Positive Re-enforcement Only Techniques

[poodlefan]

Aatainic:

I think pgm's point is that there is something instinctively superior to a dog that doesn't get external rewards.

I've met a couple of dogs like this - they work to their own set of rewards and they were difficult to control. It may be great that a dog enjoys its work however, if I want to control when, where and how it works, I'm going to need some form of external system to get it to STOP. Yes, I could probably use an aversive but I'd rather be more positive.

Give me a dog that looks to me for feedback any day, trained or untrained.

[poodlefan]

PQM:

When one treats it as the whole story, I am inclined to believe that one ends up with a dog constantly in need of treats.

Can I suggest PQM that you have a more narrow definition of 'reward' than I do. I know of dogs that have never been trained with food but whose handlers use 'positive motivational methods'. The key is to find our what motivates the dog and there's plenty of dogs around who do better on praise than treats.

Acknowledgement is something that can only occur within a social relationship. Rewards as understood within behaviorism (the science of behaviorism that is)give no consideration to that relationship.

Any one who thinks that training occurs outside of a relationship with the dog (be it positive or negative) ain't never trained one IMHO.

[pgm]

Poodlefan: Can I suggest PQM that you have a more narrow definition of 'reward' than I do. I know of dogs that have never been trained with food but whose handlers use 'positive motivational methods'. The key is to find our what motivates the dog and there's plenty of dogs around who do better on praise than treats.

PGM: I am not talking about rewards - my concern is not primarily with motivating certain behaviors, it is with developing understanding. You talk of reward in terms of 'motivating' - but say I have a narrow definition? Shaping behavior is the most superficial aspect of training - which is why teaching a dog to do something is actually (so long as you know what your doing) the easiest part of training. Getting reliability on the other hand is difficult.

Poodlefan: Any one who thinks that training occurs outside of a relationship with the dog (be it positive or negative) ain't never trained one IMHO.

PGM: really? Rspca quotes a study undertaken by behaviorists (who else?) into the effects of the ecollar. Part of this study involved subjecting dogs to random electrical shocks from the ecollar. Who but somebody completely ignorant of the principles of training with an ecollar would subject dogs to such an experiment? Who but a complete moron would think that the results thus obtained shed any light on the use of ecollars in training? Answer: a group of behaviorists.

Edited February 14, 2005 by pgm

[sidoney]

It is possible to train outside of a relationship. Chickens, fish, and so on can be trained - little to no social relationship with the trainer. (Also can be trained remotely.) However I wouldn't WANT to train outside of my relationship with my dogs - a great deal of what I do with them seeks to strengthen that relationship - gives me lots more options and also a great deal more enjoyment.

pgm, I don't think that anyone is arguing that pure behaviourism is the be all and end all of psychological theories. However I WOULD argue (and have done so, repeatedly) that there is much in it that is useful. Also it gives useful ways of talking about concepts that can be used in training. Even though much of the naming is laden with connotation (eg. reward, punishment) - one has to see beyond the connotations.

Actually, thinking about it, maybe there are some that don't see all sides of behaviourism - for anyone who may be in that situation, it's worthwhile finding out about the negatives as well as the positives (about anything).

Edited February 14, 2005 by sidoney

[pgm]

Sidoney: Also it gives useful ways of talking about concepts that can be used in training.

PGM: well sidoney, as we are both more than aware, I don't agree that talking in behaviorists concepts are useful in training. They are probably not harmful so long as one is aware of their limitations, but I don't agree that what one gains is offset by what one loses in this kind vocabulary. I think you would agree however, that the way we talk about things very much shapes the way we think about things.

Eg. 'positive punishment' - adding something adversive that reduces the likelihood of behavior being repeated.

I don't use 'positive punishment' in my training, so I don't see how it is of use.

[Guest Tess32]

PGM, does your theory about worth of OC theory relate to normal, pet dog trainers learning to training their own dogs, or just to professional trainers with years of experience?

As a person who has only trained my own dog, without theory I'd be lost. Not because I don't have knowledge of my own dog's motivations and behaviour, or am incapable of using my own brain, but because as an amateur, I can't afford to make a huge number of mistakes and experiment without guidance.

