Operant Conditioning (explained)

[eeeefarm]

Terminology.

Often someone posts something referencing operant conditioning, mostly “positive reinforcement”. But there is more to operant than that. It helps if everyone is on the same page when using technical language, but often it becomes obvious that the terminology is not well understood, so herein is a brief refresher course.

Back in the day a guy named Pavlov conducted an experiment involving food, dogs, and bells. Pairing a neutral stimulus (bell) with an unconditioned stimulus that causes salivation (food) results in salivating at the sound of a bell. This is Classical Conditioning. It’s also what you do when you “load the clicker” in clicker training. You pair the sound of the clicker with the food reward.

A guy named Skinner coined the term “Operant Conditioning” for the process of conditioning responses by means of reward or punishment. In Operant Conditioning the reinforcement or punishment follows the behaviour to either strengthen it (reinforcement) or discourage it (punishment).

Positive reinforcement typically is the “click/treat” method most people are familiar with, but the other “positive” is positive punishment, i.e. an unpleasant consequence that follows the behaviour to discourage its recurrence. This could be as mild as a disapproving word, or perhaps a dose of water from a spray bottle, but to be effective it needs to be something the animal would prefer to avoid, otherwise known as an aversive.

Negative reinforcement is the application of pressure of some sort, which is removed when the animal performs the required behaviour. This is how horses typically are trained….e.g. pressure from the rider’s legs is removed when the horse moves forward, pressure from the bit is removed or lessened when the horse turns or stops. Negative punishment is removing something the animal wants, as in walking away from a nippy puppy, thus denying your attention that he craves. This could also be crating as a “time out”, denying him his freedom. Withholding expected food or praise is another example.

But here is where it gets tricky. Sometimes we misinterpret what the dog finds salient. If the response to bad behaviour lands our dog in his crate, it is the dog who decides if this is in fact negative punishment. If he has misbehaved because he is fed up with our training session, being confined to his crate might actually be preferable to the training, thereby becoming a positive reinforcement of the misbehaviour, since we are no longer bugging him to perform.

The most important element of training is observation. You can do everything “correctly”, but if it isn’t working you need to figure out why and modify what you are doing. Often people are not reinforcing what they think they are reinforcing. Ultimately it is the animal who decides what is reinforcing and what is punishing, and their response tells you what you need to know.

You don’t have to be a clicker trainer to employ operant conditioning. Many people who didn’t know the jargon have been using the principles since well before Karen Pryor popularized clickers and “all positive” training. And you don’t need a clicker, either. A marker word will suffice when using positive reinforcement. A clicker is more precise, but in practice isn’t always easy to employ. But if we want to use operant terms when discussing training it’s useful to know the precise meaning of them. To reiterate, positive means adding something, negative means taking something away. Reinforcement means increasing the likelihood of the behaviour being repeated, punishment means decreasing the likelihood of the behaviour being repeated. Clear as mud?

[elbrant]

And here I thought the "other positive reinforcement" was love, praise, "atta boy", pets, etc. Tell me something... was this a question, or a lecture?

[eeeefarm]

It was a clarification. These terms get bandied about quite often, and it helps if we are all talking about the same thing!

And yes, positive reinforcement can be anything the dog finds salient. Praise, toys, walks, cuddles, whatever. In fact, a great strategy for a dog that loves walks is to do your training session before your walk, so the walk acts as the reward, and you also have the option of "jackpotting" him by immediately taking him for his walk when he does something especially good.

Just to add, food is the most common reward because it is easy and works with most (not all!) dogs. Once paired with the clicker, the click itself becomes reinforcing because it signifies reward to come later, but may actually stand on its own. Using a marker word also tells the dog the specific thing that you will be rewarding and can become salient itself.

[elbrant]

So... this is more of a blog post, I'm not sure lectures and blog posts are appropriate for the forum. After all, we are offering opinions and advice based on our experiences, to questions. Since this isn't directly related to a question that someone has asked.... I'm left to assume that you just want everyone to know that you are an expert in this concept... ?

