Clicker = training???

My frustration has nothing to do w/ not being able to convince the masses or anything like that as there have been some great discussions here; it has more to do with the fact that it appears that on this board there is no relatively "safe place" to discuss the methods I use w/o constant questioning and ocassional antagonism from those who use more aversives than I do (yes the CM board has people who post that are critics, but it doesn't seem as thickly populated as the clicker discussions do).

The reason I use a clicker and mostly +R has everything to do with the idea that I don't see any reason to use +P if +R works just as well and with fewer side effects. That's truly the reason I do what I do - it has nothing to do with the scientific side and nothing to do with my personality or a "cult" mentality as has been implied. It has everything to do with seeing the results I want without worrying about my dog's psyche - remember most of us in the clicker side of things have "been there, done that" with other methods and have found them to be flawed for various dogs we've worked with. We aren't unfamiliar with "traditional" methods by any means.

In closing, some things you should remember when using punishment as explained by my applied animal behavior prof and vet behaviorist:

- Punishment must be a consequence of the behavior
- The behavior must be punished, not the dog
- Punishment must follow the behavior *every* time
- Punishment should never occur without the behavior
- Punishment should not be related to the handler
- Punishment must interrupt the behavior
- Punishment should be of the right intensity; not too low as to create a punishment "callous" but not so high as to cause severe pain, visceral reactions and high anxiety
- Punishment must occur w/in 0.5 seconds of the start of the behavior (a longer delay will make it less effective)
- An alternative to the undesired behavior should be provided to the dog
- Punishment should never be used to teach a behavior
- Behaviors induced via negative stimuli cannot be reduced w/ positive punishment

If properly implemented, the use of punishment to reduce a behavior should only be required a few times; if what you're doing increases the behavior (regardless of the behavior or "punishment") the behavior is being reinforced somehow and you should re-evaluate your method. Re-evaluation is also indicated if the behavior does not decrease after several applications of the percieved punishment.

If "uh-uh" is thought to be neutral enough to be taught as a typical "no reward marker", why can't "eh-eh", bump with the leash, a frown, or a finger tap on the rump? What do these gestures/sounds intrinsically mean to a dog that would discount them from being neutral?

most dogs find these things to be intrinsically aversive-- it's not pleasant to have your neck tugged on, or to be tapped, or to realize your owner isn't happy-- you don't have to teach the dog these are no-reward markers because they carry negative motivational power on their own. Eh eh is kind of a gray area. Have you ever heard a dog give a Roar-bark to "correct" another dog? a very short, sharp sound. Delivering a short, sharp sound at most dogs is understood by them as an aversive Roar-bark. Have someone say  Eheh and uh oh and listen to the difference in the sharpness of the noise. A sharp Eh eh is actually a great aversive to use in an emergency to stop a dog from doing "whatever", if you feel you need to use one, because dogs so readily understand it.

So the difference between a "no reward marker" and a "correction" is the "no reward marker" contains information but no motivational capacity; whereas the "correction" carries motivational capacity by its very nature.

Mind you, I'm not against the carefully timed, carefully planned, occasional use of a correction or aversive with certain dogs in certain situations. But most of dog training can be more effectively accomplished without resorting to the endless string of No no collar pop no no bad dog so many dogs are subjected to in the name of training.
 
 

ORIGINAL: ron2
And maybe that's a lesson to be learned here for people who doubt the power of clicker training and the positive operant conditioning. For someone like me, raised in Hell, to see and affect changes with this knowledge, which I have also used on co-workers and crews should show it's power and effectiveness.

Ron, thanks for making it real in terms of why someone might feel so appreciative and connected to rewards-based training. How we choose to interact with our dogs does have so much to do with how we feel about the plight of dependant creatures.

ORIGINAL: Xerxes

So the way I see it is that clicker training is used to train dogs to perform behaviors on cue.
...

That's the creation of a new behavior with the desired end result.

Other disciplines seem to want only to extinguish undesired behaviors-but not replace them with anything. It seems to me that this method quashes and actually represses the dog's sentience and ability to make decisions on it's own. Which would, IMO, seem to disagree with the foundational basis of the philosophy, which is that dogs are basically wolves.

Any thoughts?

You really have to think about a context before answering this question. Teaching what -- what's the goal? In what environment? I urge those of you who haven't read the [linkhttp://forum.dog.com/asp/tm.asp?m=349845]'Working Dogs Crossover to Companions' [/link] to check it out. This discussion really illustrates what is involved in educating a dog (or a human) to live in a dynamic environment... When it is useful to resort to "dry" training responses to work out the mechanics, and when you just have to have the sensibility to dive into it with the dog, and allow nature and other members of your social group to be your teachers as well. It emphasizes the intricacies of how our mind works within social pressure (that not only dogs, but us and our kids have to learn to adjust to), and it outlines some of the roles that our companion dogs have to fulfill: dog as a member of a human family, dog as a member of a human community, dog as a member society of other animals, dog as a biological organism in the world of physical stimulation, dog performing a special function (herding dog, guard dog, etc.) I'd love to see more discussions like that!