When training, an amateur will come up against a number of questions - the order of doing things, is this bit right, will this be counter productive and so on. Where do you go? OC may or may not provide a *complete* framework for training, but most of my relatively simple questions can quickly be answered and answered effectively.

You said you don't need a theory to train a dog, but you DO unless you have years and years of experience.

How many pet owners struggle because they DON'T know simple OC theory?

How many of them yank the chain without understanding how punishment

works?

How many people smack their dogs for "coming" too late instead of rewarding?

How many people give up on their dogs because they are "stubborn"?

All these people are unaware of training theory and their "common sense" simply is not equipped to handle it.

Nat

[sidoney]

I think Nat has hit upon something important - we who are having this discussion here do have some understanding of and experience in dog training.

I completely agree with her when she suggests that a better understanding of reinforcement and punishment would help the average dog trainer (edit, ahem "owner"). It's not a great deal to remember, just a few simple guidelines to follow.

I would argue that this understanding would also help in raising children - particularly young children.

(oh, btw, re use of language - the thread title - "positive re-enforcement" - has a strange sort of dichotomy, eh?)

Edited February 14, 2005 by sidoney

[abzndbonnie]

woah this thread has been going for AGES! dont let me interupt!

[Bonniescot]

I for one am enjoying this thread.......

Keep it going folks!!

[pgm]

Tess#2; the theory of OC goes something like this:

Positive reinforcement: adding something that reinforces behavior.

Negative reinforcement: taking away somthing that reinforces behavior

Postive punishment: adding something that leads to a decrease in behavior.

Negative punishment: taking away something that leads to a decrease in behavior.

This knowledge is implicit in understanding how to train animals in the same way as anyone who can speak has an implicit knowledge of the rules of grammar. The theory is just an explicit description. Useful if you don't know anything about animals, pretty much useless after that. Why?

Because the art of training is surely contained in the knowledge of just what is reinforcing and what isn't. Where in the theory of OC does it give you this knowledge? Whether you realise it or not, this knowledge is given to you by the examples trainers use. Every animal is different, knowing what, and in particular, in what context, something will work to reinforce behavior is something only experience will tell you. This knowlegde which we gain from experienced and successful trainers is by far and away the most impotant aspect of training.

You can train a dog perfectly well without having any explicit knowledge of OC, just as most of us, myself included, can speak and write without any explicit understanding of grammar. You cannot train a dog however, regardless of how well you understand OC without the knowledge of what works and what doesn't work in terms of reinforcement. Nowhere in the theory does it tell you, in any particular situation, what will work as reinforcement and what wont.

And as training always occurs within a particular context, OC, as a very general and to mind, superficial description of behavior, wont help.

Tess32: I can't afford to make a huge number of mistakes and experiment without guidance.

PGM: if the theory of OC was all you had, then you would surely be experimenting on a regular basis. As I said, the theory doesn't tell you anything in regard to particular situations. The examples that trainers give, is what counts.

[Vickie]

PGM,

out of curiosity, I was wondering if you'd mind sharing what type of dog you have and what type of training/competition you do/have done.

[pgm]

vpzn, like Sidoney I have a Hungarian Vizsla that I am currently training for obedience competition. I use a combination of ecollar and Koehler methods to train. Have almost completed training for open, and will shortly be moving on to Utility. I dont plan on trialling until all the training is complete.

[sidoney]

I find that the most misunderstood part of dog training is the TIMING of the various reinforcers, punishers, etc. I find that people normally know at least a few things that their dogs like and don't like (although not the full range, and also some don't seem to realise how environment/distraction influences this, eg. bring kibble to their first night training), but again and again, I see little understanding of how and WHEN to use this knowledge to influence behaviour. For example, and clicker teachers will know this one, soooo many times people will click just AFTER a behaviour. It takes awareness and practice to get the timing right.

They also often don't think about how what they are doing is reinforcing/punishing behaviour. Eg. make a mistake in a course, handler slumps - this punishes the dog for trying (as vpzn(&t) has mentioned). Or, very common, dog whines or barks, handler pats it to get it to be quiet - they have just rewarded whining. When people realise that more than food can reinforce, and more than a kick can punish, then they tend to become more aware of how their actions influence behaviour.

(BTW pgm, how's about a piccy of your boy? I'm always in for a Vizsla ogle! )

Edited February 14, 2005 by sidoney

[sidoney]

Here's a piccy of my 8 year old girl doing agility. Gotta love that cheesy Vizsla grin!

