Can Someone Please Explain To Me All The Training Methods!

[AusDog]

I would hate to be living in a house where the dolphin decided it just wanted to swim flat out inside the house, because it was having one of those, "I'll do what I want moments". not sure holding a piece of fish is going to stop him

Most training such as with these exotic animals is based on denying a resource whether it be food or freedom, and then controlling when the animal has access to it.... Maybe you can be 75% affective at training a dolphin in the wild by offering it an easy feed that it doesn't have to expend to much energy hunting for. But when it decides not to do what you expect of it.. It's not going to destroy your home

Edited October 7, 2009 by AusDog

[Aidan]

What do the Brelands and Baileys do when the species they are training decide they aren't 'into it' at the time or moment?

We have more options with dogs, that does not make the Brelands and Baileys a poor model for how to use positive reinforcement though.

AusDog's argument that we can't use the training of bears, dolphins etc as a model because they are trained in sterile, controlled environments is invalid when you consider that they were also trained - to the satisfaction of the military - in open environments also.

[AusDog]

What do the Brelands and Baileys do when the species they are training decide they aren't 'into it' at the time or moment?

We have more options with dogs, that does not make the Brelands and Baileys a poor model for how to use positive reinforcement though.

AusDog's argument that we can't use the training of bears, dolphins etc as a model because they are trained in sterile, controlled environments is invalid when you consider that they were also trained - to the satisfaction of the military - in open environments also.

I agree in principle.. however I think you will find the conditioning commenced in controlled environments first..or the animal was actually born in captivity... please correct me if I am wrong

[Erny]

What do the Brelands and Baileys do when the species they are training decide they aren't 'into it' at the time or moment?

We have more options with dogs, that does not make the Brelands and Baileys a poor model for how to use positive reinforcement though.

AusDog's argument that we can't use the training of bears, dolphins etc as a model because they are trained in sterile, controlled environments is invalid when you consider that they were also trained - to the satisfaction of the military - in open environments also.

I'm not sure what you're getting at here Aidan.

Certainly I'm not comparing positive reinforcement with positive reinforcement nor suggesting positive reinforcement can't achieve much good. I use R+ a lot in my own work with dogs, but I will use any of the remaining 3 quadrants where appropriate. Unless I've lost the plot somewhere along the line, I think the subject of dog training methods (including P+ and R-) came up by another poster who made the comparison with animals such as the dolphin, who of course are trained using only R+ and P-.

I'm also not sure what "to the satisfaction of the military" means.

[Are You Serious Jo]

The Brelands paper isn't a good example to illustrate issues in dog training though, as the focus was on a different concept. I can't remember the specifics, haven't read it in ages, but it is about when animals don't behave as expected for other reasons, not related to training.

Classical conditioning is also a part of dog training, even if not deliberately taught, so don't forget that as well.

[SkySoaringMagpie]

Bub - "Don't Shoot the Dog" by Karen Pryor has the advantage of not containing a bunch of DOL training forum baggage

It's also a classic in the literature and explains Skinner etc in a readable style.

[Aidan]

Maybe you can be 75% affective at training a dolphin in the wild by offering it an easy feed that it doesn't have to expend to much energy hunting for

Well that was the very interesting part of the discussion with Bob Bailey on dolphins trained in open environments for the military -

they had to perform very long duration behaviours, hard work. At any time they could hunt food, and maybe they even did, but they still came back to the boat for their reward. Maybe it was more than just the free fish? Bob regrets that he was never able to secure funding to look at why this was so.

What's more, they could train WILD CAUGHT dolphins to be deployed, for the military, within 90 days. I'm pretty sure the military insists on better than 75% effectiveness.

I would never suggest that a dog is a dolphin, but if you're looking at what can be done with +R (rather than what can't be done, which of course we should also investigate), then this is the sort of stuff to consider.

[Aidan]

I'm also not sure what "to the satisfaction of the military" means.

The primary role that dolphins play in the military today is to detect mines using sonar, where electronic sonar is not as effective (as I understand it). This is a life-saving operation and the dolphins become very valuable. They must work reliably, in a hostile environment, for long durations. I would suggest that the standards applicable are very high, making this quite a good model for the use of +R in dog training.

