Operant conditioning

[JohnLloydJones]

Elsewhere on the boards I read, once again, a discussion (and I use the term diplomatically) about training methods; much of the hot air comes down to a dogma about use (or not) of +R (aka click-and-treat) methods. The atmosphere is heavily laden with the assumption that if you don't us +R (and only +R) then your training methods are cruel and mean -- you use Punishment. I believe much of this is a misunderstanding of terms. In operant conditioning, the terms have a clear meaning.

 

Reinforcement: something which causes an increase in repeating the behavior that was displayed.

Punishment: something which causes a decrease in repeating the behavior that was displayed.

 

 

In the argot of an operant trainer:

 

My dog does something I like (i.e. desired behaviour) and say "Yeah, good girl" -- That's R+.

My dog does something I don't like (i.e. undesired behaviour) and say "Eh!" -- That's P+.

 

I look happy until my dog does something I don't like; then I stop looking happy. -- That's R-.

I scowl until my dog stops doing something I don't like; then I stop scowling. -- That's P-.

 

There, now I have admitted that I use all four quadrants; scold me.

 

[Edited to fix typo pointed out by Shetlander]

[Shetlander]

The atmosphere is heavily laden with the assumption that if you don't us +R (and only +R) then your training methods are cruel and mean -- use use Punishment. I believe much of this is a misunderstanding of terms. In operant conditioning, the terms have a clear meaning.

 

Really? I don't recall many people making that assumption. It seemed to be more some liking a clicker +R approach and others finding it not needed or detrimental.

 

Your explanation of +R +P -R -P are not correct, I believe. My understanding is:

 

+R Behavior earns something positive, e.g. cookie, praise

 

+P Behavior earns something negative, e.g. scolding, collar jerk

 

-R Behavior loses something negative e.g. ear pinch stops when dog takes dumbell

 

-P Behavior loses something positive, e.g. food removed, access to yard ends (time out)

[Sekah]

Reinforcement: something which causes an increase in repeating the behavior that was displayed.

Punishment: something which causes a decrease in repeating the behavior that was displayed.

 

My favourite part of R+/P- discussions is that people do not accurately track whether or not a particular method has actually caused an increase/decrease in a specific behaviour. This isn't apparent until later once the subject can be observed again. People get hung up on quadrants without actually observing the result, rendering quadrants useless. (Did the leash pop actually result in a decrease in lunging behaviour?; did the cookie actually result in increased eye contact? or are you simply nagging/feeding a dog?) In my experience, people with current science-based animal behaviour backgrounds are rankled by the overuse/misuse of labelling of the operant conditioning quadrants and are trying to distance themselves from it.

 

Try not to get too hung up on labels. My personal barometer is ascertaining whether I'm being effective and whether I am improving my relationship with the subject.

[Shetlander]

Much better explanation of the four quadrants than my attempt, Sekah. Thanks!

[JohnLloydJones]

Your explanation of +R +P -R -P are not correct, I believe. My understanding is:

 

+R Behavior earns something positive, e.g. cookie, praise

 

+P Behavior earns something negative, e.g. scolding, collar jerk

 

-R Behavior loses something negative e.g. ear pinch stops when dog takes dumbell

 

-P Behavior loses something positive, e.g. food removed, access to yard ends (time out)

The goal of operant conditioning is to modify behaviour. Thus an R+ response is the addition something that increases the frequency of that behaviour (e.g. praise). A P+ response is the addition of something that decreases the frequency of that behaviour (e.g. scolding). The verb "earns" is not strictly part of operant terminology, but if we treat it as a neutral reference to a positive response, I don't see exactly where we differ.

[Shetlander]

My dog does something I like (i.e. desired behaviour) and say "Yeah, good girl" -- That's P+.

My dog does something I don't like (i.e. undesired behaviour) and say "Eh!" -- That's R+.

I look happy until my dog does something I don't like; then I stop looking happy. -- That's R-.

I scowl until my dog stops doing something I don't like; then I stop scowling. -- That's P-.

 

I am not good at these discussions and should not keep trying, but here goes...P+ is positive punishment, something added to decrease the behavior and R+ is positive reinforcement, something added to reinforce the behavior. Unless your dog dislikes praise and enjoys hearing "Eh!" you have reversed the R and P in the first two examples. Same thing in your examples of negative punishment and reinforcement. You have them reversed.

[GentleLake]

^^ This is my understanding, as well.

 

R = reinforcement; P = punishment.

 

There is also P+. My dog pulls on the leash; I pop the collar with the leash (or I beat the dog, in a more extreme example).

