Investigating Learning Theory Further: Beyond P,R,+, and -

Many of those terms are new to me, so I am taking the liberty of searching for definitions.  Please correct me if my sources don't get it right!

Premack Principle:
The Premack Principle, often called "grandma's rule," states that a high frequency activity can be used to reinforce low frequency behavior. Access to the preferred activity is contingent on completing the low-frequency behavior

I wonder if NILIF works this way?

Hull's Drive Reduction Theory:http://tip.psychology.org/hull.html
Hull developed a version of behaviorism in which the stimulus (S) affects the organism (O) and the resulting response (R) depends upon characteristics of both O and S. In other words, Hull was interested in studying intervening variables that affected behavior such as initial drive, incentives, inhibitors, and prior training (habit strength). Like other forms of behavior theory, reinforcement is the primary factor that determines learning. However, in Hull's theory, drive reduction or need satisfaction plays a much more important role in behavior than in other frameworks 

Behavior Systems Theory: http://www.psychology.sbc.edu/harris_legacy.html
Moving away from Learning-Process model, theorists have proposed the Behavior Systems approach. According to the behavior-system’s approach, the full repertoire of behaviors that an animal expresses naturally can be categorized into different categories or systems. For example, there is the category of food-getting behavior. Behaviors that fall into this system relate to general and focal searches for food, and all behaviors that the animal uses once food is within its reach. The Behavior Systems theory believes that certain behaviors can be conditioned more readily with stimuli that relate to that particular system. Garcia and Koelling (1966) were the first experimenters to have empirical support for this notion. In their experiment, two stimuli were paired with two aversive stimuli in a between design. The taste of flavored water was paired with either nausea or shock; a tone was paired with either nausea or shock. Garcia and Koelling found that selective associations were formed. Taste could be associated with nausea but not shock and the tone could be associated with shock but not nausea. In the rat’s natural environment, the taste of a particular food is not going to lead to shock, but taste of poison is a good predictor of nausea.

Response Deprivation Hypothesis:
A hypothesis that states when access to one activity is restricted to below baseline levels, the person will engage in the targeted activity at a level exceeding baseline rates in order to gain access to the deprived activity. Restricting access to below baseline levels, then, serves as an establishing operation.

Dog_ma

According to the behavior-system’s approach, the full repertoire of behaviors that an animal expresses naturally can be categorized into different categories or systems

That reminds me of the work by Lorenz in animal behavior and learning. A real simple example that I can think of is the fact that dogs mark with urine. This is something that they don't seem to learn in any response to the environment but seems to be a "hard-wired", so to speak, part of being a dog. The effect, of course, is marking territory not only to advise others but also as landmarks for themselves "ah, this is the way back to the den." Such behavior promotes survival which would explain it's value as a behavior.

I also wanted to throw in the non-linear dog theory. Even with it's imperfections, I think it had some useful insight to dog-dog behavior.

The Premack Principle, often called "grandma's rule," states that a high frequency activity can be used to reinforce low frequency behavior. Access to the preferred activity is contingent on completing the low-frequency behavior.

This is one reason we say that food doesn't always need to be the reinforcer for desired behavior.  Let's say you have a big, goofy Lab who just loooooves swimming so much he could be a dock dog (preferred activity).  He isn't terribly food motivated.  You could do your obedience exercises (low frequency behaviors) next to the lake, and his reward would be getting to go swimming! 

Kim_MacMillan

stimulus a good/better reinforcer (including belongingess, species-specificity, temporal/situational constraints, etc)

I still think the dog in each individual case defines it.

I've heard of classical conditioning being separate from learning theory or OC. I understand that OC, as often described, is about an actual learning process.

So, start at the beginning of that. In associating a marker with reward, at first the marker has no meaning. The dog hears the sound whether it is a click or a blast from bagpipes and a treat is delivered. The first instance may have no connection. What about the second or third or whichever instance the dog has identified the marker as a predictor of reward? Is that not some kind of learning process or response to the environment? Sure it uses the principle of reward but isn't the increased attention or anticipation of that sound a sign of a learning process?

Another thought I had, which may or may not appeal to the spiritual aspect of our debates or even our relationship with our dogs.

As can be pointed out here and even in some papers I have read, the dog is not, as it were, an automaton, "blindly" responding to whatever, even though OC and, imo, it's cousin, classical conditioning happen every day. To the point, do dogs have a soul? They certainly have personality, which lends to suppose a soul, I suppose. In so doing, we acknowledge that they have a sovereignty. A freedom of will, if you will. If that is the case, then they do what we ask because they like us, they really, really like us.

Developementally, they are often likened to a child, though I think some are fairly smart and would think that description could be misleading in the wrong context.

