REWARD MARKER VS "good dog".......

Well, I don't post much, but browse occasionally and thought this was interesting that Joel is actually a marine mammal trainer but is seemingly using the clicker in a way that other marine mammal trainers don't or any animal trainers for that matter....as far as I have come to know, at any rate. It should be clarified, simply so that people learning don't get more confused.

A few concepts which I felt like throwing out even though some of them have been thrown out, but thought I'd word it in yet another way...my take on the subject:

It is really, really important to keep in mind that reinforcement is what drives behavior. Cues do not.

Operant conditioning Angelique, is indeed in everything. It's in the air we breath. LOL. It is how all organisms with a brain learn, how behavior is modified. It is the only way behavior is modified. Even the little fishies in my aquarium learn that most times when I approach the aquarium, food is on it's way. They all swim rapidly to the surface in anticipation. There is a conditioned reflex in those little critters. It is very simply consequences for any action. Do this and this happens. Do that and that happens. All behavior has consequences.

Clicker training, (using a clicker) how the vast majority of "trained" trainers utilize it is based on Pavlovian response. It is classical conditioning, as opposed to operant conditioning. Classical conditioning is the creation of a conditioned reinforcer or conditioned reflex in the animal to realize value in an otherwise neutral thing that has no inherent value in itself. But comes to represent something of value to the animal because it is tied to something really wonderful (a primary reinforcer) through association or (Pavlovian response) The value of that thing will be lost if the association to the primary reinforcer (the meat or tug game or whatever) is not paired with it. You might get away with it a few times, (clicking and not treating) but after some time, the association will be lost. This is one of those laws of behavior.

The only time this use of bridging a conditioned reinforcer (the clicker) to a primary reinforcer (treat) is necessary is when learning a new behavior. Once they have had enough history of being reinforced, they no longer need the communication that what they just did earned them the reward. They already know. The cue (introduced a little later) has by this time been tied with the behavior and is used to elicit the behavior. The reinforcement history is enough to propel them to do it again. The clicker is then dropped and not used for a behavior which is being performed quite regularly.

THEN the behavior is put on a variable reinforcement schedule. The clicker is not used in conjunction with a variable reinforcement schedule Joel, because the former, as was mentioned is only needed in the beginning of learning a new behavior....something unfamiliar to the animal. And the variable reinforcement schedule, conversely is not used in the beginning because reward must be (if you want the most bang for your buck) on a continuous schedule at first to make the animal rule out any other incidental behaviors he may be doing at the same time...so eventually the animal can stop guessing at which behavior you're targeting. This consistent, rapid, high rate of reinforcement is what separates the wheat from the chaff, so to speak, to the animal.

Only when he is "getting it" consistently is a variable schedule used....to keep the animal trying harder, keeping interested and thus, strengthening the behavior. In other words, a clicker as a marker and a variable schedule of reinforcement are diametrically opposed in usage...incongruous or irrelevant...however one wants to word it.

Clicking without reward following will make the clicker lose it's meaning. This is behavioral law and what Pavlov discovered when ringing the bell before food for his dogs. When he stopped ringing the bell, after a time, they stopped salivating at the sound of it as they had before when the conditioned reflex had been developed to the sound of the bell.

Duration training:

There is a no reward marker, a reward marker and there is something that some trainers use, myself included called a "keep on going" signal. That is just to verbally let them know that they're on the right track. And encouragement type of thing. But the big reward needs to be reserved for the degree of the behavior you want, regardless of what increments you use. (for example, baby steps for straighter sits, or faster recalls). It's like a reward or NRM. A NRM isn't simply an "eh eh" or a "too bad." They have to know that they WOULD have gotten a treat but didn't. And they don't connect that necessarily simply by words alone. To condition or prime them to the NRM, you can hold a treat in your fist and swish it past them, giving your verbal just before. So, when they hear that NRM, they come to associate it with disappointment. Same thing with the reward marker. It needs to be connected strongly with a consistent consequence....a reward that is a reinforcer to the dog.

