Spin-off: what makes a positive trainer a positive trainer?

mudpuppy, you've stated three different things - in the first post from which i responded with a "no, it hasn't", you stated:

"It's been thoroughly proven that "punishment", i.e. a correction applied after the "misbehavior" occurs, is the least effective method of modifiying behavior."

okay, so i thought we are talking about behavior MODIFICATION..... then in the following post, you state:

"Punishment by definition STOPS a behavior. If you want your dog to perform a behavior, lets say HEEL"

whoaaaa, now you've switched to 2 other things - stopping a behavior and teaching new behavior........... which one do you wish to discuss? 1) modifying a behavior 2) stopping a behavior or 3) teaching a new behavior? you also state:

"Scientists and psychologists and behaviorists have definitely proven that punishment is the least effective way to modify behavior in every species they've ever studied."

first off, i wish you would cite your references. we could examine them ro see whether or not they are polarized. by that, i mean, as an example of polarization, the way that bush's scientific advisors convince the bush administration that there is no global warming...... funny how special interest can bend things into any perspective they wish..... science is not static, it is not stuck in the mud, mudpuppy.... science is fluid.... it was once thought that newton's equations of motion were the holy grail..... but that was upset by modern physics..... which is still in a state of flux..... if rigidity is what you wish to see, then rigidity is what you will see...... secondly, when you bring up the word punishment, why is it that people such as yourself immediately associate a leash pop? i'd really like to know that..... is the image of connecting a leash pop and punishment stuck in the mud as well, causing any other information to simply be drowned out?.... thanks in advance.

"It's been thoroughly proven that "punishment", i.e. a correction applied after the "misbehavior" occurs, is the least effective method of modifiying behavior."

 

This statement is flawed. First of all, punishments are stimuli that STOP a behavior, so "a correction applied after the "misbehavior" occurs" isn't a punishment at all, since it isn't the catalyst for stopping behavior. Aversives applied after the behavior has stopped, aren't punishment, in the OC sense, they are simply punative. I agree, those would seem to me to be the least effective (rubbing dog's nose in his own pee on the carpet, for example).

A punishment must occur DURING the behavior in order to STOP the behavior! So, a leash pop applied DURING leash pulling can be effective. Of course, applying a leash pop after leash pulling has already stopped would be totally ineffective. Same goes for rewards, in operant conditioning the stimulus and the response MUST be connected to have any meaning. Rewarding a dog for sitting, after the fact, while he's walking, does nothing to support the sit! This seems obvious, to me. And, since, I've never seen anyone recommend applying corrections or punishments AFTER the behavior has already been corrected, I don't understand the intent of the statement.

Back to topic:

The version of "positive trainer" that I don't like, suggests someone who gravitates towards clickers and rewards, focuses on "training" protocols, and prefers the use of equipment (collars and leashes) to being hands-on with a dog. It also implies that social factors like pack dynamics are disregarded, and each dog is seen as an independent and greedy player, motivated by particular and discrete things that can be produced (treats, toys, praise, privileges). This attitude of "positive trainer" that I'm describing feels formulaic to me, and devoid of intuitive and social aspects of relating with other beings. I am not at all attracted to that kind of trainer mentality, as it feels to me like a manufactured position that, ironically, uses the language of science to cover the projecting, onto dogs, of unresolved emotional human baggage.

For me, however, the version of "positive trainer" that I am interested in, would be someone who strives to not only educate themselves with new and old, high and low, knowledge, but is also aware of their own personal power and physical presence and strives to work most efficiently, kindly, and appropriately with an animal.

 A punishment is something which decreases the likelihood that a behaviour will be repeated. 

So, rubbing a dogs nose in his mess is not punishment.  Stopping a dog from doing something he should not be doing is not punishment, unless you use an aversive as well.

Punishment is used in behaviour modification.  It is used to stop (ie, eliminate, not just stop at that moment) unwanted behaviour, allowing desirable behaviour to continue.  It is a consequence of a behaviour - ie, it immediately follows the behaviour.  Example - Dog pulls, dog is punished (leash correction) dog stops pulling. 

The definition of "stop" and "behaviour" in MP's posts was fluid, but not inconsistent.  While I do not always agree with MPs posts, I do at least find that they are concise and they make sense, which is more than I can usually say for my own....

 Ixas, what I have learnt from dealing with everything from frogs and lizards to kangaroos and cows, is that it is formulaic. It's formulaic because of a wonderful little thing called convergent evolution that just blows my mind. It's the very same methods I would use to deal with human adults, as well. I think it's wonderful that it's formulaic. It makes it easier for everyone. If it's not working, you know that's because you haven't found the right motivation yet. That's where it goes the way you want positive training to go. It goes that way BECAUSE it's formulaic and you can say "Oh, well this always works on every animal ever, so what am I doing wrong?"

I think the kind of positive trainer you first described is going to do no better than a traditional trainer in the same mindset of "This is how you do it every time". It's vital for success in any training to know your animal and what drives them.  

corvus

 It's vital for success in any training to know your animal and what drives them.  

 

That's a "formula" I can agree with! Though I have to add that it's equally vital to know oneself!  

I'm a formulaic guy. f(x) = x^2. So f '(x) = 2x. Ahh, sexy.

