Short-comings of Operant Conditioning?

 I think most animals can be clicker trained.  Fish, chickens, horses, cats, dogs, dolphins, pigs, donkeys, mules....

I certainly don't use a clicker for everything; why would I?  But for me, it's not a case of it doesn't WORK that well.... more a case of, "why use a sledgehammer to crack a nut?"  As you say, if it's a complex behaviour or something that you reeeeally need to stick, that's when a clicker comes into it's own.  The rest of the time, much less specific methods are perfectly adequate.

I am curious as to how much time was spent charging the clicker (with both species).  Did you play a box game or something similar so they could get the hang of "how it works"?  Also, how many treats are given and how many seconds between rewards.... 

And I am also wondering if there is not something Kit would work for better than "a food treat he likes".  It may be that the thing he wants MOST out of life is, mostly, to be left alone.... for folks to stay out of his "space".  This makes him a poor candidate for clicker training, because you would have to, in effect "punish him" before you could reward, every time.  It's also an awkward reward to deliver, and timing is everything. 

It sounds like, with Kit, you have been "capturing" as opposed to "shaping" and IMO you should have been "jackpotting" him on those isolated occasions when offers the behaviour.  "Shaping" is where you start very very small.  You wouldn't ping when he offered the behaviour; you would ping when he offered the teeniest chunk of that behaviour.... and keep pinging, gradually upping the criterai. 

This maybe why there has not been much increase in that behaviour and why Kivi is still appearing a little mystified.  To me, the latter is a sign something has been missed or done incorrectly, because a hallmark of marker training IME is that suddenly the animal appears completely UN-mystified - a light bulb goes on and it's like: "oh THAT'S what you want me to do! EASY!  Why didn't you say so before??!!"  And then they start "throwing" the behaviour at you - at which point you put it on cue.

This is offered respectfully,  and no offense is intended.  It is not the technology, it is the ability of the individual applying it.  Chuffy makes a great point about the communication aspect.  To many times folks get frustrated with the technology because they are missing a vital element.  It may be the antecedent is not clear and concise, it may be that the reinforcer is not appropriate.  I find that to be an issue many, many times.  It is not a reinforcer if it is not changing the behavior.  In this case, your dog is more subtle than most.  You will have to get creative to determine what does the dog self select to do and how can you put conditions on that.  If you can set up training opportunities that involve access to desired things or activity, you may be able to establish the chain of events to get things moving.  It may be like inertia.  You have to over come the resistance before you have a chance of actually attacking the behavior you wish to target.

Get another pair of eyes or a video camera.  Set up times when the dog is with you and when the dog is alone.  Study those tapes or discuss with your observer over and over again.  Make sure you are looking for subtle cues.  You may find that key you are looking for.  Another consideration,  wait time;  in many many cases we humans do not allow a creature of any pursuasion to "think" about it.  You can just wait for behavior to occur or you can engineer the situation so the choices open to the dog are limited. 

Try and focus on the behavior you are getting focusing your observations on what occurred immediately before and after.  That is a good place to concentrate your efforts.  I offer these comments because this is what I see not only in my obedience classes, but daily in my job with people.

corvus

So I guess part of my gripe is that whole "whatever the animal will work for" cop-out that hasn't helped me much with an animal that doesn't hold with "working for" anything. Where is operant conditioning when you have an animal that won't work for rewards, won't stand punishments, and doesn't care if you withold something good

I could be unfair and turn it around on you.You're the one who professionally, or for collegiate purposes, observes animals, specifically birds. Why do animals do what they do? How similar is a dog to a hare? Should we train a dog in the same way we act with a hare? OC is not a philosophy or political party, it is a description of what is happening. And just because an animal is not reacting to the stimuli you are cognizant of presenting does not mean that they are not reacting to stimuli. That would be like saying that since something is happening behind me where I cannot see it, it's not happening at all.

Then again, do different animals have a different general psychology? For dogs, would it be to investigate, leading with the nose and for hares wouldn't it be to get to safety first and wouldn't those two motivations be drastically diferent and wouldn't that affect how we want to achieve a control over their actions?

What if the click were frightening to a hare but not to a dog and it simply had to do with both hearing response and a general method of how to deal with the world. A dog, when in doubt, sniff, a hare, when in doubt, hide. Should we judge the clicker to be a failure with the dog or other animals because it was a failure with on hare?

Is the notion that a animal does what works a "cop-out"? What is the final analysis of an animal that doesn't do what you want? Is that proof that the maxim is a cop-out or have we not investigated fully?

