Deprivation In Training

[Red Fox]

As per the title.

I believe that the word "deprivation" is all too often misunderstood and misinterpreted by those who do not understand it, nor the role that it plays in training.

Thank you.

[huski]

Cosmolo started a thread on this not long ago - when I'm not on my iPhone I will find the link

[Red Fox]

Ah cool, I might do a search then. I think it's about time to bump it

eta found it: http://www.dolforums.com.au/index.php?show...;hl=deprivation

Edited June 25, 2010 by SecretKei

[corvus]

The purpose of deprivation in training is (has to be) to increase the appetite for the reward on offer. Deprivation is the classic example of an unconditioned establishing operation. EOs work to increase the effect of a stimulus. So in the case of food deprivation as an example, the dog is hungry, therefore they become more motivated to gain a food reward and the food reward makes them feel better than it does when they aren't hungry. What's more, they suddenly increase the frequency of behaviours that in the past have been associated with food rewards. There is a paper freely available about it here: http://www.ncbi.nlm.nih.gov/pmc/articles/P...n00026-0061.pdf

The reason why I make the "sweeping generalisation" that I do not think deprivation is necessary in training the average dog is that there are other ways to develop an EO than deprivation. One can play around with cues and anticipation and create a very strong anticipatory response with a conditioned EO. Drive trainers already do this. It is my view that if you're clever enough about it you don't need to deprive, and that is not based on my own experiences alone, but also those of people that do compete with dogs of varying drive levels. I wouldn't do it unless I felt I had to and that's my ethical standing on it. A "hunger" for something is, in general, an aversive state and that's why it works. Furthermore, I have not found it to have a very large effect at the level I would be comfortable using it.

Having said that, there is a time and place for it. They use it extensively in the military to gain a level of reliability that would otherwise be impossible. The level to which they do it is outright cruel if you ask me. The nature of an EO is that you get a temporary effect, as in, once the dog is sated the effect fades. So to use this method to get a constant effect it is necessary to keep the dogs constantly hungry, and that's what they do. They have to use an essential resource because the stress created in this working environment would kill off any desire for a non-essential resource (e.g. a tug toy). And besides which, the work being asked of the dog is very active and strenuous and they are going to get sated for game or bite rewards very early. You can't deprive them much if you're using it as a reward daily.

As I have said before, I do use deprivation for my hare. He is not a naturally opportunistic animal and he is so easily frightened that it's very difficult to create anticipation for anything good. I have found it to be necessary to motivate him to get him to even want to work for food in the first place. Having said that, I don't give him less food. I just give it to him in smaller portions so that he gets a little hungry just before I feed him.

[Bubitty]

Hmm so is it deprivation if I used to hand them out all sorts of additional junk for free and now I stop?

Or is that purely making the dog work harder for treats?

[Staranais]

Hmm so is it deprivation if I used to hand them out all sorts of additional junk for free and now I stop?

Or is that purely making the dog work harder for treats?

We had another thread about this a few days ago, and I said then that I believe deprivation is a sliding scale. I believe that everyone uses SOME deprivation in their training. To not use deprivation at all you would have to give your dogs all treats, all toys, all attention whenever the dog wanted them.

My girl only gets certain foods, and certain toys & games, during training sessions. She does not get them at other times. This is why she will work for them. If she got these things for free whenever she wanted them, she would be less inclined to work hard for them.

This is the very basis of reward training, IMO - the dog can only access the reward contingent on its behaviour, and it is deprived of the reward if it does not behave as desired.

[huski]

What Staranais said

Corvus I am assuming you must give your dogs access to food, toys, your praise and attention whenever they want? Because if you don't, you are depriving them. I deprive my dog of food every day unless we are training or I choose to feed her; she's hungry all the time and I don't free feed her. I don't feed her whenever she wants food, either.

Edited June 28, 2010 by huski

[corvus]

Staranais, I think you are using a different definition of deprivation to me. In psychology I guess most things are defined by the effect that they have. So I'm defining deprivation by the effect of increasing appetite and therefore motivation. If I delay a reward, it's pretty debatable whether my dog feels that "loss" particularly keenly or not. He might if he were in an emotional state that made him more sensitive to a sense of having lost a reward, but most of the time he would not even react except to try to do something else that will get him the reward. There'll be no detectable change in his body language that might indicate a change of emotional state, for example. If there's no increase in motivation or anticipatory behaviour I would assume it is not deprivation in the sense that the term would be used in motivational theory and so on - which is what we would presumably be trying to achieve in dog training.