As it was said, you can delay the click in order to develop duration....one second more at a time. But still, follow with a reinforcer. If the dog is unable to "get" the duration of something because he's losing focus, instead of clicking without a reinforcer to follow, try reinforcing more frequently for much tinier responses or baby steps. And then more gradually than you have been, add a second longer. Don't expect 10 or even 5 seconds more all at once. Then you avoid the risk of losing the value of the very effective, precise communication tool that classical conditioning provides.

I would think that any animal trainer would find it advantageous to utilize of the laws of behaviorism rather than sloppy guess work. We can all speculate and label dogs as "dominant" or "stubborn" any way we want. We can imagine what they're thinking, what we think they think we're thinking, what kind of hierarchy, (if any) they have and what kind of energy and attitude we all have. But the truth is that none of that has been concluded, that the only thing that has been proven to work consistently and across the board are the laws of learning because they are animals capable of learning.

Always good to see you posting, Carrie.

Actually, my issue isn't so much with "operant conditioning is everything", but with the insistance that it is the only thing, the only form of learning present. I also don't lump classical conditioning and operant conditioning together. Two separate fields of study by two different scientists.

I don't ever discount social learning. Unless the subject is isolated from all social contact and information exchange, social learning will be present.

It's how social animals find food, save energy, and avoid danger = The untimate reward of living long enough to pass on their genes. There is little instant gratification in nature.

Operant conditioning, classical conditioning, and social learning all exist and work in harmony with each other...unless you (not you personally) are a radical behaviorist (sometimes known as a "Skinnerite, as opposed to a Cesareenie ). Ever read the debates on that?

Funny, Joel's show is not called "Clicker Training U", it's called "Good Dog U". Now, is praise, or is it not, a reward?

It's the one I use the most, works great.

Bart the Bear was trained with a "goood boooy!!!" by his trainer, Doug Suess when he did something correct. So Doug would sometimes give him a pop, but not everytime. Yet everytime he did a behavior correct, Doug would let him know with a "goood boooy!!!" and social reinforcement, pop or not.

Now, that sounds more like what Joel is describing. He is marking a correct behavior with a social communication (which is also a reward) but not as a predictor of a food treat.

Know what? I'll bet a lot of animal trainers do it both ways.

Finding food, avoiding danger, choosing to run or not to run from something or after something is still all choosing behavior which has conseqences. Dig into the garbage pile (behavior)......get food. (consequence) (Operant conditioning) Run from the Grizzly bear and dive into a hole in a log (behavior) Being safe (consequence) (Operant conditioning)
Hear the human scraping out the usual, nightly left overs into a bowl. (wonderful noise.....scraping is good) (conditioned reinforcer) Meat, potatoes and green beans (primary reinforcer) (Classical condtioning) In nature, pariah dog hears or sees humans throwing out garbage. Yeah! Humans are good. All social. All operant or classical condtioning at work.

Operant conditioning and classical conditioning are two slightly different things. That is true, as I mentioned in my post, describing the difference. But they're both the parts of what is going on which causes us all to learn or to change behavior. One is simply or basically consequences to behavior. "Do this and this happens." And the other is when something with no inherent value comes to have value due to an association made with it and something else that does have value.

It is a behavioral law of learning that all animals with a brain learn by these two factors; operant and classical conditioning.

I'm not sure what you mean by "social learning." I take it to mean that you're offering the dog normal, everyday communication we use with our dogs and with everyone we interact with. "Good dog" or a scratch behind the ears or "let's go for a walk." (something dogs love to do) Those things are ALL consequences for behavior. They are all potential reinforcers. They have the opportunity to be tied to a behavior. And they have the ability also, to be tied to a primary reinforcer, there by making them a conditioned reinforcer. (classical conditioning)
If you say, "good dog" to a dog who doesn't get much in the way of a primary reinforcer or that receives a lot of punishment, even mild, that verbal praise can have some meaning as a reinforcer. It signifiies that the dog isn't likely to get a collar correction, if collar corrections are used regularly. Punishment and praise don't usually happen at the same time. So, naturally, it's a pretty dang good thing to most dogs. But not reliable. It's not something I rely on to raise the odds to a significant degree that I'll get a dependable, repeated response in the future.