Hi Corvus,

Maybe since you are very conscious in your use of classical conditioning and OC, you can best describe and distinquish the two as you attempted to apply the two.  I see a lot of overlap and how a dog learns depends on the dog temperament, natural drives, and the dog's perception of what is going on.  I often think that with Clicker Training its dependent on changing the natural drive of the dog, somehow changing the want to be more intense where the original want made the dog balanced and well behaved.  E.g. changing the emotional response from the stimulus from satisfying a basic need as hunger to the response of creating the emotional response of greed.  If you can expand on the definition by their practical use, that would help the discussion be balanced.

"Classical conditioning is what happens when an animal learns associations among things. Learning associations means learning that things go together. When one thing happens (you burn your dinner), another thing will follow shortly (the smoke detector goes off). You say "Let's go for a walk", your dog gets all excited because he has learned that this particular phrase precedes going out for a walk. Your cat runs to its food bowl when he hears the can opener because this often signals feeding time. In each case, there is a predictable relationship among the events and the animal learns to respond to the first event in anticipation of the second event. Your dog learns that a treat comes after the sound of the clicker.

Operant conditioning is a set of principals that describe how an animal learns to survive in its environment through reinforcement (consequences). This is learning in which behaviors are altered by the consequences that follow them. If your dog "sits" and you give him a treat, he'll be more likely to repeat the behavior "sit". On the other hand, if the dog "sits" and is knocked in the head, it will be less likely to repeat the behavior. These responses were operantly conditioned. B.F. "Fred" Skinner first coined the term in 1938 in this book The Behavior of Organisms. "

 I think I see what Ron's saying.  Saying "shortcomings of OC" doesn't really make sense, because it's not a method as much as it is a description of what is going on... WHY some behaviours INCREASE and why some decrease.    I though what you meant was "shortcomings of marker training" and replied as such.  Sorry if I misunderstood.

I agree that classical conditioning is often discounted in favor of looking at things from a strictly operant conditioning belief system. The same can be said regarding social learning, which is often discounted as having little value in the face of well-used OC.

The problem (IMO) seems to be coming from a kind of "OC brainlock" of the absolutists.

Everything I've observed in dogs (and a lot of other social animals), tells me classical conditioning, social learning, and operant conditioning are all affecting an animal's behavior and choices (to varying degrees), at the same time.

Unless you're dealing with a subject which is isolated (in a box) from all social contact and information, all three forms of learning, as well as intelligent choice, will be present.

One is not "better" than the others. One cannot claim to define everything on it's own. It simply functions in it's appropriate time and place - in harmony with the other two.

Angelique

Everything I've observed in dogs (and a lot of other social animals), tells me classical conditioning, social learning, and operant conditioning are all affecting an animal's behavior and choices (to varying degrees), at the same time.

And that's a good point. For example, I trained Shadow to walk with mushing commands without using treats or corrections. I would tug in the direction (in harness) and say the command. As DPU described it, an eventual association of the command and the action. And I would suppose that it would matter what effect is desired as to what method is used. In traditional sled dog training, a new dog is placed in an established team and they literally learn the ropes (or ganglines) as they go along. Nothing fancy and it takes a while but then any worthwhile training will take a while.

I think, however, Corvus might be more interested in a natural laboratory sort of study. What does a dog do to placate humans without a lot of human intervention or direction? Plus, her training demands are not the same as some of ours. She's not doing agility, Schutzhund, SAR, therapy, or CGC work. Her animals are simply coexisting with her and all she asks is that they give her space when she wants it. A clicker would indicate a direct human intervention. But, dogs are always learning from their environment and we are part of their environment, so, we are training, even when we don't think we are. You get what you reinforce. That says a lot, because you don't always realize that you are reinforcing something. Not that everything a dog does is a head-long rush to pleasure or a fear-filled flight from punishment. But they are creatures of habit.

Maybe it would be better if there was a clarification of the issues or direction of this investigation.

I put a question mark at the end because I couldn't work out how to summarise what I wanted to say in one sentence.

Kivi has his "ah-ha!" moment, he just often repeats behaviour in a way that suggests to me he doesn't really understand what he's doing. It's my fault for not being as subtle as he is, but that's sort of what I'm getting at. I'm finding I'm communicating better without the clicker in most cases. In fact, sometimes I think the clicker confuses the situation, precisely because I'm better at communicating without it than with it so that every time we work I inevitably click something I didn't mean to click or not click something I realise half a second later I should have clicked. I suppose what I should have said is that I was experiencing unexpected short-comings with operant conditioning based on the kind of person I inherently am. I have great timing, that's not the problem at all. I just click sometimes when I think he's about to do what I want him to and then it turns out that he wasn't after all or the click actually distracts him from following through, and sometimes I don't click because I was holding out for something marginally different or longer and realise half a second later that now he's confused because he really thought that was it and now he thinks he has to try something different.