[Kelpie-i]

Hi Bub.............. (feels like I'm talking to my OH )

I'm going to try to stay on track here for you....

Nekhbet covered off on the what the quadrants are and just to add to that, they are usually refered to when discussing learning theory. The words "negative" and "positive" are scientific based, meaning to remove and to add, as was explained. This is why you will see some write them as P+ or -P etc.

The above are not so much "methods" or "techniques" of training but rather facets involved in learning. You can refer to the quadrants when discussing the learning theory of any organism, including humans and other animals.

When you talk techniques you might discuss things like luring, shaping, guiding all of which may include 1 or more constituents of the quadrant. These are your methods/techniques for creating the behaviours. The quadrants refer to the reinforcment or punishment schedules used in conjunction with the techniques to either increase or decrease behaviours.

I hope that clarifies it for you somewhat and I am sure that others will add or refine the above.

[Aidan]

The above are not so much "methods" or "techniques" of training but rather facets involved in learning. You can refer to the quadrants when discussing the learning theory of any organism, including humans and other animals.

When you talk techniques you might discuss things like luring, shaping, guiding all of which may include 1 or more constituents of the quadrant. These are your methods/techniques for creating the behaviours. The quadrants refer to the reinforcment or punishment schedules used in conjunction with the techniques to either increase or decrease behaviours.

I hope that clarifies it for you somewhat and I am sure that others will add or refine the above.

Good post I apologise for taking things off-topic above, what the Brelands and Baileys did was use a technique called "free operant shaping", using positive reinforcement to shape a behaviour by successive approximation. This forms a large part of what dog trainers call clicker training.

[Kelpie-i]

technique called "free operant shaping", using positive reinforcement to shape a behaviour by successive approximation.

Yes...we also do this with luring and shaping by SA. Although some trainers see "luring" as the devill re-incarnate

[Erny]

... making this quite a good model for the use of +R in dog training.

Fair enough. And again I agree that R+ can achieve much in training. However what if on one time one dolphin did not do the job required? Did the military rely on only one dolphin?

In the dog world, one wrong thing even by just one dog is sometimes not forgiven by the general public and the law. And of course there's frequency to be taken into account. IE dogs are a part of our lives and therefore their behaviours are encountered on a day to day everyday basis. Consequently the % of times when 'critical' mistakes can occur has to be higher.

I'm not meaning to be argumentative, I'm just not sure if you're getting my point (which would be my communication error) or perhaps I'm not getting yours, save that 'this is what R+ can do'? But reliability over a life time by just one animal doesn't seem to be taken into account.

Edited October 8, 2009 by Erny

[Bubitty]

Thanks Kelpie – I

I am curious though, with you guys who do hard core training, do you really try and ponder these theories etc?

Are they being brought up on Dol as a debate or are they given consideration when you teach your dogs?

Reason I ask is as the average Jo, if I want to teach something, I google the trick and I teach it. Or I just do what seems vaguely ok and see if I have success.

Are you all very well planned in terms of how you teach?

[Aidan]

Hi Erny, I sent you a PM before I saw this, felt I should clarify here for the benefit of anyone else who may be wondering.

I'm just not sure if you're getting my point (which would be my communication error) or perhaps I'm not getting yours, save that 'this is what R+ can do'?

That wasn't really my point. I was really only addressing the assertion that bears, dolphins etc were only trained in controlled, sterile environments which is not true. In fact, wild-caught dolphins have been trained to perform difficult tasks in hostile environments for long durations to a military standard of reliability in a short space of time in the open ocean using these methods.

Whether that makes these methods appropriate for dogs in the home environment or not is not something I have commented on.

[Staranais]

Dolphins, bears, tigers, etc are all trained in cages and sterile environments, they don't live in a family home..not a very good argument on why a particular method should work on dogs in my opnion.

You might be interested to hear that the Brelands and Baileys were training a wide range of species in open environments using +R; e.g dolphins in the open ocean, cats in urban environments.