 

Negative punishment (P-) might be a puppy nips, you remove toys, yourself, all attention away. It's a consequence to unwanted behavior that removes something the dog wants.

[Eileen Stein]

JLJ, go back and read your first post. You reversed your R+ and P+ in your verbal example, and the R- and P- in your scowling example. It's obvious from your otherwise sound explanation that the reversals were inadvertent.

 

But I have to add that very, very seldom have I seen the use of these terms -- whether correctly or incorrectly -- lead to better understanding of dog training.

[JohnLloydJones]

Shetlander, on 11 Jul 2014 - 12:13, said:

I am not good at these discussions and should not keep trying, but here goes...P+ is positive punishment, something added to decrease the behavior and R+ is positive reinforcement, something added to reinforce the behavior. Unless your dog dislikes praise and enjoys hearing "Eh!" you have reversed the R and P in the first two examples. Same thing in your examples of negative punishment and reinforcement. You have them reversed.

I need new glasses. I'll fix the OP.

[GentleLake]

Here's a very nice chart that clearly defines all the terms: http://careforreactivedogs.com/start-here/

[Tea]

Eileen, I cannot even red this + or - stuff without freaking out about the horror which was my high school.

[Eileen Stein]

Eileen, I cannot even red this + or - stuff without freaking out about the horror which was my high school.

 

[Root Beer]

As someone who utilizes both operant and classical conditioning in training on a very regular basis, I don't often think or talk in terms of quadrants. And I rarely talk about quadrants when I work with students in training classes. I speak far more in "plain talk" even if the matter at hand could technically be described in behavior theory terms.

 

Sometimes basic understanding of quadrants is helpful to me on occasions when I am faced with a training challenge that I really need to take some time to consider carefully in order to formulate a new approach. In those cases, I am likely to be thinking in terms of the value of the reinforcer, what the dog actually perceives in a given situation, the degree of the criteria that I am asking for, and the possible effects of circumstances beyond my control on the work at hand, but my understanding of how those things function is deeper and clearer because of an understanding of quadrants.

 

Interestingly, I sometimes talk to my High School students about dog training because they find the fact that I train dogs to be very interesting. I often get the question, "are you like the dog whisperer?" I tell them that I am pretty much the opposite of him and they want to know how and why. Inevitably, and without fail, when I start to explain how I train, they immediately bring up the behavior quadrants! From that point I can use the quadrants to explain how I train, what I do when I want a dog to stop doing something (which is always a burning question), and what the dog whisperer does and how I am different. Understanding the quadrants has given me the ability to speak their language on the subject, and I always enjoy those discussions.

I have found that having knowledge of the quadrants is the most useful in discussion with other dog trainers. Whether one agrees with behavior theory or not, reference to the quadrants can provide a good way for different trainers to compare and contrast their training approaches. Of course, there is a good bit of misunderstanding out there, but it is useful and helpful to a notable degree.

[Eileen Stein]

Inevitably, and without fail, when I start to explain how I train, they immediately bring up the behavior quadrants!

I find that very depressing.

I have found that having knowledge of the quadrants is the most useful in discussion with other dog trainers. Whether one agrees with behavior theory or not, reference to the quadrants can provide a good way for different trainers to compare and contrast their training approaches.

Well, yes, I suppose it's a good way for trainers to "compare and contrast their training approaches" in the sense that they can say (as the website Gentle Lake referenced does) "For ethical reasons . . . we use only the R+ quadrant of operant conditioning." Me humane, others inhumane. But beyond using it as a litmus test of ethics and humanity, I don't see how fitting a method into a quadrant adds any clarity or insight. It just shifts discussion from the concrete to the abstract, and the concrete is where dog training takes places. For example, that same site Gentle Lake linked classes a leash correction with a prong collar as R-. I should think it would be more common to class it as P+ -- I certainly would, if I were sucked into using quadrant terminology. But what the heck difference does it make? Isn't it more productive to talk about the method itself, its advantages, disadvantages, observed and likely effects, available alternatives and their observed and likely effects, etc., without the tedious and tendentious process of assigning it a place in the almighty quadrants?

[Debbie Meier]

always laugh when people say they only use the R+ quadrant....my question is, exactly what quadrant are you using when your withholding that cookie that your dog clearly wants until the dog offers you the behavior your asking for?

[Root Beer]

I find that very depressing.Well, yes, I suppose it's a good way for trainers to "compare and contrast their training approaches" in the sense that they can say (as the website Gentle Lake referenced does) "For ethical reasons . . . we use only the R+ quadrant of operant conditioning." Me humane, others inhumane. But beyond using it as a litmus test of ethics and humanity, I don't see how fitting a method into a quadrant adds any clarity or insight.