I have played the box game with KT and done fun free-shaping to get him thinking. I think he does better without my direction with clicker training. But I'm still the idiot that clicks.

With Kit I've been feeding him little bits of dried strawberry and that's all I've been doing. Some days he greets me at the door of his enclosure waiting for the strawberries and some days he takes off and hides and won't come out. Chuffy, you're quite right in that what he would work for is to be left alone and that makes it kinda impossible to work with him using that. I haven't been quite truthful because he will also work for the sake of communication if I'm not mistaken, which is its own reward with or without the clicker or the treats. He works to make his desires known to me so I can fulfill them. As far as he's concerned, the ping means I'm giving him a piece of strawberry, but more interesting to him is that strawberries are a new favourite and if he wants them he has to somehow tell me that he does and he already knows how to do that his way, so why would he try anything else? And that's where I get stuck. He gives me so little to work with in the first place that I am having trouble giving him a reason to try anything else. He knows I have strawberries and he knows he can get them by coming out and looking hopeful regardless of pings. I try to leave him wanting and not come pinging very often to keep him interested, but he doesn't want food in general enough to work for it with any kind of predictability.

So with him, in the end I am in the same place I'm in with every other animal in my house. That is, we're trying to make ourselves understood to each other and the most effective way of doing that is finding a way that seems to communicate what it is we want. Clickers are generally considered the bridge we need here, but as I've said, I tend to either react too quickly or make a bad judgement call, which never happens if I'm just classical conditioning because it's so passive. I'm a little voice in their ear saying "A=B".

Anyway, I realise that there's slightly more to it because communicating a need and having it met is operant conditioning, really. I'm not entirely sure what my issue is. I just wanted to start something and I figured you guys would get my mind into gear and force me to find words for what I am finding. 

 I think what I'm trying to get at is I think there's actually a fair bit of motivation in an animal just to communicate. When Kivi first discovered 2 days after he came home with us that he could ask for a treat by sitting on his butt, I could see the excitement and wonder in his eyes. It was not getting the treat that was making him so excited, it was realising that he could actually talk to us. He was sitting whenever he wanted anything from us no matter what it was. The more tools we gave him to communicate the happier he was and the more attached to us he became. Now he is 6 months old and of course he is finding ways to tell us what he DOESN'T want.

Anyway, I have noticed over the years with Kit that when he wants something, he finds ways to make that known to me. He uses exaggerrated gestures and his body language becomes very simple. He's talking baby talk to me to make me understand. In the end, I sat back and let him find ways to make himself understood because he was clearly so much better at it than I was. Now I look at my dogs and even my rabbit, who is especially good at expressing herself, and I think, we all just want to be able to talk to one another and make ourselves understood. It's hard enough as it is, so why add another sound to it all? Of course, I know the answer to that question because I have added another sound and am glad I did, but for the most part, I find I am talking to the animals and they are talking to me and our reward is when one of us makes ourselves understood and for a moment there is no huge gorge between us and we are very close. That is not just rewarding to the one that gets what they want, I think.

Anyway, I often feel that when I am rewarding behaviour or punishing behaviour that I'm bridging that gorge in little rickety zig-zag lines on a bridge that may collapse at any moment, whereas when I make myself understood with nothing but my desire and my animal's desire to understand one another, that gorge isn't there at all.  

Just as an example, we were using some great treats to train Kivi's emergency recall. It was very nearly conditional when he abruptly decided he didn't like those treats anymore and within a week it's all fallen apart and we have to rebuild with the major setback of him having discovered it's sometimes worth declining to come to the emergency recall. However after I got back from this disastrous trip to the dog park that resulted in our rickety bridge partially collapsing, my little fur monster pushes himself into my arms, tucks his head under my chin, and presses against me whining. I instantly forget how blue I was feeling and for a few moments there is no gorge and no need for rickety bridges that collapse with little warning. If only I could make that gorge disappear every time I wanted to teach something.

So I am beginning to think that my beef with OC is that it doesn't feel "real". In the end, clickers, treats, ignoring behaviour, punishments... they don't make my gorge disappear. Classical conditioning does.