But as I recall they were also the ones that couldn't train a raccoon to pick up a coin and put it straight in a piggy bank, since their training methodology had no way of countering the racoon's instinctive drive to self reward by "washing" the coins first.

That's right, in 1961 the Brelands wrote the classic paper "The Misbehavior of Organisms" which described instinctive drift and discussed the limitations of Operant Conditioning (at that time).

To see that as a failure of their training methodology overall rather than an observation of a phenomena that had not been accounted for in the theory of Operant Conditioning in it's formative years is to miss the point somewhat. But no, they were not able to extinguish the coin "washing" behavior. Has any balanced trainer successfully attempted to do it since?

I'm not sure if anyone could. I suppose it may be possible to get the desired performance using punishment, but I suspect since the behavour is very instinctual, it would take more punishment than I would personally be comfortable with - I like racoons more than I like TV commercials.

However, unlike you I would call not being able to condition an animal to do something when the animal has a strong instinct to "disobey" a huge failure in training methodology.

As an example, my dog has a very strong instinct to chase and bite things. I like that - I bought her because of it, because I can use that in her training. But I do need a training methodology that allows me to override that instinct when I need to do so. In our case so far, it's been mostly a "positive" training process (using drive training to attempt to provide a similar adrenaline and endorphin reward for doing what I want as she would get from chasing and biting other things). Hopefully this will be enough. However, I will not rule out using punishment in the future if necessary to proof her recall. I do not want a puppy who has been hit by a truck or put down for biting someone or something, or that has to be always kept on leash in some situations and "managed", since she can't be trained to overcome her instincts.

I would never claim that the Brelands weren't superb trainers. But training is continually evolving. I think the tide is turning again - towards balanced training, with consideration for what each dog needs to achieve a reliable level of control along with a good relationship with their handler, and away from positive-only.

Edited for sense!

Edited October 8, 2009 by Staranais

[Kelpie-i]

Thanks Kelpie – I

I am curious though, with you guys who do hard core training, do you really try and ponder these theories etc?

Are they being brought up on Dol as a debate or are they given consideration when you teach your dogs?

Reason I ask is as the average Jo, if I want to teach something, I google the trick and I teach it. Or I just do what seems vaguely ok and see if I have success.

Are you all very well planned in terms of how you teach?

Each individual trainer will have their preferred methods that (as far as they are concerned) are tried and true. There really is no right or wrong and it all depends what you are training for, ie to create, increase or decrease a behaviour. There are 'die hards' on all fields whether it be those who use only food and/or those who believe food is bribery and does not belong in training (these are very old values but some out there still feel this way).

I think the people here on DOL enjoy a debate and discussing (and sometimes arguing ) over things they believe strongly about and/or methods that they swear by. Not just that, there are many different new things be learned and ponder. The quadrants seem to be one of them, especially which "side" of the quadrant you sit on.

If the average Jo wants to teach something new to their dog, I would say use what works for that dog. Going onto Google to find how to teach a dog to rollover is a good way to find the myriad of techniques available to achieve this. It could help you or it could confuse you, depending on the way you interpret the information and your application of the technique.

A trainer and/or someone who works with dogs on a daily basis, it is important to know and understand the learning theory of the animal as well as have a tool box of methods to be able to use effortlessly, so continued research, discussion with peers and experimenting is important.

Edited October 8, 2009 by Kelpie-i

[corvus]

Plans are good, but flexibility is better, IMHO. I'm currently training my wild hare to target an extendable metal stick. Considering his temperament and what drives him, it HAS to be positive or he will never do it. Anything I do with him I have planned quite carefully in terms of what I want as the end product, what I should use as rewards, how long my training sessions should be, how far apart they should be, and how I should go about shaping the behaviour I want over time. All that pretty much goes out the window the moment I step into his cage. What I do all depends on his mood and how he is interacting with me right now. If he is running around the cage, I leave without doing any training at all. If he is hiding in his box and doesn't even peak out within ten seconds of me getting there, I leave without training. I could list a dozen ways his behaviour influences what I do next. It has to be that way because he's such a flighty animal.