If "good dog" comes to mean that something the dog really likes is likely to follow, then it becomes a conditioned reinforcer. For instance, I often am pleased with something one of my dogs does. She gets out of the way quickly when she's been standing by the door, wanting to go out. She backs up rapidly and sits and waits when she sees me coming so I can get to the door easily. That behavior is rewarded by "good girl" and a happy voice that she likes and opening the door so she can go out. When my dogs stop barking on cue for me, I show pleasure and they recognize that and like it. Usually, when I'm pleased and happy, good things happen, not bad. So, the dogs like that. That, you can call social learning, right? Sure it is. But it's still the laws of behaviorism at work which are driving behavior. It's going to be either operant or classical condtioning which is going on. Do this and something happens, either favorable or unfavorable. Something will happen which will cause the dog to either tend to repeat the behavior or let the behavior extinguish or change the behavior in some way. Behavior is never static. It will evolve constantly in some manner, minutely or largely or anywhere in between. And that is dependent on operant or classical conditioning....on what happens following a behavior. Depending on just how the dog views the thing that happens when he performs a behavior, that dictates just how the behvaior will change. My dogs like praise and it works a little bit to increase the likelihood of a change in behavior. But they get a lot of attention anyway, in every day, normal social interactions. It's not out of this world, fantastic. It's good, but it's not that good. But if I give them a piece of Porterhouse steak with mushroom sauce spread all over it when they give me a really good recall, you better believe it has more value to them than "good dog." It just strengthens and quickens the response next time and increases the reliability of the response. Get the timing right in training and choose appropriate rewards for various things (they do vary) and it all makes for more reliable, qiucker, more precise responses to cues.

So, if you mean the definition of "social learning" to be confined to verbal or physical communication, body language, happy voices....yes indeed. That's all incorporated in everything we do with dogs. It's why we have dogs in our lives...to have this phenomenal relationship with another species that we do, a species hard wired through evolution and domestication...designed especially to live with and be sociable with us. They understand so much about us. They probably read us better than we read them. I think they understand more than we give them credit for sometimes. They're just flat out phenomental in their social ways. But all that doesn't delete the laws of learning. Behaviorism, operant and classical conditioning is right in there, at work all the time...constantly. And that is the case with every living organism.

Every single move we make, every single thing we do is either done because of operant or classical conditioning. And it can vary between individuals. For instance, for one person, starting the lawn mower on the first try is reward in itself. Someone might really love the sound of the lawn mower or the smell. So that consequence of pulling the starter string is operant conditioning. For me, I don't particularly care about the very sound itself. It has no inherent value to me. But I do like what it represents....that I get to have my lawn mowed and it will look great. The nice, cut lawn is my reward. So, the sound of the engine starting on the first try is classical conditioning. I pull the string. It starts. Yeah, That means I get to mow my lawn. For my son, it could be a conditioned punisher. He really didn't want to mow the lawn in the first place. But now he has no excuse. Everything is either a good or bad consequence and it is either a direct consequence or a representation of a consequence.

Think about some of the little things you do during the day, whether socially relevant or not.... the tiniest movements or actions. Is there anything that has no result? If you can answer "yes," then you can say that not everything is operant or classical condtioning.

LOL Carrie! It's all your fault that I ended up studying the sciences in the first place! I used to just go by instinct and intuition.

Think of social learning as an information exchange in order for one being to learn from another being's (original) operant conditioning experience.

If I were to step into the Skinner box with the subject, communicate or demonstrate knowledge that the other subject does not possess, the experiment changes by adding a variable = Pre-existing knowledge passed from one being to another through observation or some other information exchange.

Higher social animals take this to a relatively complex level. Humans, with their verbal and written language skills, are very good at passing information to another human, and thus the second humand doesn't have to do experience reward or punishment personally, in order to learn from it.

