Jump to content

Topic For Discussion - Keep It Nice, Folks ;)


persephone
 Share

Recommended Posts

Quite simply, it is NOT positive punishment.

http://www.dog-training-excellence.com/images/operant-quadrants.png

that just shows how limited those diagrams are ...looks like it needs another quadrant as 'negative' also doesn't cover it (I don't 'take' the treat away, I refuse it and this triggers frustration)...how about indirect positive punishment :laugh: ...

it seems that some people have just issues with accepting that 'not giving something that is wanted or needed' is a very powerful form of punishment and widely used in all realms even for torture.

Link to comment
Share on other sites

  • Replies 173
  • Created
  • Last Reply

Top Posters In This Topic

You are not 'adding' anything, so using the quadrants there is no way that withholding a reward could ever be any sort of positive punishment.

I think what willem is trying to say is that he doesn't distinguish or recognize positive or negative punishment as separate entities because both result in a negative consequence (and hopefully subsequent extinction of behaviour)

Link to comment
Share on other sites

Quote: it seems that some people have just issues with accepting that 'not giving something that is wanted or needed' is a very powerful form of punishment and widely used in all realms even for torture.

Rubbish. Some people just have issues with you using scientific terminology incorrectly. We are talking about effective dog training not animal cruelty. Did you read the part where I said that I train for 7-8 successes out of every 10 attempts?

Honestly it's like saying I'm a vegetarian who eats chicken and fish. It's just incorrect terminology.

Edited by The Spotted Devil
Link to comment
Share on other sites

You are not 'adding' anything, so using the quadrants there is no way that withholding a reward could ever be any sort of positive punishment.

I think what willem is trying to say is that he doesn't distinguish or recognize positive or negative punishment as separate entities because both result in a negative consequence (and hopefully subsequent extinction of behaviour)

thanks for giving my layman thoughts some credit :D ...however it is more that I differ between the phase when you start positive reinforcement training rewarding your dog and the later phase when he is used (conditioned) to it. It is the later phase, where he is used to this kind of training and expects earning the reward that I believe the current terms are limited, actually the subject matter experts seem to ignore the difference. Refusing the treat / reward at the start: I wouldn't call it punishment at all as I believe that it doesn't have a big impact on the dog's behaviour (also a lot of experts would call it a form of negative punishment)...it is just the reward that affects the behaviour (IMO). The later stage is totally different - the dog is now used or even addicted to the reward and refusing the treat becomes a very powerful form of punishment; holding the reward back will trigger now negative feelings / perhaps frustration ...and I'm the one who triggers it, I'm the one who gives / adds frustration to the dog...so yes, IMO punishment, but it doesn't fit smoothly in the existing quadrants.

Link to comment
Share on other sites

Exactly in most cases here we are talking about not giving a treat, not starvation. Although I accept that there are people that withhold meals in order to make an animals more receptive to training with food rewards.

I let my dog have 2-3 attempts at an exercise and if she has no success I'll change it to be easier. She tends to lose enthusiasm or otherwise just go quite wild and fail to engage the brain if she doesn't have success after a few goes.

Link to comment
Share on other sites

You are not 'adding' anything, so using the quadrants there is no way that withholding a reward could ever be any sort of positive punishment.

I think what willem is trying to say is that he doesn't distinguish or recognize positive or negative punishment as separate entities because both result in a negative consequence (and hopefully subsequent extinction of behaviour)

thanks for giving my layman thoughts some credit :D ...however it is more that I differ between the phase when you start positive reinforcement training rewarding your dog and the later phase when he is used (conditioned) to it. It is the later phase, where he is used to this kind of training and expects earning the reward that I believe the current terms are limited, actually the subject matter experts seem to ignore the difference. Refusing the treat / reward at the start: I wouldn't call it punishment at all as I believe that it doesn't have a big impact on the dog's behaviour (also a lot of experts would call it a form of negative punishment)...it is just the reward that affects the behaviour (IMO). The later stage is totally different - the dog is now used or even addicted to the reward and refusing the treat becomes a very powerful form of punishment; holding the reward back will trigger now negative feelings / perhaps frustration ...and I'm the one who triggers it, I'm the one who gives / adds frustration to the dog...so yes, IMO punishment, but it doesn't fit smoothly in the existing quadrants.