In my mind, a plan is vital because it guides my actions so that I am thinking rather than reacting. HOWEVER, being flexible is a great way to get the best out of your animal. The more you listen to them and allow them to teach you, the better you both understand each other.

I think that people here are forgetting that not every dog is a working breed. I know someone who does agility with Basenjis. In her opinion (and the opinion of many other Basenji owners), you have to use positive methods with them because they are like cats. If you punish them, they will want no truck with you and your training. This person I know does very well with her Basenjis, or at least, one in particular who has a good temperament for agility. All the same, some days her Basenjis get into the ring and embarass the hell out of her. What can she do? She can't punish them or they will decide they want no truck with agility rings. It probably won't damage their relationship with her, but it will damage their enthusiasm for agility. All she can do is come back another day. Just like all I can do when my hare is hiding is come back another day.

If there is something I need to be super reliable, I don't mess around with punishing consequences. I make sure my dog is set for success and reward the hell out of it every single time. Often that will have to mean starting in a near sterile environment and gradually adding distractions. Not every dog takes kindly to punishing consequences for non-compliance. Who could blame them? Some dogs you can't punish during training at all or they shut down and do nothing.

[MonElite]

If there is something I need to be super reliable, I don't mess around with punishing consequences. I make sure my dog is set for success and reward the hell out of it every single time. Often that will have to mean starting in a near sterile environment and gradually adding distractions. Not every dog takes kindly to punishing consequences for non-compliance. Who could blame them? Some dogs you can't punish during training at all or they shut down and do nothing.

That potentially can work with a dog that has no history of self reward.

If the dog perceives the ingrained self reward higher then your super duper reward what do you do then?

[Kavik]

Thanks Kelpie – I

I am curious though, with you guys who do hard core training, do you really try and ponder these theories etc?

Are they being brought up on Dol as a debate or are they given consideration when you teach your dogs?

Reason I ask is as the average Jo, if I want to teach something, I google the trick and I teach it. Or I just do what seems vaguely ok and see if I have success.

Are you all very well planned in terms of how you teach?

When deciding on how to teach the seesaw, I did several things (none of which related to the quandrants )

* Looked up seesaw performances on Youtube - some people even have diaries or videos demonstrating how they taught it -> so decided what I wanted my final performance to look like

* Discussed with other competitiors the best way to teach

* Decided how to break the exercise down into separate components and how to train each component (most required shaping)

* Decide which component to teach first and which ones can be taught concurrently

* Decide how to link the components together and any intermediate steps needed to get to final performance

* Look at how to proof performance to get independent performance, understanding and speed

[Tassie]

In answer to the OP - perhaps one additional aspect which maybe hasn't emerged all that clearly so far. IMHO, we need to consider what bit of training we're talking about.

On the one hand we have training in the sense of the teaching of 'tricks' - by which I include most of the performance disciplines, which are essentially tricks - I'd exclude herding from that, tracking to some extent, and I don't know enough about gundog work - but probably that too. These are essentially done in the interests of pleasing us, giving the dog and us something productive to do, and building the relationship between dog and handler. These aren't life or death situations, and for these I much prefer to use R+/P-, and to say, that as trainers, we should be asking ourselves all the time - have I made my expectations clear to the dog? If not, how can I do that better?

On the other hand, there are particular behaviours which while they may be instinctive to the dog, we cannot allow because they could result in injury or death to ourselves, or other people or animals, or at the very least make living with the dog unpleasant. In order to modify or extinguish these behaviours, we may need other tools out of our training tool box - which may include P+, almost certainly will include R-, and/or may be addressed by other techniques such as classical conditioning, or management techniques,

I sometimes think we muddle ourselves by not distinguishing between teaching rules and boundaries and behavioiur modifications in order to be able to live harmoniously and safely in human society on the one hand, and teaching that great range of behaviours which are not strictly necessary, but are highly desirable, and just fun for all. If I don't successfuly teach my dog a change of position for obedience, or a fast and independent weavers - it really doesn't matter - but if I don't fix or at the very least find a way to manage the car-chasing - that matters.

JMHO.