This girl, Shaula, was one of the first fully clicker trained obedience titled dogs in Australia. She gained her CD title at, hm, about 14 months, with her first trials at about 12 1/2 months. She trialed in Novice 4 times, for 4 passes, 3 of which were wins. Her mother Amy gained CDX with several wins and places but that was before I knew about clicker training. Shaula has several progeny competing in the obedience rings, from 2 repeat matings with Huxley (obedience champion) - one of these puppies I have kept, that's Cedar. Cedar and her siblings have no less than 4 obedience champions in their pedigree.

Anyway, after gaining Shaula's CD title, we didn't do any doggy competition for some years, until we came back but this time doing Agility, and again, she is doing well (despite limited competitions) - as is Cedar.

Oh BTW, photo credit pinnicle photography, www.pinnicle.com.au

Edited February 14, 2005 by sidoney

[sidoney]

I will add that back when I was clicker training Shaula for obedience, they were exciting and challenging times to be clicker training. There were not many people doing it and traditional trainers looked VERY askance at it.

Our knowledge was gained via the 'Net through a wide-spread network of other clicker trainers. What we were doing, at least in terms of general dog training, was new and we often had to work from first principles and make up the method as we went along, or adapt and apply from other related situations.

I was lucky in that I had a friend (who had Shaula's sister Adele) who didn't live too far away, and we used to clicker train our dogs together.

[Clicking Mad]

...

Edited January 7, 2008 by Clicking Mad

[poodlefan]

PQM I wonder how much of our debate about 'rewards' is a matter of semantics rather than any actual difference of opinion.

Perhaps you might assist me to get my head around this by explaining to me what, in your view' "motivates" your dog to work with you - for example in obedience competiton?

[pgm]

clickingmad,

dogs have more than one instinct. The instinct to kill is just one of those intincts - an instinct that has been turned down in many breeds. Bordercollies a have very strong stalking instinct - but their instinct to kill is not so strong.

Dogs also like to play, there is no killing in play. Dogs I think enjoy cooperative social activity. I think working in a cooperative activity with humans can be very satisfying.

Listening to an experienced sheepdog trainer on another list is quite revealing to the very different mentality that is brought to dogs. Apart from an occasional pat on the head for a job well done sheepdogs are not rewarded in any overt sense for their work. He tells that in training and working his dogs that he has never once stopped to praise his dog - it would break their concentration. The thought of giving his dogs treats would be the height of absurdity.

[sidoney]

Herding is what sheepdogs were bred to do. Likewise, I don't need to reward my dogs for hunting and pointing game - that's what they want to do.

Sam, the instincts in these dogs has been modified over centuries of breeding.

I'd be interested in hearing pgm's reply to poodlefan's question.

[Vickie]

Listening to an experienced sheepdog trainer on another list is quite revealing to the very different mentality that is brought to dogs. Apart from an occasional pat on the head for a job well done sheepdogs are not rewarded in any overt sense for their work. He tells that in training and working his dogs that he has never once stopped to praise his dog - it would break their concentration. The thought of giving his dogs treats would be the height of absurdity.

PGM,

I think that is a fairly narrow view of what a reward is.

I agree that BC's don't usually want to kill/eat sheep. Their instinct rather is to control them and bring them to the handler. As a handler there are many ways to reward your dog with sheep, just as there are many ways to punish...just in the way you step and whether or not you allow them to do what is instinctual. And to these dogs, that reward is more important than anything else...without them the exercise becomes pointless, both to the handler and to the dog.

I think that using herding as an example can be very difficult for someone who has no experience in it to grasp. For instance on Saturday I rewarded my dog by stopping her while she was working sheep. To the average person, this would make no sense at all as a dog training concept and is probably fairly unique to herding.

I know you don't like theory, but are you familiar with the theory of applying & releasing pressure? They theory (basically) is that you can reward your dog by releasing pressure at the right moment & let them "have their sheep". Applying pressure at the wrong time will often result in disaster. Herding is very much about rewards and only a good handler with theory as well as experience will be able to reward the dog.

I'd be interested in reading the discussion that you are currently reading, maybe you could pm me the link?

Edited February 15, 2005 by vpzn