@eeeefarm said in What's in a name?:

Withholding expected food

Here's a question: This "scientific method" concludes that it's acceptable to "withhold food"? Like sending a child to bed without their supper, right? How long are forcing your dog(s) to go without food? Is it a matter of minutes? Or hours? Are you removing their food bowl or not filling it with their dinner? Or is this just a way to be a bully? What exactly lies behind the idea of not giving your dog(s) the piece of chicken (they can smell in your pocket) simply because they disappointed you in some way? Do you think that the dog is cognisant that they have wronged you in this way?

Our collective society frowns upon denying food to our children, labeling it as a form of abuse. Wouldn't denying food to an animal (in our care) be a form of abuse? Perhaps even animal cruelty?

You know... since you brought it up.

[eeeefarm]

@elbrant said in What's in a name?:

How long are forcing your dog(s) to go without food? Is it a matter of minutes? Or hours? Are you removing their food bowl or not filling it with their dinner?

Um, no. Just not giving them the treat they might expect to receive if they had performed the expected behaviour. Interestingly, some professional trainers suggest to their classes that they starve dogs that aren't all that food motivated so that they will be more interested in food rewards at class. Happened to a friend of mine because her Border Collie was disinterested in food treats as motivation at dog class. Don't expect all trainers to be well informed about breed differences. Border Collies are more motivated with praise rather than food. Basenjis, OTOH as a breed are more likely to respond to food.

In other disciplines, however, starving animals to get their compliance is common. Any honest dolphin trainer will admit that it's hard to get a sated dolphin to perform, so they withhold food to obtain the performance. We don't tend to do that with pet dogs, even if some professional trainers suggest it.

[eeeefarm]

@elbrant said in What's in a name?:

Since this isn't directly related to a question that someone has asked.... I'm left to assume that you just want everyone to know that you are an expert in this concept... ?

Not really. As stated, I wanted to clarify the terminology. It has come up on a number of past occasions and I thought it might be useful to post on it. As far as posing a question, well, lots of people post things that aren't really questions, e.g. brags about what their dog has achieved in the show ring or posts about how they are learning various techniques related to showing or breeding. Also information pertaining to food or dogs for sale or dogs in shelters needing homes, etc. I didn't know the forum was restricted to only questions, but if so I will be happy to cease posting things I think might be informative and interesting.

[Roux]

@eeeefarm I for one enjoyed the post, it's interesting and sometimes intriguing to see opinions, methods, theories, etc with dog behavior and training. For those with less experience and joined this group to learn, posts like this AND the replies that come after are very useful.
I do think this should be a place to share "cool stuff" and everyone should feel free to discuss how they feel about it. Is that not the use of a forum?

[eeeefarm]

@Roux Thanks. I was hoping it would be useful to clarify terms that are often used incorrectly.

One of the things about terminology is the way people perceive labels. You hear "punishment" and immediately many picture the poor dog being whipped to within an inch of its life, but punishment in operant conditioning terms is anything that makes it less likely that the unwanted behaviour will recur. It doesn't denote a violent thrashing of the animal! Also, many use punishment, either positive or negative, without recognizing it as such. Again, it helps to clarify what it actually is.

Reinforcement is also subject to misinterpretation but people are not put off by the term.

Again, I would like to emphasize that the dog is the ultimate arbitrator of whether your action is reinforcement or punishment. Observation will tell you if you are on the right track.

[elbrant]

@eeeefarm said in Operant Conditioning (explained):

some professional trainers suggest to their classes that they starve dogs that aren't all that food motivated

This is just disturbing... in today's society, this type of behavior would be considered animal cruelty. Certainly we can find a way to encourage compliance without depriving an animal basic care.

[eeeefarm]

@elbrant said in Operant Conditioning (explained):

@eeeefarm said in Operant Conditioning (explained):

some professional trainers suggest to their classes that they starve dogs that aren't all that food motivated

This is just disturbing... in today's society, this type of behavior would be considered animal cruelty. Certainly we can find a way to encourage compliance without depriving an animal basic care.

I know, it shocked me too when I heard it. The first instance was a friend of mine whose Border Collie was disinterested in food rewards, and she was advised by the instructor not to feed the dog anything on class days so there would be an incentive to accept treats. The second instance was my niece, and in this case a German Shepherd dog that again didn't want to take treats, and she was given the same advice by a different trainer, in fact in a different city, so it appears that it isn't unusual. Both these dogs work well for praise and in both cases the owners declined to starve the dogs....