And I can't help but think that opinions on this will do more to reveal opinions about ourselves. If the dog possesses an independent, sovereign soul and has, to an extent, free will, then at best, we can barter. Which leads to what is commonly called OC, a description of how an animal learns and how what they learn is retained or reinforced.

If we assume that we know what they should be doing and we institute a stimulus to stop something, OC-wise, a punishment, then we assume to be that which stops them from doing what they were "driven" to do. And for the punishment to work requires accepting that OC is valid. Actually, it works that way, whether one accepts the precepts of OC or not. And the dog, with the soul, would have to accept this limitation. What if the soul of the dog did not care to accept the correction. They have sovereignty, right? Or, no?

Or, the dog has a soul but is not as able as humans to disregard the mechanics of OC.

ETA:

I know I keep mentioning OC but if we are going to go beyond it, it seemed like a handy vantage point from which to go forward, if that is possible.

 I'd be doing a PhD on animal behaviour right now if I hadn't decided to get a job that actually pays for a while, so I'm up for any discussion. I would really love to discuss classical conditioning, though. I think this has been sadly neglected in our discussions and I think we probably use it more than we really talk about. The majority of what my dog knows came from classical conditioning, and I've enjoyed it for how easy I've found it to be, even for my rabbit and hare.

I also wanted to throw in the non-linear dog theory. Even with it's imperfections, I think it had some useful insight to dog-dog behavior.

what were some of the imperfections you saw?

what were some of the useful insights you saw?

i had fired off an email to the author because there was an option to do so, with comments.... and i had read somewhere in that blurb that they are working on simplification of terminology used for the everyday joe and jane sixpack to be able to understand.

To the point, do dogs have a soul?

I know I keep mentioning OC but if we are going to go beyond it, it seemed like a handy vantage point from which to go forward, if that is possible.

We actually had a thread before just on the non-linear dog theory but I can't find it right now.

lostcoyote

what were some of the useful insights you saw

Though a bit ragged at first, I appreciated the author's use of topology (set theory on steroids) which she actually borrowed, I believe, from a mathematical model for biological evolution. That dogs have "fitness hills" that represent points of security or stability. And they do what it takes to maintain or remain on these fitness hills. That sometimes, especially in introduction to a stranger dog, there will be a scuffle as each dog tests the limits of the other dog to see how far they will carry violence. As each dog offers an appeasement signal, they establish an equilibrium, of sorts, experiencing security in the notion that the other dog will not escalate past a certain point. Weakness in the theory is failing to account for such breeds as Akita. Akitas are notoriously one-owner dogs. They don't form packs or associations as most dogs do. Once they don't like another dog, the not-liking continues forever, amen. And they don't fight for "dominance" or an equilibrium, they fight until the opponent is no longer breathing.

Another weakness is that the subjects were primarily dogs in her home or dog seen at the dog park she went to. I think many factors and a few antecedents were not accounted for.

An interesting part of the theory is that, for example, one dog desires ownership of a stick. Another dog concedes the stick because playing chase is more valuable. She doesn't give much thought to two dogs who truly desire the stick at the same time.

Also, she is correct in pointing out that the early study of wolves were flawed, their use as a model for dogs is grossly inaccurate, that the primary authors of those wolf studies were men, with prejudices and a weltanschauung all their own, as a product of upbring, the state of the world, etc. Then she goes as far as to suggest that men, themselves, are the problem with most theories and that our perceptions are clouded by testosterone, so to speak. There could be some credence for that.

Also, I disagree with the notion of just letting dogs of various unknown backgrounds "just work it out." Dogs do not exist in a vacuum, they exist alongside Man.

I'll try again to find that previous thread.

Or, better yet, I have the study linked in my favorites.

Many things are of value in the study, though.

The fact that a true leader dog rarely actually fights. Most fighting is amongst the sub-ordinates. And that most fighting is not a fight for dominance but a struggle for elquilibrium. The introduction of a new dog at a specific time in the emotional space of the dog at that time creates a perturbation in the fitness hill of peace that it had. The dog will interact to determine that it's fitness hill remains undisturbed. Dogs set up a language with each other. For example, A Sibe, which carries his tail curled over his back may find other dogs reacting to him until they realize that he is not challenging, it's just how his tail hangs. Once they have developed a colloquial lingo amongst themselves, everyone rests on their adjusted fitness hill, or mound of stability.

And yes, I agree, the language could be re-written for laymen a little better. Most people have not studied topology. You should see the Lorentzian group in hyperbolic trigonometry as applied to topology. That will put hair on your chest.