 

No, Eileen - there is much, much more to it than that.

 

In fact, very few of the best training discussions that I have taken part in or been party to have been about ethics at all.

 

There is certainly a time and a place to discuss ethics in dog training, but if that were all we were talking about, nothing would ever get done. We discuss training goals (what we wish to accomplish), the what and the how of getting the training accomplished, results (yes, I know +R trainers are "supposed" to be unconcerned with results, but the vast majority of us missed that memo), challenges faced along the way and how we dealt with them, etc.

 

 

It just shifts discussion from the concrete to the abstract, and the concrete is where dog training takes places. For example, that same site Gentle Lake linked classes a leash correction with a prong collar as R-. I should think it would be more common to class it as P+ -- I certainly would, if I were sucked into using quadrant terminology.

 

All the terminology does is give us a common language by which we can discuss how the tool works. Some trainers use a prong to apply an aversive when the dog is doing something they don't want (example: use it to flip a dog backwards if the dog does not recall at the end of along line). Some use the prong to apply an aversive in the first place and then take it away when the dog is doing what they want (example: keep the prongs employed until the dog moves into heel and then allow them to release). The collar is actually used in both ways by different people.

 

I know that people tend to read ethics into such statements, but to discuss how the device can be used as -R and how the device can be used as +P is simply a discussion of how the device works. And trainers who enjoy, or find value, in discussing the whys and wherefores behind different training approaches and methods find this terminology useful.

 

I find it useful.

 

 

But what the heck difference does it make? Isn't it more productive to talk about the method itself, its advantages, disadvantages, observed and likely effects, available alternatives and their observed and likely effects, etc., without the tedious and tendentious process of assigning it a place in the almighty quadrants?

 

I guess I simply don't consider consideration of what quadrant (or quadrants) a particular approach or method falls into to be any more tedious than it would be to consider what food group a particular food item falls into (yes, I will actually enjoy a discussion of corn as a grain that we commonly consider a vegetable, or the tomato a fruit that we commonly consider a vegetable and I find it interesting, and a bit amusing).

 

One can certainly discuss the ethics of training that falls into one quadrant or another, but the quadrants themselves are not ethical terms.

[Root Beer]

always laugh when people say they only use the R+ quadrant....my question is, exactly what quadrant are you using when your withholding that cookie that your dog clearly wants until the dog offers you the behavior your asking for?

 

If the use of the "cookie" (food, toy, access to something the dog wants, praise, etc) serves to increase the behavior that the dog is in the process of learning, then it is, by definition +R, even if one is withholding it until the dog does what is intended.

 

If the use of the "cookie" causes the dog to stop doing something, then it is -P.

 

In point of fact, most, if not all, trainers who base their training almost exclusively in the +R quadrant do utilize some degree of -P. We know it, we have absolutely no problem with it, and we actually don't intend to lead anyone to believe that we don't. From a practical standpoint, the use of -P tends to be so sparse that it simply does not describe the vast majority of what we do. So, it tends not to get mentioned.

[juliepoudrier]

Eileen,

I think this is just one of those cultural difference things that cannot be easily bridged.

 

J.

[Root Beer]

Eileen,

I think this is just one of those cultural difference things that cannot be easily bridged.

 

J.

 

I think you are correct, Julie.

 

I know it can be bridged in some cases. I have a lot of friends and colleagues who would likely think more along the lines of what Eileen is saying and we work side by side and communicate with one another very well and we can discuss training and agree on some things and disagree on others and somehow it works.

 

I think it is a lot more difficult online.

[Tess's Girl]

I know that people tend to read ethics into such statements, but to discuss how the device can be used as -R and how the device can be used as +P is simply a discussion of how the device works. And trainers who enjoy, or find value, in discussing the whys and wherefores behind different training approaches and methods find this terminology useful.

 

I find it useful.

 

 

 

I guess I simply don't consider consideration of what quadrant (or quadrants) a particular approach or method falls into to be any more tedious than it would be to consider what food group a particular food item falls into (yes, I will actually enjoy a discussion of corn as a grain that we commonly consider a vegetable, or the tomato a fruit that we commonly consider a vegetable and I find it interesting, and a bit amusing).

 

This is what I think. To me, it's not tedious and it is very useful to me to discuss why or why not a particular training method I employ is working.

 

I also come from a psychology background. In my undergrad work I took classes on learning and cognition, I worked in the pigeon lab and in the dog cognition lab on campus. In working with those animals (we used the clicker with the pigeons and the dogs) and working on studies you have to be able to somehow quantify what is happening, and that's where the quadrants come in. They give a universal language to discuss very different behaviors and techniques.

