Lee Charles Kelley
05-18-2005, 09:46 AM
I wanted to post a reply to Mario's last post in the thread that has been closed (due, apparently, to my own bad manners?) I've been trying for half an hour to open a new thread but the website isn't responding. I know (or hope) that some people are interested in this topic (others got bored with it).
Anyway, here's my reply (with Mario's questions highlighed in red):
Mario: If I misrepresented what you said or thought about conditioning then this may be because I really don’t know your stand on conditioning and classical learning theory. And certainly, from a number of your posts one could have taken away the idea that you don’t think conditioning is a good training tool. If I indeed misinterpreted your posts than I gladly apologize.
To clarify my position, hopefully, once and for all, a.) there are definite weaknesses in the philosophy underlying operant conditioning techniques (as stated by not just me but by the man who invented clicker training), especially when it comes to dealing with strong instincts and drives, b.) except for teaching two or three obedience commands (the sit, the recall, and maybe the heel), food is an unnatural form of inducement/reward and is actually detrimental to the real learning process (because it increases emotional tension), and c.) all behavior comes out of a desire to reduce emotional tension and is reinforced when that tension is reduced.
One tiny, further explanation: there's nothing wrong with increasing emotional tension during training because in many cases the higher the emotional tension, the deeper the learning when the dog finds a way to reduce it.
You have seem to have latched on to certain statements of mine while ignoring others (like the fact that detection dogs are always more reliable when they’re reinforced every single time, which contradicts one of the rules of operant conditioning -- no one has addressed that yet).
Mario: And here again is one of the reasons why I am confused to your stance on conditioning. While I think your methods work because you are using conditioning, you suggest that BS people may interpret your methods being based on conditioning. Thereby you implying that conditioning is indeed not a good way of training dogs, while your method of training dogs is good (even though it is based on conditioning).
Do you at least see why I may have been confused by your statements?
Not really, though I'm somewhat baffled by yours. As I explained I don't think Learning Theory holds up across the board. That's not to say that a great deal of what I do can't be interpreted as being the result of operant conditioning. Clearly it can. I even use some of the terminology (reinforcement, counter-conditioning, successive approximation) when describing what I do. But I think the underlying principle is quite different. For instance, I've said on several occasions that looked at from the point of reducing emotional tension, there is only one of the four ways that animals supposedly learn that is close to being valid, and that is through negative reinforcement. There is no such thing as positive reinforcement or P+ or P- (although I may be overstating this; I do tend to hyperbolize at times to make a point).
Certainly there can be little doubt for anyone who has read The Dog Who Loved Too Much, or who has had clients call and say "The behaviorist wants to put my dog on drugs," that conditioning techniques fail quite often at dealing with problem behaviors.
My other objection to what I view as the o.c. craze is the prevalence of modern puppy classes, which are almost always about food and getting the pups to calm down before they "learn" anything. Dogs actually learn better when they're highly stimulated, not calmed down. Of course, that would be impossible in a puppy class! Can you imagine getting a group of five to fifteen puppies all hyped up and excited and try to teach them anything? That's why I prefer working with dogs that are at least six months old or even older before starting an obedience program. Let a puppy be a puppy!
Mario: Lee, oddly enough what you describe sounds an awful lot like the effects of strong opiates being injected into a human being. A short hit of euphoria, dulled pain which then crashes into a very uncomfortable feeling. Hmm, and this is not supposed to be a self-rewarding behavior? Seems to me like this is intensely self-rewarding, regardless of the punishment that follows the reinforcement.
Well, Mario, you're welcome to interpret it that way. I think you’d be ignoring the observable evidence in favor of what you already believe. And no, it's not like strong opiates being injected into a human being. I can only extrapolate from my own experience, but I remember very clearly what my first shot of morphine was like (when I broke my leg in a car accident). I was suddenly floating, calm, and peaceful with no desire to do anything, and quickly fell asleep. If that's what the dog who’s seeking a female in heat experiences from his sudden shot of endorphins, I don't see it in his demeanor. And remember, I said that the endorphins are there to dull physical pain, not to create euphoria.
I've also had an intense fight or flight experience where the adrenalin and endorphins came into play and it wasn't like the shot of morphine at all. I know that endorphins were released because half an hour after the incident I realized that my fist, which I’d been forced to use to get away from someone who was intent on harming me and my dog, had been injured in the fight. I hadn't felt the pain while I was in that state but felt it quite vividly afterward. This explains why physical pain doesn't deter the dog (he doesn't feel any), but it hardly qualifies as "intensely self-rewarding" the way a shot of morphine would.