The terminology doesn't relate to you, it relates to the reward/ punishment.

The punishment is removed (negative reinforcement)

The punishment is given (positive punishment)

A reward is given (positive reinforcement)

A reward is removed or withheld (negative punishment)

Yes, it's true that you control the punishments and rewards but the dog doesn't think "that B**** didn't give me my treat" they think more in terms of "this behaviour did not result in my getting my treat"

Edited by aussielover
Link to comment
Share on other sites

The quadrants are confusing as they relate to a very specific response more easily explained in a controlled scientific setting (ie Skinner Box). Reinforcement does not equal reward. It means the behaviour is more likely to occur again. Likewise punishment doesn't mean a whack on the head. It means the behaviour is less likely to occur.

Link to comment
Share on other sites

You are not 'adding' anything, so using the quadrants there is no way that withholding a reward could ever be any sort of positive punishment.

I think what willem is trying to say is that he doesn't distinguish or recognize positive or negative punishment as separate entities because both result in a negative consequence (and hopefully subsequent extinction of behaviour)

thanks for giving my layman thoughts some credit :D ...however it is more that I differ between the phase when you start positive reinforcement training rewarding your dog and the later phase when he is used (conditioned) to it. It is the later phase, where he is used to this kind of training and expects earning the reward that I believe the current terms are limited, actually the subject matter experts seem to ignore the difference. Refusing the treat / reward at the start: I wouldn't call it punishment at all as I believe that it doesn't have a big impact on the dog's behaviour (also a lot of experts would call it a form of negative punishment)...it is just the reward that affects the behaviour (IMO). The later stage is totally different - the dog is now used or even addicted to the reward and refusing the treat becomes a very powerful form of punishment; holding the reward back will trigger now negative feelings / perhaps frustration ...and I'm the one who triggers it, I'm the one who gives / adds frustration to the dog...so yes, IMO punishment, but it doesn't fit smoothly in the existing quadrants.

Many domestic dogs IME have rubbish resilience in learning. They quit as soon as something doesn't work the way they thought it would. As someone who spent 3 years watching this in minute detail and examining emotional states and behaviour in different dogs doing the same task, I will say that there is something going on there and I think for some dogs, it's a big deal when they don't get rewarded. I have seen dogs hide under furniture and get weirdly avoidant because they didn't get rewarded, presumably when they expected they would. When you see that kind of thing, you have to start wondering if there is some punishment occurring, although it's hard to separate it out from emotional states.

However, this is not all dogs, thank goodness. It's probably about 20% of the population or so if my experiences are anything to go by. I would also be quite shocked if training style didn't have something to do with it at least in some cases. Even a dog with a little resilience will brush it off and try again. If you are giving more failed trials than successful trials, there's a good chance the dog will walk out on you because your training session is not fun no matter how resilient they are. And some dogs need a very high success rate to keep their interest. But unless we can document some avoidance of behaviours that have not been rewarded, I'm not willing to call no reward a punishment. It's really unlikely to be accurate.

Link to comment
Share on other sites

Willem

The later stage is totally different - the dog is now used or even addicted to the reward and refusing the treat becomes a very powerful form of punishment; holding the reward back will trigger now negative feelings / perhaps frustration ...and I'm the one who triggers it, I'm the one who gives / adds frustration to the dog...so yes, IMO punishment, but it doesn't fit smoothly in the existing quadrants.

there is a really grey area in the middle, where "no reward" might become an unpleasant experierence or aversive. The whole idea is to not encourage an unwanted behaviour by rewarding it. Tho that is not the same as trying to discourage the behaviour.

Depending on how you've trained your dog - your dog may get frustrated and try to leave training, or your dog may get really excited about trying new stuff. This is something you can (I have) move your dog on. Ie a dog that used to be frustrated by something can get excited by it.

