen before:
So what I'm talking about here is DRA (Differential Reinforcement of Alternative behaviour), sometimes referred to as DRI (I = Incompatible). So you can see that it's on the hierarchy, but isn't considered as humane as "pure" R+, presumably because you're dealing with conflicting behaviours and we know that conflict creates stress.
But, looking higher up the hierarchy, there we find Antecedents, which is the environment and everything in it. So if we manipulate the environment so that the behaviour doesn't happen, AND employ R+, and preferably deal with the dog's arousal state which is driving the behaviour, which also comes under antecedents, then we're as high up that hierarchy as we can possibly be - assuming we've already ruled out the behaviour being driven by anything on that top layer.
All this going to say that we can absolutely make a behaviour less likely to happen without punishing it in any form whatsoever.