Quote:
Originally Posted by Lee Charles KelleyHe's trying with all his might to reduce his emotional tension. That's how and why all behavior and learning take place, under all circumstances and in all situations, even those described by Keller Breland.
Lee, I assume you are still referring to me. In fact, I do not know that all behavior and learning take place because of conditioning. As I stated before, I don’t think that all instinctive behaviors need to be conditioned to be learned. In fact, it is obvious that this can’t be the case.
I don't understand your response. The "he" in my post was about the dog. Did you think that was referring to you? And the import of my text is that all behavior and learning take place through an attempt to reduce emotional tension; which, in my view, stops up the obvious (to me) gaps in Learning Theory. We don't need to go 'round and 'round again on the fact that instinctive behaviors aren't learned. We agree on that. The question on the table is if they're not being reinforced after the first time they appear, whether that reinforcement comes from an outside agent or happens internally, why do the behaviors continue to repeat themselves, particularly when some of them are severely punished or put the animal in danger. Your explanation, as far as I can tell, is tautological.
Mario: Lee, I do not know what made the pigs behave the way they did. Nobody here suggests that by providing an animal with food (be they rewards or just general food) leads to healthy, happy animals without behavior problems.
Good! So we agree. Using food can be detrimental to the learning process.
Mario: I can certainly teach my dog to sit by using food rewards.
Because as I said in a previous post, when you hold out a piece of food a dog will naturally either jump up to get it or when that doesn't work, he'll sit. (He may also bark, circle around, or just sigh and walk away.) So holding out a bit of food naturally stimulates a dog to sit. The behavior is "hard-wired" and fairly predictable. There's nothing inherently wrong in using food to train the sit.
Mario: However, if I chain my dog up in a small kennel, completely deprived of other stimulation, then the dog will certainly lack certain things it needs for its survival. What does that have to do with failings of conditioning?
If you've successfully conditioned your dog to sit using food and have carefully followed all o.c. protocols and the dog produces the behavior reliably for a while but then gradually chooses other behaviors over the sit, then it does have something to do with conditioning failure, obviously. That’s exactly what happened in the cases the Brelands wrote about.
Mario: That’s why I said that it is very hard to interpret Breland’s results without actually knowing the experimental setup. Look, if the pigs were treated as well as chicken are in egg production, then it is no wonder their behavior broke down.
Well, I think the Brelands described the set-up pretty well in their article.
Anyway, here's my reply (with Mario's questions highlighed in red):
Mario: If I misrepresented what you said or thought about conditioning then this may be because I really don’t know your stand on conditioning and classical learning theory. And certainly, from a number of your posts one could have taken away the idea that you don’t think conditioning is a good training tool. If I indeed misinterpreted your posts than I gladly apologize.
To clarify my position, hopefully, once and for all, a.) there are definite weaknesses in the philosophy underlying operant conditioning techniques (as stated by not just me but by the man who invented clicker training), especially when it comes to dealing with strong instincts and drives, b.) except for teaching two or three obedience commands (the sit, the recall, and maybe the heel), food is an unnatural form of inducement/reward and is actually detrimental to the real learning process (because it increases emotional tension), and c.) all behavior comes out of a desire to reduce emotional tension and is reinforced when that tension is reduced.
One tiny, further explanation: there's nothing wrong with increasing emotional tension during training because in many cases the higher the emotional tension, the deeper the learning when the dog finds a way to reduce it.
You have seem to have latched on to certain statements of mine while ignoring others (like the fact that detection dogs are always more reliable when they’re reinforced every single time, which contradicts one of the rules of operant conditioning -- no one has addressed that yet).
Mario: And here again is one of the reasons why I am confused to your stance on conditioning. While I think your methods work because you are using conditioning, you suggest that BS people may interpret your methods being based on conditioning. Thereby you implying that conditioning is indeed not a good way of training dogs, while your method of training dogs is good (even though it is based on conditioning).
Do you at least see why I may have been confused by your statements?
Not really, though I'm somewhat baffled by yours. As I explained I don't think Learning Theory holds up across the board. That's not to say that a great deal of what I do can't be interpreted as being the result of operant conditioning. Clearly it can. I even use some of the terminology (reinforcement, counter-conditioning, successive approximation) when describing what I do. But I think the underlying principle is quite different. For instance, I've said on several occasions that looked at from the point of reducing emotional tension, there is only one of the four ways that animals supposedly learn that is close to being valid, and that is through negative reinforcement. There is no such thing as positive reinforcement or P+ or P- (although I may be overstating this; I do tend to hyperbolize at times to make a point).