I'm a believer in allowing an animal access to resources that matter to them. When my hare is waiting for his next meal he still has hay to chew on if he gets desperate. When my dogs are not training they still have access to lots of tug and chew toys. Mostly because if I let them have the toys I train with Erik would rip them into tiny little pieces. So I do not view this practice as a form of deprivation, because the means are there for them to fulfil their needs without me. What makes training toys so awesome is not that I deprive them of them when we are not training, but that I pair them with super awesome fun, so they become an establishing operation in themselves. If I let them sit down and rip bits off whenever they wanted to I'd be disrupting that conditioned EO and that's why they would become less motivated to work for them, not because they were sated all the time. At least, that's my view. One of Erik's favourite toys is not a toy I habitually put up when I'm not working with him. He does like to rip bits off it, but he'll get tired of it and leave it on the floor and ignore it for days, then come back to it and so forth, but the moment I pick it up, it's game on and he'll do anything for the chance to tug on it with me. For him, me picking up a toy has become a signal for imminent super awesome fun. His enjoyment of tug with that toy is not in the least diminished for having near constant access to it.

Bub, I would think that asking for work in exchange for goodies that have previously been free is probably deprivation. I did this with my hare the first time I was trying to get him to work for food. He still got all the food he wanted, but he only got yummies directly from my hand. It turned out he wanted some yummies in his diet enough to work for them, but I had to use truly amazing hare lollies to convince him he wanted my yummies. I started giving him yummies for free again and that's where all the current problems of refusing to have anything to do with me started. It's multi-faceted, though, and the treats are only one aspect of the problem.

[Staranais]

Staranais, I think you are using a different definition of deprivation to me. In psychology I guess most things are defined by the effect that they have. So I'm defining deprivation by the effect of increasing appetite and therefore motivation.

Well, I'm just using the common dictionary definition.

But even by your definition, I would say you are depriving your dogs, in a very mild manner, by only allowing them to access certain rewards contingent on their displaying particular behaviour.

To put it bluntly: the rewards only work since your dog cannot access them as much as he would like to when you are not training. Premack.

If the dog could access the reward at all times, (in other words if he was never deprived of the reward), then his appetite and therefore motivation to earn the reward during training sessions would go down. This is the change in behaviour that results from not depriving your dog. This would be an observable, objective result of giving your dog access to the reward whenever he wanted it.

You say you need not deprive your dog of his toys for him to want to work for them - this is because when the toy is sitting around the house, it is not nearly as exciting as when you are playing with it. In other words, the reward is the exciting interaction with you, not the mere toy. And the dog is deprived of this exciting interaction except when he is performing for you.

I do exactly the same thing, as we talked about in another thread. I can leave my toys around the house and my girl will still work for them when I ask her to - this is because she is working for the play "experience", not the toy alone. So between training sessions the deprivation is still going on - she is just being deprived of excited praise & interactive play from me, rather than the actual physical toy. If I played with her like that all the time, to her hearts content, she would have no need to work for it.

Get what I mean?

[Vickie]

So I'm defining deprivation by the effect of increasing appetite and therefore motivation.

Let's say I asked my dog to run a sequence with no presentation or hint of reward. Let's say I then got a toy out & showed it to them, teased them a little with it & then repeated the sequence. Am I increasing appetite for the toy & therefore motivation to do the sequence faster? I think yes. It would not increase appetite if I had hidden the toy away for weeks, they would forget it even exists. It is the sight of the toy that increases their appetite and the delay in allowing them to have it which inspires motivation.

If I delay a reward, it's pretty debatable whether my dog feels that "loss" particularly keenly or not. He might if he were in an emotional state that made him more sensitive to a sense of having lost a reward

emotional state? do you really think dogs are capable of that thought process to become sensitive about what they could have had?

but most of the time he would not even react except to try to do something else that will get him the reward.

only if you are shaping. If I delay a reward when my dog is sitting in front of a jump/sequence, they are not offering behaviours...they are quivering in anticipation of their release word, the longer I deprive them, the more they blast off.