Perhaps "positive" trainer is a misnomer as trainers who are labelled as positive actually use any one of the quadrants that are necessary. Their toolbox is indeed wide and deep. A big toolbox is not several arrays of punishments or corrections. It is several rewards, several deprivations, several ways of ending an aversive in the wake of desired behavior. As well as several aversives, punishments, corrections, whichever word that you mean for doing something to stop a behavior. In the truest behavioral sense, a + punishment stops a behavior and creates a change to not engage in that behavior in the future. A + reward reinforces a behavior to happen again, to earn future rewards. And this is supported by the unalterable fact that every creature does what works for them. A positive trainer will strive along the path of + reward because, due to the effects of other quadrants together, some behavior will simply extinguish when the creature knows which way the reward lies. It also makes it easier to use -P or deprivation of a desired reward to extinguish a behavior once the creature knows which way the reward lies. And -R, or the cessation of an irritant in the presence of the desired direction also becomes more effective. It's not that a positive trainer won't use +P, it's just not the first tool out of the box.

 

"whoaaaa, now you've switched to 2 other things - stopping a behavior and teaching new behavior........... which one do you wish to discuss? 1) modifying a behavior 2) stopping a behavior or 3) teaching a new behavior? you also state:

"Scientists and psychologists and behaviorists have definitely proven that punishment is the least effective way to modify behavior in every species they've ever studied."

first off, i wish you would cite your references."

 

what's the difference? it's all behavioral modification- you can think of teaching the dog to heel as stopping the dog from doing anything other than heel (i.e. a punishment-based approach); or as teaching the dog to assume and hold a specific position (i.e. a reinforcement based approach); or modifying a behavior (changing where the dog walks in relation to you).  And I'm not going to do your research for you- there's forty-odd years of research into behavioral modification readily available to anyone who looks. Why don't you start with Karen Pryor's work? she has some easily understandable books that explain the concepts and provide references. And then go try them out on your friends-- try playing the "training game" where one person pretends to be a dog, and the others try to train the "dog" to do or stop doing something, using a variety of methods. See what works the best.

And I'm not going to do your research for you- there's forty-odd years of research into behavioral modification readily available to anyone who looks.

thanks for the confirmation - lol - and your other reference is biased.

It's been thoroughly proven that "punishment", i.e. a correction applied after the "misbehavior" occurs, is the least effective method of modifiying behavior.

wrong

Scientists and psychologists and behaviorists have definitely proven that punishment is the least effective way to modify behavior in every species they've ever studied.

wrong again.

example number 3:

we recently got 2 new puppies. jasper (who is now 7 months of age) and smokey (3 months)

mandi does not like to have other dogs stick their noses into her food bowl while she is eating. when we forst got jasper at 2 months of age, he did not know any better and tried to poke his head into mandi's bowl while she was eating. she snarled at him and he got spooked. then he tried it again the next day and she snarled at him again. he never tried again. (

(+P conditioning here - sorry to burst your bubble of perception but your 2'cd claim quoted above is simply untrue.)

mandi taught jasper not to stick his head into her food bowl while she was eating - and it was quick and it was very effective. and guess what? she has done the same thing with smokey - 2 snarls later, smokey no longer will stick her head into mandi's food bowl.

 i'm pretty well convinced that you get your information from polarized sources which have special interest at play. indeed, horses are not the only animals that can be "made" to wear blinders.

Wow talk about  a well analyzed subject, on both sides of the issue. Interesting reading for sure.

PRT is exactly what it says it is and really anything thrown in the mix in the form of punishment is a modification. If you use a choke collar, prong collar, rolled up paper you are certainly not using positive reinforcement training. If you use them you are spinning it to make it seem so but those items have no place in positive reinforcement training.

As far as reliability through a training method my border collies have rock solid recalls, are fine off leash or on. Two are doing agility and Mozart is doing great at it. Roc will be going for his CGC shortly and will get it. Capp could easily pass the therapy dog test if I so desired and all I ever used was a clicker, treat bag and some patience.

The toughest part is patience. Unlike punishment and pain the results may take a bit longer.

OK, feel free to hammer away at me.

ron2

I'm a formulaic guy. f(x) = x^2. So f '(x) = 2x. Ahh, sexy.

 

 

ahhhhhh so solve for d (dog), when d's behavior is a function of cat, c...

 

d'' + d' = c + 1

lostcoyote

ahhhhhh so solve for d (dog), when d's behavior is a function of cat, c...

 

d'' + d' = c + 1

 

The dog's first and second derivative actions are equated to or are in response to the actions of the cat plus an unknown variable, arbitrarily assigned the unity value, or 1. In terms of vector analysis, the arc of the dog's path at a single point is defined by the first derivative, d'. The single point at which that slope of d' changes is d ''. Both derivatives of the changing vector are a function of or defined by the motion of the cat.

Now that is sweetly done. Talk dirty to me ...

We may not always agree on things but I likes me some math.

 

c= (d' + d '';) - 1. Is that a fancy way of saying the cat is one step ahead of the dog? Or certainly hopes to be?

ahhhhhh so solve for d (dog), when d's behavior is a function of cat, c...

d'' + d' = c + 1

solving for d...

d=k1e^(-c) + 1/2c^2 +k2 where k1 & k2 are boundary value constants

 (this ain't algebra)

 

cross check:

d=k1e^(-c) + 1/2c^2 +k2

so

d' = -k1e^(-c) + c

d'' = k1e^(-c) + 1

therefore,

d'' + d' = c + 1

 

and indeed, despite whether the dog chased the cat, the cat still got the mouse by being 1 step ahead :) 

lostcoyote

and indeed, despite whether the dog chased the cat, the cat still got the mouse by being 1 step ahead :) 

but while the cat was busy with the mouse, the dog got into the cat poop o.O

lostcoyote

d'' + d' = c + 1

lostcoyote

d'' + d' = x + 1

Ah, you changed constants on me. And you provided more than the initial info. But still, it looks quite elegant. And certainly entertaining.

We should probably start a thread for mathematical descriptions for this tomfoolery. You and I might be the only ones reading it.

I have a feeling that there is a multivariable calculus to describe what happens when the dog catches up with the cat.

 

 

the typo has been fixed