Certainly there can be little doubt for anyone who has read The Dog Who Loved Too Much, or who has had clients call and say "The behaviorist wants to put my dog on drugs," that conditioning techniques fail quite often at dealing with problem behaviors.
My other objection to what I view as the o.c. craze is the prevalence of modern puppy classes, which are almost always about food and getting the pups to calm down before they "learn" anything. Dogs actually learn better when they're highly stimulated, not calmed down. Of course, that would be impossible in a puppy class! Can you imagine getting a group of five to fifteen puppies all hyped up and excited and try to teach them anything? That's why I prefer working with dogs that are at least six months old or even older before starting an obedience program. Let a puppy be a puppy!
Mario: Lee, oddly enough what you describe sounds an awful lot like the effects of strong opiates being injected into a human being. A short hit of euphoria, dulled pain which then crashes into a very uncomfortable feeling. Hmm, and this is not supposed to be a self-rewarding behavior? Seems to me like this is intensely self-rewarding, regardless of the punishment that follows the reinforcement.
Well, Mario, you're welcome to interpret it that way. I think you’d be ignoring the observable evidence in favor of what you already believe. And no, it's not like strong opiates being injected into a human being. I can only extrapolate from my own experience, but I remember very clearly what my first shot of morphine was like (when I broke my leg in a car accident). I was suddenly floating, calm, and peaceful with no desire to do anything, and quickly fell asleep. If that's what the dog who’s seeking a female in heat experiences from his sudden shot of endorphins, I don't see it in his demeanor. And remember, I said that the endorphins are there to dull physical pain, not to create euphoria.
I've also had an intense fight or flight experience where the adrenalin and endorphins came into play and it wasn't like the shot of morphine at all. I know that endorphins were released because half an hour after the incident I realized that my fist, which I’d been forced to use to get away from someone who was intent on harming me and my dog, had been injured in the fight. I hadn't felt the pain while I was in that state but felt it quite vividly afterward. This explains why physical pain doesn't deter the dog (he doesn't feel any), but it hardly qualifies as "intensely self-rewarding" the way a shot of morphine would.
Quote:
Originally Posted by Lee Charles KelleyHe's trying with all his might to reduce his emotional tension. That's how and why all behavior and learning take place, under all circumstances and in all situations, even those described by Keller Breland.
Lee, I assume you are still referring to me. In fact, I do not know that all behavior and learning take place because of conditioning. As I stated before, I don’t think that all instinctive behaviors need to be conditioned to be learned. In fact, it is obvious that this can’t be the case.
I don't understand your response. The "he" in my post was about the dog. Did you think that was referring to you? And the import of my text is that all behavior and learning take place through an attempt to reduce emotional tension; which, in my view, stops up the obvious (to me) gaps in Learning Theory. We don't need to go 'round and 'round again on the fact that instinctive behaviors aren't learned. We agree on that. The question on the table is if they're not being reinforced after the first time they appear, whether that reinforcement comes from an outside agent or happens internally, why do the behaviors continue to repeat themselves, particularly when some of them are severely punished or put the animal in danger. Your explanation, as far as I can tell, is tautological.
Mario: Lee, I do not know what made the pigs behave the way they did. Nobody here suggests that by providing an animal with food (be they rewards or just general food) leads to healthy, happy animals without behavior problems.
Good! So we agree. Using food can be detrimental to the learning process.
Mario: I can certainly teach my dog to sit by using food rewards.
Because as I said in a previous post, when you hold out a piece of food a dog will naturally either jump up to get it or when that doesn't work, he'll sit. (He may also bark, circle around, or just sigh and walk away.) So holding out a bit of food naturally stimulates a dog to sit. The behavior is "hard-wired" and fairly predictable. There's nothing inherently wrong in using food to train the sit.
Mario: However, if I chain my dog up in a small kennel, completely deprived of other stimulation, then the dog will certainly lack certain things it needs for its survival. What does that have to do with failings of conditioning?
If you've successfully conditioned your dog to sit using food and have carefully followed all o.c. protocols and the dog produces the behavior reliably for a while but then gradually chooses other behaviors over the sit, then it does have something to do with conditioning failure, obviously. That’s exactly what happened in the cases the Brelands wrote about.
Mario: That’s why I said that it is very hard to interpret Breland’s results without actually knowing the experimental setup. Look, if the pigs were treated as well as chicken are in egg production, then it is no wonder their behavior broke down.
Well, I think the Brelands described the set-up pretty well in their article.