But even ants can use social learning by laying down a scent trail for others to follow once the first ant has found a food source. And bees do a little dance for their sisters, to show the direction to pollen. So, with these two insects alone, by passing information to their colonies each individual doesn't have to go through the trails and errors of personally experienced OC to learn where the food is.

Right. I understand what you mean by variables. They're inescapable in just about everything. And yes, information can be had from one individual from another. My father made me study Greek history when I was in high school since the school was lousy. That I learned from a book and questions he'd give me. I didn't have to go to Greece or go back in time and experience the things I read about. So, can you give us a few examples of where this applies to dogs with humans? Dogs don't learn from imitation of humans very well. They do from other dogs though. And in these examples, can you show how operant or classical conditioning is not applicable or is not being used by the subject doing the imitating or observing of the subject who has already learned the thing from operant/classical conditioining? Or where a consequence or result of the action of the imitator is not good or bad? And therefore having nothing to do with repeating or not repeating the action or response in the future? (usually an indication of having learned something or having modified behavior)

I'll give you a weird example: One day my horse would not step over a log when I was out riding in the woods. It was rather high....above her knee. She simply would not. She'd try. She'd move her leg until it bumped the log but would not step over it. So, I got off and told her to watch me. I stepped over the log with quite a theatrical flourish. And then told her, "See? You can do it too." And she did! I was sure she was learning by observation and imitation of my leg going up and over the log. It could be that. Or it could be that I was ahead of her....out at the ends of the reins and she was simply doing what she always did.....follow me. Was she analyzing how I stepped over the log or was she simply following me and already knew how to step over the log but was insecure and needed me to go first? LOL.

With Ares, it was only a few times. maybe a half dozen.

The issue with Ares was that he forged. He wavered a bit between one step ahead of me and four steps ahead of me. I needed him precisely at heel position. One step ahead of me just wasn't good enough. He wasn't losing focus, he simply didn't get that the one step ahead of me made a difference, and the clicker was the only way to give him a "keep on going" signal, because of the timing that was necessary. Initially I clicked and treated. Then I clicked, took one more step and then treated. Then I clicked, took two steps and treated. When we got to 5 steps before the treat, I didn't need the click anymore.

It's entirely possible that in those 5 steps he was merely waiting for the treat. I saw it as clicking and not treating, but he might have seen it as hearing a click and then waiting for a treat to land in his mouth. All I know for a fact though is that it worked.

All the science stuff makes my head hurt. If connecting with the dog, reading the dog from moment to moment, adapting myself accordingly as I go, keeping my training fluid and dynamic to the point where it's more like a dance is "sloppy guess work", then so be it. But I have multiple dogs whose performance indicates that indicates that the sloppy guess work is effective.

Training dogs is part science and part art. I am better at the art end of it.

I'd say both.

I think OC and CC are almost always present. I don't know of any experiments which have removed either of these two. I do see examples of trainers removing SL from an experiment by covering their eyes. You can also remove SL by isolation.

A NRM could be an of example SL. But again, OC and CC will still be present to some degree.

Remember when we took the dogs to the beach? By walking with each other, Lyric learned through our demonstrated behavior, that we had joined up and Tia was okay. That's why the parallel walk works so well...most of the time.

Me too, and I agree.

Yet, in order to understand the "why", I've found the sciences to be important in these discussions as it is a common language and/or point of reference, so to speak.

Yea. And I do understand some of the science stuff, but without all the big words. The art is much harder to verbalize, so the science does dominate the discussions. Maybe we need a thread to discuss the art of dog training?

Yes, I remember how it was much easier for Lyric to "behave" when he walked along with Tia and the other dogs. It helps them realize, I think that "it's all good." Same with the horses. I had one that would NOT go over a wooden bridge when I was riding alone. I had to get off and lead her. When my niece rode my other mare, and she went over, (because she was braver) then the one I was riding said, "No problemo." LOL. After a few times of that, she went over it from then on whether alone or not. It always helps dogs or horses to see that the other guy is there too and can do it. In all kinds of ways, feeding off the other guy can help. It can also work against you too though, when one is a scardy cat about something, sometimes that rubs off on the more rational one. LOL. All that is not real newsworthy though. And it still involves O/C or C/C. Something comes of it. There's a result to the behavior. Condition/response.