If your dog is starting to "expect" a reward... (Susan calls that "egg and bacon boy" the dog is ordering you round like his personal wait staff), you need to change something - usually increasing the difficulty of the task slightly. Slightly longer, or slightly further away, or slightly more distractions eg balloons bouncing by (ok that might be a lot of distraction).

Another one of Susan's trainer helper ideas is about the vending machine vs the pokie machine. When you first start training something new - you are pretty much like a vending machine but as soon as your dog starts to get the hang of the "trick" (task, training objective), you need to start rewarding more like a pokie machine. Payouts need to be less predictable - eg always pats and praise, but not so much treats or tug rewards. The way I do this is aim to reward average or better performance.

For some tasks/tricks - I've been brilliant at detaching from the "expectation of reward" but for other tasks (eg heeling) I've been crap at it. She pokes my hand like a pidgeon pokes the bar in the box and I pay. I don't even notice I'm doing it sometimes until I run out of treats... But our start line stay... I don't need any treats for that. And she loves that game. And she doesn't get upset if she doesn't get the reward, she just gets more eager to try again. Like the pokie player hoping for a jackpot.

Or a kid on the footy field saying "kick it to me"... a good player doesn't leave in a sulk if he doesn't get the ball, he works harder, tries different things to shake his marker and be more available for a kick or get to the ball first. And that's down to good coaching/training. Cos the really little kids - will absolutely leave in a sulk if they don't get the ball.

The scientists have specific definitions for re-inforcement and punishment and operant conditioning vs classical conditioning (which has a new name now like "respondant conditioning" or sometimes "involuntary conditioning") or just Pavlov on your shoulder.

Eg a dog salivating in response to the sight/smell of food is "involuntary" then pre-cuing the arrival of food with a bell - leads to salivation in response to the bell. Takes a lot of repetition to get this response and it will lose effect over time if you stop the training.

Anyway - what a layman thinks the word "punishment" means is not what an animal behavior scientist or a psychologist thinks it means. Eg most people would think that putting a person in jail is punishment but a scientist in the field only thinks it's punishment if the person's unwanted behaviour is reduced when they get released. Ie jail is not "punishment" for some people, in a behavioural scientist's book.

You won't get anywhere arguing about the meaning of these words with those of us who have studied training science.

There is a phenomenon that Susan Garett calls "transfer of value" where the joy of a particular dog's favourite reward - food or play or chasing sheep or sniffing, can be transferred to some task you want the dog to do. There's an anticipation of reward that goes with hearing a clicker or "yes" if you've trained that. They stop working as bridges to rewards if you use them without rewarding the dog.

My dog likes to hear the word "yes" over and over when she's doing weaves - just so she knows that she's doing it right (I think that's what's going on). Otherwise she thinks she's doing something wrong and quits or comes and yells at me. Ie - this isn't how it's supposed to be - but I can repeat the word yes yes all over the agility course and she gets more and more joy and excitement from the running... and only gets the actual reward at the end of the course. She likes running and chasing me anyway. We just try to have more direction to it.

Link to comment
Share on other sites

You won't get anywhere arguing about the meaning of these words with those of us who have studied training science.

ha, ha...I noticed that...but there are also others that call 'refusing a reward' (or 'moving treat away from dog') a punishment: http://www.dogstardaily.com/blogs/isnt-painre-defining-punishment...I guess she is not a real behaviourist :D ...on the other side: it reduces unwanted behaviour so ...anyway, our dog doesn't give a damn how it's called - but I'm pretty sure she prefers positive reinforcements, her treats, pats and workouts to less enjoyable trainings methods :D .

Link to comment
Share on other sites

the "you don't get a treat for that" or "no reward" or even "try again" part of reward based training - is a grey area. It's definitely not +P ie adding an aversive to reduce a behaviour. It might be -P, taking something a dog likes (reward) away to reduce a behaviour. If the unwanted behaviour is reduced - then technically it's punishment.

But if you're rewarding and encouraging an incompatible behaviour with the undesirable one - then it's a re-inforcer of the desirable behaviour, and punisher of the undesirable behaviour (eg training a sit to prevent jumping up - the sit gets rewarded +R and the jumping up gets co-incidentally punished -P no reward for that).