Anyway, as it appears that most dogs do not act out of a need for "dominance" leads one to question any dominance theory and it's usefulness. The main gist of the theory is that dogs do what they do to maintain stability, which makes sense since they are social animals and one must have an equilibrium to maintain a social order.

And, as far as disregarding or moving "beyond" scientific analysis, I cannot. Call it a failing of mine. I see scientific analysis as a common language by which to describe or state something. Anything else is a subjective inner monologue that may only have value to the speaker. From there, we would return to the neverending battle of semantics, amphigory, and the odd non-sequitorial metaphors which do nothing, imo, to reach a conclusion.

I cannot define a soul. Many here describe themselves as having a spiritual relationship with the dogs. Allow me to stick my neck out and say that you can only have a relationship with another soul. I, for example, cannot imagine having a relationship with a rock.

corvus

 I'd be doing a PhD on animal behaviour right now if I hadn't decided to get a job that actually pays for a while, so I'm up for any discussion. I would really love to discuss classical conditioning, though. I think this has been sadly neglected in our discussions and I think we probably use it more than we really talk about. The majority of what my dog knows came from classical conditioning, and I've enjoyed it for how easy I've found it to be, even for my rabbit and hare.

I like that idea, too.  I'm especially interested in the application of classical conditioning with aggressive behavior, a la Jean Donaldson's suggestion about feeding the dog even while it is growling at a scary looking human to change the dog's basic perceptions about scary humans.  

spiritdogs

I'm especially interested in the application of classical conditioning with aggressive behavior, a la Jean Donaldson's suggestion about feeding the dog even while it is growling at a scary looking human to change the dog's basic perceptions about scary humans. 

When I give Shadow treats for remaining calm in public in the face of a reactive dog or distracting scene, I am not just rewarding what I value as "good" behavior. Out in the wild, what I value as good behavior might get him killed. In the wild, he might be safer returning challenges and scaring off would be attackers. But my dog is a dog, symbiotic to man, and has to be safe in our society. Therefore I reward the behavior that is calm, even if it is anti-survival for another species. Anyway, the secondary effect that I am looking for is classical conditioning wherein these formerly upsetting scenarios instead become good things. Being in public and calm is good. Meeting other dogs and being calm is good, odd. scared dogs aside.

I wish I knew the magic number of time when a learned behavior becomes classical, or how many repetitions of an event that is not designed as an OC event become classical. For some dogs, it might the first time on something. So, for example, if the first time you ever get a dog and start on a walk, and define the rules for that walk as being near the heel position, then that is all the dog will ever know. By never allowing pulling ahead from day one and being committed to altering any behavior not conducive to LLW. Some of it will be primarily classically conditioned because those are the universal rules of your walk. Other times, using a specific OC event, change a behavior to one more conducive to LLW. But all roads point to LLW from Day One.

hiya ron, thanks for your sumary on that paper. i read somewhere that dogs do not form true packs (we often use that term); in a true pack such as wolves, the alpha shows dominant postures, marks his territory with confidence, walks with stiffe legs and high moderately wagging tail, sniffs nose to nose first and the nose to genitals, etc. the sumbissive dogs tag along in contended fashion with more subdued behviors while the submissives perform ritual passive behaviots such as groveling with the head and tail down, crawling on their bellies and trying to lick the lips of the dominant.

pack behavior in dogs is less complex but, because rules can't be followed in a more exacting fashion comparatively speaking, the behavior is less predictable. frew dogs ever gain the ability to form a true and wild pack. only dogs that have reverted to the wild do such as dingos.

free ranging dogs such as ones found in urban areas don't seem to form packs with dominance heirarchies; most of them are lonesom travelers although two neighboring dogs will often range together. the only time that dogs congregate out in the open range is when a female is in heat.

i think that in a truer sense of the word, pack implies a degree of mutual cohesion that is generated by virtue of the need to defend a food source, territory, and themselves against other predators but by virtue of ssociation, we, to this day, simply call a group of dogs in our own household, a pack largely because that is what we've been taught and have accepted as what constitutes a pack. if i were to let all four of my dogs roam free, i'm not so sure they would all remain together as a cohesive unit, barring them being picked up by animal control folks

as far as what s relationship is, what you've described is open to interpretation and definition as well..... one could argue that one has a relationship with a rock if there is special meaning behind a rock on a fireplace mantle... although i suspect you place human psychological attributes onto that term - such as a sexual relation, or a friendship, etc.... our message exchange herein may be considered in relation to one another, as could my feelings of value i might place on a gemstone.

lostcoyote

our message exchange herein may be considered in relation to one another, as could my feelings of value i might place on a gemstone.

During the 70's, there was a craze in having pet rocks. They are easy train. Don't each much. Non-existant medical costs.