Edited June 28, 2010 by Vickie

[Tassie]

Staranais wrote

In other words, the reward is the exciting interaction with you, not the mere toy. And the dog is deprived of this exciting interaction except when he is performing for you.

Yes!! This is how it seems to work for me too.

[Vickie]

It is the sight of the toy that increases their appetite and the delay in allowing them to have it which inspires motivation.

Just to qualify this...appetite can also be increased by things other than sight of the reward. ie... a sound/word used in association with the promise of a reward, a routine used in association of a reward, a request for a behaviour that has in the past always been heavily rewarded.

LOL, I think I am going way off track...but was just thinking about what I wrote & realised that I actually would not expect to see from my dogs a significant difference in performance based on whether I produced a toy or not. Still believ in the theory though .

[corvus]

Let's say I asked my dog to run a sequence with no presentation or hint of reward. Let's say I then got a toy out & showed it to them, teased them a little with it & then repeated the sequence. Am I increasing appetite for the toy & therefore motivation to do the sequence faster? I think yes. It would not increase appetite if I had hidden the toy away for weeks, they would forget it even exists. It is the sight of the toy that increases their appetite and the delay in allowing them to have it which inspires motivation.

Ah, this is a bit more complex than it appears, I think. Are you increasing appetite through deprivation or are you increasing appetite through value and expectation? I think the latter, because you say that it is the sight of the toy that increases the appetite. How could you say that 5 seconds of deprivation has a stronger effect than the promise of imminent reward? I predict if you got that toy out and didn't habitually allow your dog to have it once she'd seen it, then you would not see that increase in motivation. UNLESS she badly wanted it for its value alone and your teasing served to frustrate the heck out of her. Frustration is a dicey thing, though. If you frustrate them too much their motivation plummets. They have to believe they can win to put in more effort. Would you agree?

emotional state? do you really think dogs are capable of that thought process to become sensitive about what they could have had?

Well, in my field we call it affective state so that no one thinks we're anthropomorphising and saying that animals are capable of consciously identifying an emotion. I say emotional state because affective state sounds wanky and unintuitive, like practically everything psychologists come up with. There was an interesting study done by Burman et al. in 2008 on sensitivity to reward loss as an indicator of negative affective state in rats. It's a bias. They have to have an expectation of what they will get to feel they have missed out. And they feel the loss more keenly when they are in the pits.

only if you are shaping. If I delay a reward when my dog is sitting in front of a jump/sequence, they are not offering behaviours...they are quivering in anticipation of their release word, the longer I deprive them, the more they blast off.

Again, how can you say that is the deprivation and not the anticipation of reward that is creating that effect? You would have to start from scratch with something that has no conditioned value in surroundings that are neutral and an activity that has no history of reward to tease them apart. That anticipatory effect is very powerful. It can overcome the effects of chronic stress and depression.

The theory is sound by my way of thinking, Vickie, I just think you're more likely describing a conditioned EO related to anticipation rather than the unconditioned EO of deprivation. If I condition my dogs well enough with a solid reward history I would expect them to perform equally well with or without dangling a motivator in front of them. But if we were talking about dogs that haven't been conditioned that way I think the theory would hold.

I don't think there is any sense arguing whether delaying a reward is deprivation or not. By my way of thinking, it's neither here nor there, as when people talk about deprivation in training they are not talking about delaying a reward for a few seconds, or only delivering dinner if the behaviour meets criteria. At least, not the discussions I've seen on DOL. I don't think there's much point arguing whether starving a dog for 24 hours has the same appetite increasing effect as delaying a treat for 5 seconds. If it did, no one would ever starve their dog for 24 hours.

[corvus]

Staranais wrote

In other words, the reward is the exciting interaction with you, not the mere toy. And the dog is deprived of this exciting interaction except when he is performing for you.

Yes!! This is how it seems to work for me too.

My problem with that reasoning is that it assumes the dog is in a state of "hunger" for interaction with you whenever you are not playing tug with them. Now, don't get me wrong. Erik certainly has ways of telling me he would badly like to play tug now, please. But it's related to arousal levels and expectations. He's not perpetually trying to get me to play with him. Only when he's aroused and has nothing constructive to do with his energy.