ORIGINAL: ron2

Some dogs may come up with their own behavior chains. Dog fights another dog, gets pinned, then released, commanded to sit, sits, gets treat. It's possible for the dog to think "pick a fight, accept the pin, get up and sit, ah, a treat. Let's see if I can do that again." As opposed to a recall, which always gets a treat and takes less effort.

Yes. It's those pesky behavior chains that I think are particularly problematic for exclusive R+ training. You chose a pretty complex chain, but I see simpler ones all the time.
"jump on person, sit, get a treat"
"sniff the ground, look attentively at handler, get a treat"
"pull on leash, pause and look attentively at handler, get a treat"
when told to sit/stay "start to lie down, have handler come back, sit, get a treat"
"bark, bark, bark, bark, bark, bark, bark, bark, bark, bark, bark, bark, bark, bark, bark, bark, pause to catch my breath, get a treat"

These are the most common examples of when I think it's useful to let the dog know that the "jump on person", "sniff the ground","pull on leash", "start to lie down", "bark, bark.............................." portion of the chain is undesireable, and where I don't think upping the value of the reward is the answer. If the dog has chained these things together, then upping the value of the reward reinforces the entire chain. Maybe one could argue that better timing could mitigate these chains, but I'm coming from a background of having lots of drop outs from the local R+ only school show up in my pet training classes. I happen to know the head trainer at this R+ only school, and she is a superb trainer. She is also articulate, and possesses a delightful sense of humor, and at one time we visited nursing homes with our dogs together. I've had many a productive discussion with her. When I said above that I've seen some wonderful R+ dogs, hers are the ones I was referring to. BUT, quite a few of her students don't achieve her wonderful results. And then they show up in my low-brow training club classes, and generally they've come up with one of two solutions to their problems if I can't show them a pretty quick way to fix their problems. And if they haven't been successful with exclusive R+ under expert instruction, there is nothing new about those techniques that I can teach them. One of their solutions is to "manage" the problem by confining the dog whenever guests arrive, exercising the dog by sending him out into the back yard to run around because walking the constantly pulling sniffing dog is too tiresome, confining the dog where he won't be stimulated to bark or at least where the neighbors won't hear him.... The other solution is to "rehome" the dog where he will be "happier." For these dogs, I think that there are several slightly aversive techniques that can show the dog that jumping on people is not part of what is being rewarded, but that the sit nicely will be rewarded profusely, that sniffing and pulling aren't part of the way to get rewards while on walks, that barking incessantly and then pausing isn't the way to get the treat. Again, maybe an excellent R+ only trainer can get these results without the aversive, but I deal with real people with real dogs that are destined to either spend a huge portion of their life in a crate or are destined to be sent away unless the trainer is given some other tools than the ones that have failed them.
My original goal in entering this discussion was to offer a different point of view than that the only possible bad outcome of unsuccessful clicker training is that a dog will offer a bunch of cute alternative behaviors in rapid succession. There are worse outcomes.
Apparently I offended some people with my use of the expression "drinking the kool-aide". I apologize. I didn't mean to compare anyone to a cult member, and I'm sorry for my intemperate choice of words.
On the other hand, it amazes me that a group of people who pride themselves on never ever doing anything remotely aversive to dogs and whose feelings were hurt by my kool-aide comment have no problem with the insults that have been issued to anyone who argues a point of view other than exclusive R+. This thread started out with the opinion that anyone who used anything other than R+ was repressing a dog's sentience and ability to make choices. It included "smacking" children as a comparison to using anything other than R+. It has included the use of "dog friendly" to describe R+ methods with the clear implication that anything else would be "unfriendly" to dogs. It has included the clear implication that anyone who doesn't accept exclusive R+ training as the single best answer in all situations must be arguing out of ignorance/inexperience. It has included a description of abusive parents who never gave positive feedback and lied and reneged on promises as "the (apparently only?) other side" of exclusive R+ training. All of this gives the appearance that either people don't actually recognize aversive techniques that they are using, or that they have no problem applying them to people who disagree with them.

ORIGINAL: mudpuppy
So the difference between a "no reward marker" and a "correction" is the "no reward marker" contains information but no motivational capacity; whereas the "correction" carries motivational capacity by its very nature.

Mind you, I'm not against the carefully timed, carefully planned, occasional use of a correction or aversive with certain dogs in certain situations. But most of dog training can be more effectively accomplished without resorting to the endless string of No no collar pop no no bad dog so many dogs are subjected to in the name of training.

You are singular, mudpuppy, in that you are willing to openly discuss the reality of P+ in the context of rewards-based training. Thank you for that.