Incidentally, the hare doesn't mind the ping of my clicker. This evening his ears were rotating every which way and he was cheerfully accepting head strokes between strawberries and pings. He wouldn't be that relaxed if he didn't like the noise. Although you're right about seeking safety if in doubt. 

corvus

 I think what I'm trying to get at is I think there's actually a fair bit of motivation in an animal just to communicate. When Kivi first discovered 2 days after he came home with us that he could ask for a treat by sitting on his butt, I could see the excitement and wonder in his eyes. It was not getting the treat that was making him so excited, it was realising that he could actually talk to us.

This is the great thing about NILIF - people think NILIF is all about controlling the dog and bending him to your will.... it IS good for encouraging the good habit of "checking in" and listening and generally co-operating to get what he wants, but it ALSO EMPOWERS THE DOG enormously, which is rewarding in and of itself IMO.  It's a great confidence booster for insecure dogs I think.

corvus

It's hard enough as it is, so why add another sound to it all?

Where is houndlove.... Sighhhh... I recall once houndlove described living with a group of other people and none of them coud communicate using their first language because they were all from different places and didn't all know those languages.  So they managed by using a second language they all knew... it wasn't "native" for any of them, but they were all pretty fluent in it.  Marker training is like your second language and Kit's second language and Kivi's second language.... Sure if it doesn't feel comfortable to you, don't continue with it.  It's just another tool to try. 

But whether you LIKE it or not and whether it feels "real" or not, OC is happening all the time, in how you respond to your environment.  If something results in a pleasant outcome, like a "thank you" or a delicious meal, or a hug, or a sense of satisfaction - you will do that thing again.  If something results in an unpleasant outcome, you will work to avoid doing it again.  You can accept it and use it - or not, as you choose.  Neither stops it from happening.

Oh, I think we will continue with the marker training, but whether I like it or not was not really my point. My point is, there are a lot of ways to bridge that species gap. I think it interesting that clickers are touted as the best way to bridge that gap and I believed that when I started, but now I don't. The idea of the clicker is to simplify the message, as I understand it. But then, all of my animals are in most cases capable of interpreting the much richer and more complicated message of my voice and body language once they have a bit of time with me.  So the clicker was fantastic when we first started out and it's great when I'm trying to introduce something a bit complicated or something hard to capture normally, but I suspect that Kivi at least is looking for more direction during clicker training. It's kind of like taking a step backwards sometimes. Like Kit, it seems to me Kivi is trying to bypass the click sometimes and trying to engage directly with me.

Is it possible that things like marker training, especially with clickers and the likes, can actually confuse the matter for the animals rather than make it clearer? That's the kind of short-coming I'm wondering if I have. I don't know if this is just me and my animals or something more common.  

I think dogs can respond to classical conditioning, too. Scenario, you recall, the dog ignores you because either you don't give treats or punishments, or they have a stronger interest to go after than anything you have and so they become classically conditioned to ignore you. As in, life offers the chance for them to decide for themselves what they wil do. Which lasts until they run out on to the highway and a lorrie doing about 88 kph ends their independent career.

They can also classically condition to recall to you because it is rewarding to them to do so. Classical conditioning is most remembered as the example of Pavlov's dog, salivating when the bell was rung because it was always rung at mealtime. This is also known as associative learning. If a, then b. Operant conditioning might be the way they learn, but classical conditioning is how it becomes a habit. Operant, imo, simply means that the dog chose the behavior because of the reward associated with it. Done enough times, it becomes a habit, ergo, classical, at least to some extent. But the process of avoiding pain and seeking pleasure is below the cognitive level and is a survival trait, imo. So, whether you use a clicker or not does not stop the process of operant conditioning. The dog is still learning via pleasure and pain. If you can't use the clicker, fine. But you are still marking, whether you realize it or not. A certain expression of your face, a relaxing of tension, even the accidental "good boy", straightening up from a stooped over posture. You are marking somehow. Otherwise, how would the dog know that "sit" is what you wanted? Dogs can read our body language but I don't think they can read our minds, though there may be someone who can come up with theories to suggest that very thing.

Well I don't think a conscious marking of behaviour is quite the same thing as unconscious. It is for the dog, but what I'm thinking is if you take everything below the level of consciousness perhaps you're actually marking more clearly. Like in my case where the conscious marker is vulnerable to poor judgement.

I'm not arguing about whether OC occurs or not. I'm talking about whether consciously using it has short-comings, whether they are as a result of the method itself or the human. Or the animal you're trying to train. 

corvus

but what I'm thinking is if you take everything below the level of consciousness perhaps you're actually marking more clearly. Like in my case where the conscious marker is vulnerable to poor judgement.

You stated that better than I did.