[Bubitty]

Bub, I would think that asking for work in exchange for goodies that have previously been free is probably deprivation. I did this with my hare the first time I was trying to get him to work for food. He still got all the food he wanted, but he only got yummies directly from my hand. It turned out he wanted some yummies in his diet enough to work for them, but I had to use truly amazing hare lollies to convince him he wanted my yummies. I started giving him yummies for free again and that's where all the current problems of refusing to have anything to do with me started. It's multi-faceted, though, and the treats are only one aspect of the problem.

Does this mean he would suffer if I bumped his rewards down a notch? So in this case, I have effectively created my own monster?

[megan_]

Bub - he won't suffer. He just might sulk a little. Dogs will, in general, do the least amount of work possible to get the reward. At the moment, he gets the reawrd for no work.

"Work" doesn't have to be anything labour intensice. I give mine treats and pats for targetting my hand with their nose.

[Vickie]

Frustration is a dicey thing, though. If you frustrate them too much their motivation plummets.

Thank you I will keep your advice in mind....so far so good

I don't think there is any sense arguing whether delaying a reward is deprivation or not.

You're right. There is no sense in arguing. I believe deprivation is an excellent training tool & will continue to use it, but probably never to the extremes some people choose to, unless I become a hell of a lot more competitive.

You don't believe it is necessary at all. Each to his own . Hopefully we will both still be able to achieve the results we desire.

[corvus]

Does this mean he would suffer if I bumped his rewards down a notch? So in this case, I have effectively created my own monster?

Depends on what you consider "suffer", and how much he has come to expect the rewards, and how much he likes them. My guess would be at the most a couple of days of feeling a bit down until he got used to the new regime. Here's an example: If you teach a rat to run to a cup full of treats on a signal, and you normally give the rat 12 food pellets as a reward, and then you suddenly give the rat 1 pellet as a reward, the rat is gonna be bummed for a little while. It'll be aversive to him because hey, what's this 1 pellet thing? He normally gets way more than that. He'll get slow on his run to his cup of rewards. But, after a few days of the new regime he'll get over it and start running to the cup as fast as he did when he was getting 12 pellets. How bummed he gets about the reduction in rewards isn't set in stone. Maybe he's being picked on by the other rats, or his cage is boring, or he's lonely, or he doesn't have a wheel to run in. All those little details might make him feel this reduction in reward more keenly and mean that he's bummed for longer. Inevitably, though, he'll perk up and get back to running as fast as he did before.

That's all based on rats, who have nothing better in their lives than running to a cup with treats in it. It's all together possible that you wouldn't see any effect in your dog at all. Maybe he doesn't anticipate the treats, just accepts them as they come. Maybe he wasn't thoroughly wild about the treats and could take them or leave them. Or maybe there's a lot of other awesome stuff going on in his life that means he's not feeling that he's missing out on anything. Maybe you start teaching him how he can control the dispensing of treats and the anticipation of being able to control the treats easily overcomes any negativity he might be feeling from not getting them for free.

So, it's not straight forward, but it's not going to be a big deal to him even if you just suddenly dumped treats from the diet completely and gave him no opportunity to earn any. He'd get used to it pretty quickly. I'm pretty freaking conservative about doing anything aversive to my animals, and I would do it. Sometimes they need a little change to show them how much more fun it can be than it is at the moment. The reason why I persist with my hare is because I know he has enjoyed interacting with me in the past and if I can just remind him that he liked it we'll both be happier. It's worth it.

Sorry, longest answer ever to a straight-forward question.

[huski]

So, by your definition Corvus, keeping super high value food for training sessions only is deprivation?

[Tassie]

Staranais wrote

In other words, the reward is the exciting interaction with you, not the mere toy. And the dog is deprived of this exciting interaction except when he is performing for you.

Yes!! This is how it seems to work for me too.

My problem with that reasoning is that it assumes the dog is in a state of "hunger" for interaction with you whenever you are not playing tug with them. Now, don't get me wrong. Erik certainly has ways of telling me he would badly like to play tug now, please. But it's related to arousal levels and expectations. He's not perpetually trying to get me to play with him. Only when he's aroused and has nothing constructive to do with his energy.

I'm probably spoilt by having Border Collies, who, while they have a really good off switch, are/should be IMHO

genetically programmed to want to do 'stuff' with their person if their person is doing 'stuff' - if that makes any sense.