And I see nothing different in science from art in training. Science is an art. Art is a science. Terminology simply helps identify or define things, just like language does in general. And it enriches things.

There are those who have an analytical mind and those who have a creative mind -- and yea, there's lots of middle ground also. I fall heavily on the creative side. It makes a huge difference. Reading your posts, Carrie, makes me believe you are much more analytical.

When I train, I have no idea if I'm using OC, CC, something-else-C or whatever. I do know that if my dog's ear moves a certain way or if his eye twitches or if his left rear foot hits the ground a half second slower, I need to do something different -- and that something might be that I need to stop an play with him, or it might be that I need to repeat the exercise, or it might mean he needs a nap, or it might mean my head was tilted and my left ear was in a funny place and causing confusion.

About Silverman...He was the one telling us we were doing the clicker training wrong. So while no his show isn't clicker training U.. he shouldn't go around telling people about things he knows little about.

As for forging. Dekka used to forge horribly. I would 'doodle around and click those moments when she wasn't forging. At first that meant I would have to change direction pretty quickly and click when she caught up but before she passed. I have had several clients do this as well and it works great. LOL I know you fixed the issue but if you have another dog in the future and don't want to risk devaluing your marker there are ways to go about it.

As to training with which side of the brain.... I find many of the things I did intuatively are the same things I do now analytically. AND it takes a LOT of creativity to train positively. You need to think out side the box and find how to help the dog find what you want. So you really need both if you want to be a great trainer. I think with only one you can still be a good trainer.

oops double post

I agree Dekka. I think it takes both to be a really effective communicator with dogs. You must analyze things and be able to discern minute changes in not only your dog's behavior, but in your own. Corgipower, I suspect you are being analytical, but just aren't aware that you are...or that you don't perhaps know the details of the terminology or just exactly what is behind this or that, but you're still probably realizing at some level what you have to do, what adjustments you need to make etc. And to train well, you have to be creative and come up with lots and lots of variables or alternatives when one thing isn't quite doing the trick and take into account the individuals.

With my Doberman, he didn't train up so routinely like my Lab or GSDs in the past. He was smart and wanting to work...BUT he insisted that I make things more interesting, more varied.....couldn't take too much repitition all at one time. He couldn't handle being serious. LOL. I'm serious. He was a major goof and was perfectly willing, but he had to put his .02 in also. Like, "Okay Mom...fine, but first, listen. I know what will work better. Try this on me, will you?" So I would and viola`! I couldn't dream of squashing his enthusiasm so I'd have to find ways to incorporate his raucous abandon and extraordinary joy of life into his tasks. But that way, he was doing everything with great gutso which I love in a dog. So, I was adjusting and fine tuning all the time. He ended up being my best trained dog ever...just a super dog. But I had to become more creative than I was in the past with other dogs.

I think besides learning particular techniques and such, it helps to have lived with quite a few dogs for a lot of years. It's helpful, but not a prerequisit for doing a decent job. I think though, that it makes it easier when things come almost naturally to you. I don't really think too consciously about what my dog is looking like or even someone else's dog. I do watch, but sort of notice things or read dogs almost instantaneously, without really thinking about it too much. You hear people who are unfamiliar with dogs unable to discern whether they're play growling or serious growling. Or whether a dog is aggressive or fearful. They can't tell. Pictures in books are good and helpful. But not much can replace years of hanging around dogs. It's like jobs. You can go to school and learn a lot, but hands on...what you do at the job is what really makes a big difference.

I have a client right now with a large dog with quite a severe aggression problem. And you better believe it that I have to be able to discern minute changes in his demeanor along with knowing technically what to do as far as a plan to counter condition him....and his owners. There is art and science, creativity and techniques all working together. I do hope they'll follow through and prevent this dog from being destroyed. After all, they caused it in the first place with that show 'em who's boss attitude. People so miss the whole point of domestic dogs.