What "no reward marker"s do you use?

ORIGINAL: spiritdogs

  People argue here all the time that you can't find something as stimulating as herding sheep or cattle.  But, for example, we have herding dogs who will find a frisbee very rewarding and can be taught not to chase cars, kids, etc. by using that as the reward for "leave it" "come".  And, we have folks who use the sheep as the reward for the dog obeying "that'll do" - just as one might teach a dog to sit and wait before going out a door - the desired activity is the reward for obedience. 

Well, sure. You can teach a herding dog all sorts of obedience commands with other rewards than sheep. I've trained my herding dogs to UD's, TDX's, and intermediate level agility titles as well as teaching them exemplary manners and all sorts of cool tricks without any sheep being involved. And, with the exception of an occasional leash tug or "hey" for sniffing while heeling or an occasional "stop" when the dog headed to the wrong jump in the directed jumping exercise, all that training has been reward based. Interestingly, my few well timed leash pops didn't inhibit my dogs in the least from sniffing the ground and pulling in the harness on their way to their TDX's.

Where it's problematic to use R+ only on a herding dog is when he's actually herding, not when we're teaching scent articles or leave it or whatever. When actually herding, the dog WILL make mistakes and come in too close and create sheep wrecks. Skilled trainers can minimize those mistakes, but they WILL happen because sheep and dogs are involved and it is neither possible nor desireable for the handler to 100 % control what those animals will do. And once the dog has started to cause sheep to run and split he just plain isn't going to offer desireable alternative behaviors to be rewarded, at least not before someone is likely to be injured, without some intervention from the handler. Herding trainers call that intervention "applying pressure" to the dog, but I think it's fair to say that that pressure is in the form of some aversion that will cause the dog to back away from the sheep. Then you can immediately reward the dog verbally if you want, but the big reward is that he gets control of his sheep once he's in the proper place. But, I have yet to see anyone who has come up with an alternative to applying that pressure/aversion first. That's the situation where herding dogs appear to need to have aversives introduced into their training, not when training leave it, or heel or whatever.

ORIGINAL: spiritdogs

Only after the entire course of training, in a controlled, then ever more distracting environment, would I consider using an aversive.  And, by then, at least you have a dog that is not going to totally shut down from the use of that aversive because he's so used to being reinforced and working by offering behavior, that one aversive is balanced by 100's of positives.  Starting with aversives shuts down the dog because he is faced with minimal positives and more negatives, so it becomes risky to move.

Holy quacamole! We agree. Sometimes, after ones best effort to teach something based only on rewards, one might have to consider using an aversive. And that won't necessarily result in a dog being totally shut down.


Edited to add a predicate because I try to uze rite grammer and too spel gud but sometimes i mess up and need to giv myself a correxshun

ORIGINAL: stardog85

My frustration has nothing to do w/ not being able to convince the masses or anything like that as there have been some great discussions here; it has more to do with the fact that it appears that on this board there is no relatively "safe place" to discuss the methods I use w/o constant questioning and ocassional antagonism from those who use more aversives than I do (yes the CM board has people who post that are critics, but it doesn't seem as thickly populated as the clicker discussions do).
.............
In closing, some things you should remember when using punishment as explained by my applied animal behavior prof and vet behaviorist:

- Punishment must be a consequence of the behavior
- The behavior must be punished, not the dog
- Punishment must follow the behavior *every* time
- Punishment should never occur without the behavior
- Punishment should not be related to the handler
- Punishment must interrupt the behavior
- Punishment should be of the right intensity; not too low as to create a punishment "callous" but not so high as to cause severe pain, visceral reactions and high anxiety
- Punishment must occur w/in 0.5 seconds of the start of the behavior (a longer delay will make it less effective)
- An alternative to the undesired behavior should be provided to the dog
- Punishment should never be used to teach a behavior
- Behaviors induced via negative stimuli cannot be reduced w/ positive punishment

If properly implemented, the use of punishment to reduce a behavior should only be required a few times; if what you're doing increases the behavior (regardless of the behavior or "punishment") the behavior is being reinforced somehow and you should re-evaluate your method. Re-evaluation is also indicated if the behavior does not decrease after several applications of the percieved punishment.

I agree with everything you've listed as caveats to using punishment except possibly that the punishment should not be related to the handler. I'm not sure exactly what is meant by that, and would like to discuss it further. My frustration has to do with the fact that as soon as any mention of the use of corrections/aversives/punishments is raised, then a segment of the population immediately jumps to smacking, never providing positive feedback, constant yanking on leashes, and all sorts of other extreme imagery. In at least three of my posts I've very explicitly pointed out that I use reward based training a great deal, and that I've had had very pleasant experiences with some exclusively R+ trained dogs. I've been very careful to use the words "some" and "mild" throughout my discussion of using aversives. If you have the time, take a look at what sorts of spectres have been raised in response to a defense of using "mild" aversives in "some" situations, and perhaps you can get an inkling of the source of my frustration.