 

To use the prong collar as an example, someone could say "The prong collar stops the dog from pulling by applying pressure" which does explain the process in a concrete way, but I want to know how it accomplishes that. So discussing the quadrants isn't tedious for me, though I can see how it is for some.

 

But not all dogs learn the same, and not all people train the same. The wonderful thing about dogs is that they are so in tune with people and so good at communicating with them you can use a variety of methods to get to the same goal. It's still up in the air about what method truly works best with dogs, and to me that's fascinating.

 

Personally I love to clicker train. I've had lots of experience with it, I think it helps my timing, and I think it's a good communication tool. Yet, I own a dog who hates the clicker and is not very food motivated. When I got her I had to rethink my process and understand what motivates her and how she learns best. And in that process I thought about the quadrants and what they meant for learning and for Tess in particular.

 

It's how I think and how I was introduced to dog training, and I think that's a lot of it. Had I been introduced a different way I would probably think about it in a different way. I do understand how some people find it tedious and why they don't employ the same thought process The conversations that I have read on this board about the different methods of training and thinking have been fascinating and overall very respectful. I enjoy reading them and learning from them.

[mum24dog]

People who trial their dogs also have their own arcane linguistic code that most people aren't interested in but it serves as a useful communication tool between them.

 

I can't see any difference between them and those who talk of "quadrants" and the like.

[juliepoudrier]

There is no difference Pam. I was simply pointing out that people like Eileen (and me) don't find operant conditioning terminology useful because we come from a different training culture. I did not place a value judgment, so no need for anyone to get their back up.

 

J.

[Laurelin]

I tend to stay out of these because I don't sheepdog and I'm a novice at other training. I will say though that I do find training discussions run a bit clinical at times and I think that is what some find off putting about terms such as +R, -P, etc.

 

I understand the terms but I still find it tedious and splitting hairs sometimes. And I find some people focus a bit too much in my opinion on where things fit scientifically or whether or not a certain method fits into their trainer box they've put themselves in vs if it works.

 

I will say though that in my dog sports classes the terms don't come up that much though if you break it down most agility is taught via +R. Maybe the first class they will briefly cover it and what shaping is and how to clicker train but in my experience that is about all it shows up. In practical discussions it is not going to show up at all. I take the conversations about this quadrant vs that a little more academically instead of being terribly helpful practically.

 

It seems to me that there is a lot more common ground than it may come across at first. I have only met one trainer who was really focused on using only positive reinforcement. She was a very accomplished trainer though with many more dogs than I will probably ever accomplish anything with. So I was not one to argue with her! But most trainers really don't seem to focus all that much on it and are open to using various methods depending on the dog in front of them.

[rushdoggie]

will say though that in my dog sports classes the terms don't come up that much though if you break it down most agility is taught via +R. Maybe the first class they will briefly cover it and what shaping is and how to clicker train but in my experience that is about all it shows up. In practical discussions it is not going to show up at all. I take the conversations about this quadrant vs that a little more academically instead of being terribly helpful practically.

 

It seems to me that there is a lot more common ground than it may come across at first. I have only met one trainer who was really focused on using only positive reinforcement. She was a very accomplished trainer though with many more dogs than I will probably ever accomplish anything with. So I was not one to argue with her! But most trainers really don't seem to focus all that much on it and are open to using various methods depending on the dog in front of them.

 

Yep.

 

I have lived in various places and trained with various groups and no matter where I was in reality no one in real life gets wound up or terribly judgmental as you find on the internet. I teach both pet and advanced obedience classes, and we don't talk about this much if at all. And unless someone is being really inappropriate or unless they ask for advice, I don't critique my peers either.

[d112358]

I will say though that in my dog sports classes the terms don't come up that much though if you break it down most agility is taught via +R. Maybe the first class they will briefly cover it and what shaping is and how to clicker train but in my experience that is about all it shows up. In practical discussions it is not going to show up at all. I take the conversations about this quadrant vs that a little more academically instead of being terribly helpful practically.

 

There are definitely ongoing sideline conversations with people coming to agility from a compulsion/correction school in the classes I've been through and helped with. Not so much quadrant talk, but stuff like the difference between a lure and a reward. Another common discussion happens around shaping confidence with a new obstacle versus slapping a leash on them and dragging them over it (flooding). It definitely helps to avoid the clinical terms; it also helps to focus on how we want the dog to run - fast, confident and independent rather than tentative and slow.