We could have an endless semantic discussion on relationship. I would say that the relationship with the rock is not the same as a relationship with the dog because the rock does not consciously interact with us. The relationship with the rock is solely what our memories and associations are with that rock. My wedding ban is a special cast 5mm band of 14k white gold. It does not interact with me, it is a symbol of vows, love, my wife, whatever I choose those associations to be. The ring is not sovereign and doesn't exhibit learning ability or behavior systems or any other characteristic that can be related to a living organism.

But our exchange brings to point what I am saying about the language of science. Many times it has been stated by those who reject the use or advice from some of us on positive reward training methods that science does not speak to the spiritual aspect of our relationship with our dogs. Some claim that they have a relationship or spiritual connection to their dogs that is outside of science. A wholly personal meaning, perhaps even having it's own internal language not easily suited to spoken language. And I don't see how we can further the understanding of learning theory (the quetion in the title of the op) when we are just now trying to define terms. I posit that a relationship is possible only with another sentient living organism and your response is that it is also possible to have a relationship with a rock. Even though I think differently, I'm not saying your wrong, either. But we've not got beyond semantics, either.

The language of science, imo, does not lessen whatever spiritual or relationship value we place on the dog, but it can describe various actions and reactions and principles that, to my knowledge, have yet to be disproven. In established terms that can be agreed upon and actually have been agreed upon by people have made a career for decades out of this very thing. We can banter in semantics all day and maybe get some enjoyment out of that but it doesn't disprove the science or offer a plausible alternate theory in evidence that can be agreed upon. Even the non-linear dog theory is largely unproven, though some of the mathematical model in it has some relevance. And yes, that could sidewind into semantics. I could say that an apple falls to Earth. It is more accurate, though, according to Einstein for a time not a full explanation, to say that the earth and the apple exert gravitational influence on each other. We could delve deeper into Quantum Mechanics and some of the fantastical cosmologies (I recommend reading "The End of Physics" by David Lindley";) but it doesn't change what is needed to perceive at our level of interaction with the environment. On some level, it is enough to say that the apple falls to Earth. In space flight, it is more accurate and necessary to say that objects of mass exert gravity upon each other.

So, if were going to explore or extrapolate on Learning Theory or any other model of behavior description or modification, we have to agree on a common language. Otherwise, it would be the Tower of Babel. IMO.

spiritdogs

I like that idea, too.  I'm especially interested in the application of classical conditioning with aggressive behavior, a la Jean Donaldson's suggestion about feeding the dog even while it is growling at a scary looking human to change the dog's basic perceptions about scary humans.

When we started using that technique in Cassidy's Difficult Dog class, it seemed really odd to me at first. Everything I'd learned or read up to that point (obviously not all that much!) stressed that behavior rewarded is behavior that will continue. So to "reward" a leash reactive dog with treats seemed completely counterintuitive. We don't WANT them to keep doing this, right? So why are we rewarding it?

But then when it was explained in class, and I read further, it started to make sense. When a dog is reactive because it's stressed or fearful, simply issuing corrections may eventually stop the behavior. But do you want your dog to not go ballistic at the sight of another dog because he fears the consequences, or do you want your dog to not go ballistic at the sight of another dog because he longer feels stress and fear in the presence of other dogs?

This is an important distinction. Counter conditioning is not about changing behavior per se, it's about changing the emotional response to a trigger, such as other dogs, so that the behavior extinguishes because the reason for the behavior, the stress and fear, is gone. It's about creating a positive association towards the trigger - a dog that sees another dog and anticipates great things has no need to react negatively - vs a dog that knows he's going to get a correction when he sees the trigger, which can escalate the behavior.

I used this technique with Cassidy and my cats too. Knowing that she was reactive on leash I had to come up with another way to control her when at about a year old, I brought home Elvis as a 3 month old, 4 pound kitten. I spent months doing supervised visits between them in a spare bedroom under controlled circumstances, rewarding her for calm behavior and banishing her from the room when she misbehaved. She learned that if she wanted to spend time around the oh so interesting kitty she had to play nice. Looking at and sniffing the kitty was reinforced, charging him or snapping at him, or even blowing me off, meant "Ooops, you're done!" and she got booted from the room. Her behavior drove her access to him, the choice was hers.

Eventually they became good buddies and I added Emmy to the pack later, but it took a LOT of time and work, way more than just putting a prong on her and yanking the snot out of her would have. But because I knew how reactive she was on leash I wanted to make sure that kitty ALWAYS = good stuff for dogs. There was a lot more I did too, distraction exercises to divert her attention off him once they were loose in the house together, and a brief bout with a citronella collar too. But that was only after many, many months of creating positive associations.