And, for the record, since CM has been mentioned a couple times now, I am apparently one of three people left in the solar system who has never watched CM on tv, never viewed one of his videos, never attended one of his seminars, never read anything written by him, and wouldn't recognize him if he was sitting in my living room establishing dominance over my dogs right now. I have absolutely no allegiance nor any antagonism toward him for the simple reason that I have only the fuzziest notion of what his views are. Whatever else you may think of what I say, please understand that it has absolutely nothing to do with any opinion whatsoever about CM, and that there is some territory between being a die-hard CM devotee and being a die-hard 100 % clicker trainer. All I'm trying to do is defend the validity of some of that territory that exists between two extremes. I'm beginning to believe that territory is truly a no man's (or woman's) land.

ORIGINAL: buster the show dog
I agree with everything you've listed as caveats to using punishment except possibly that the punishment should not be related to the handler.

Yep, that's one of my questions.

ORIGINAL: buster the show dog
And, for the record, since CM has been mentioned a couple times now, I am apparently one of three people left in the solar system who has never watched CM on tv, never viewed one of his videos, never attended one of his seminars, never read anything written by him, and wouldn't recognize him if he was sitting in my living room establishing dominance over my dogs right now.

Yep, I never heard of him till I joined this forum either, lol. He ought to pay the R+ folks here for advertising him! []

Stardog:
My frustration has nothing to do w/ not being able to convince the masses or anything like that as there have been some great discussions here; it has more to do with the fact that it appears that on this board there is no relatively "safe place" to discuss the methods I use w/o constant questioning and ocassional antagonism from those who use more aversives than I do (yes the CM board has people who post that are critics, but it doesn't seem as thickly populated as the clicker discussions do).

 
I agree.

Quick reply to Ixas and buster: I'm not completely clear on the reason for the point about not associating punishment with the handler, but I belive it has a lot to do with avoiding the potential side effect of the animal learning that the punishment will only occur when the handler is around and thus reverts to the undesirable behavior when the owner isn't home. I think counter surfing would be a prime example - you don't want the dog to know you set them up for punishment, rather you'd liek them to think it was their actions that caused the punishment to occur so they don't try to get that loaf of bread you left on the counter when you left for work.

Make sense?

ORIGINAL: stardog85

... I'm not completely clear on the reason for the point about not associating punishment with the handler, .... I think counter surfing would be a prime example - you don't want the dog to know you set them up for punishment, rather you'd liek them to think it was their actions that caused the punishment to occur so they don't try to get that loaf of bread you left on the counter when you left for work.

Make sense?

Yes, that does make sense. And it pretty accurately describes my first dog who was well trained not to counter surf. Until one day I left the Thanksgiving pumpkin pie a leeetle to close to edge of the counter before leaving for work. Only a true dog person could appreciate how fascinating the poop was the next day.

ORIGINAL: stardog85

Quick reply to Ixas and buster: I'm not completely clear on the reason for the point about not associating punishment with the handler, but I belive it has a lot to do with avoiding the potential side effect of the animal learning that the punishment will only occur when the handler is around and thus reverts to the undesirable behavior when the owner isn't home. I think counter surfing would be a prime example - you don't want the dog to know you set them up for punishment, rather you'd liek them to think it was their actions that caused the punishment to occur so they don't try to get that loaf of bread you left on the counter when you left for work.

Make sense?

Yes, that does! Thanks for taking the time to explain that! []

What "no reward marker"s do you use?

 
I use "uh oh". I introduce it around age six months. I think it's a big mistake to use anything other than +R and -R (withdrawl of attention) with puppies.  I've been known to lose it and scream NONONO in terror during REAL emergencies. Can't say that ever helped the situation, though.

Yes. It's those pesky behavior chains that I think are particularly problematic for exclusive R+ training. You chose a pretty complex chain, but I see simpler ones all the time.
"jump on person, sit, get a treat"
"sniff the ground, look attentively at handler, get a treat"
"pull on leash, pause and look attentively at handler, get a treat"
when told to sit/stay "start to lie down, have handler come back, sit, get a treat"
"bark, bark, bark, bark, bark, bark, bark, bark, bark, bark, bark, bark, bark, bark, bark, bark, pause to catch my breath, get a treat"

These are the most common examples of when I think it's useful to let the dog know that the "jump on person", "sniff the ground","pull on leash", "start to lie down", "bark, bark.............................." portion of the chain is undesireable, and where I don't think upping the value of the reward is the answer. If the dog has chained these things together, then upping the value of the reward reinforces the entire chain.

 
dogs are smart, aren't they?  training their owners in this way.
No, upping the value of the reward is clearly not going to work.  Have you tried extinction? course that takes patience.