It does depend a bit on the dog how they handle it.

It depends on the dog and their training history how they handle frustration too.

There is a balance between making a reward too easy to earn and too hard... with the right balance my dog loves training. Too easy and she gets bored and easily distracted and too hard and she gets too frustrated to learn or just nicks off. Experienced trainers call her soft, and under aroused / low drive. But if you saw her chasing a cat - you would not call that low drive. High drive is in there somewhere.

Nobody has talked about "escape training" or the -R quadrant much. I really hate that one. I've seen it used really well but your timing and training steps have to be perfect. I can't do it. What that involves is applying something unpleasant like pressure on a slip chain, or prong collar, and releasing it when the dog gets the behaviour right - to encourage (reinforce) that behaviour. You have to make it super easy for the dog to get the behaviour right especially at the start. If you ask too much - you freak the dog out. Mine would just shut down and freeze up and avoid me for days. So I learned real quick not to try to use that quadrant.

Edited by Mrs Rusty Bucket
Link to comment
Share on other sites

BAT and CAT fall under -R, I've had good success with Riley's reactivity with BAT, somewhat modified, no training tools used, just you can move away from the thing (other dog) making you uncomfortable when you show any calm, desirable behaviour.

It is a REALLY tricky quadrant though, and very difficult to explain without a demo.

Link to comment
Share on other sites

I think crazy walking or stopping if the lead goes tight is -P, you're taking something away (-) that the dog wants (reward) to reduce a behaviour (P punishment). If you keep walking when the dog is pulling you're doing +R: adding (+) something the dog likes (reward) by moving in the direction the dog wants, and encouraging (R Reinforcing) the behaviour of pulling.

You might not like the pulling but you're rewarding and encouraging and increasing it if you keep moving in the direction the dog is pulling.

I know what BAT is (Behaviour adjustment training) and LAT (look at that - and reward calm behaviour) is. But not CAT.

With my dog when trying to teach her to be calm around exciting things - I cannot use food rewards but I can use pats and praise. I can use food if she stays calm without putting on the super excited performance... Had a bit of that going on this morning.

Poodle cross at the park. Normally I would stay away but the lady missed her dog doing crap behind her and I wanted to help her find the crap. First couple of times her dog got in evil hound's face - evil hound was quite polite but the third time was just too much - her emotional bucket overflowed and she went off. But I couldn't reward the calm afterwards with food or she back chains. It ended well, evil hound managed to stay calm for a fourth in your face... and the lady picked up the crap. She was one of the ones who is happy to do that. But not so much she pays attention to her dog.

Tho most dogs do prefer to go when owner not looking cos there is less fuss and interruption to the walk that way (+P the interruption and +R sneaky crap?). I try to counterbalance that by obvious rewards and praise when she goes in front of me. But then she gets the +P of the walk interrupted while I put the crap in the nearest bin. Maybe I should give her a reward again after that.

If my dog is doing something I don't like - I have too stop and think about how I'm rewarding that (or how it might be self rewarding ie barking is self rewarding - it feels good).

Link to comment
Share on other sites

wrt crazy walking: I thought there is no reward for the dog involved at all - it is just less nuisance for the dog if she keeps the focus on where I want to go.

If you can anticipate when the dog is losing the focus and wants to pull in her own direction it actually doesn't need much force - I'm pretty sure you can do it in most cases also with much stronger dogs without using a choke or prong collar. When I start a sequence of crazy walking it's changing direction every few seconds or even every second if required so the dog has no chance to orientate finding her own direction she wants to pull...a few changes and she is just busy trying to figure out where I want to go next....you have to be quick so...then we do some meters of normal walking and when I feel too much tension in the leash again I start the next sequence.

Even my wife says it is now a pleasure to walk her - before she used always the harness with the leash attached at the front, now a normal collar does it for her too.

...following the right direction is the 'escape'...no rewards hence -R from me, but I'm not a behaviourist :D

Link to comment
Share on other sites

The crazy walking/doodling/random turns by handler, with no notice to dog, is even more powerful if followed up with a marker .. Yes or other normal verbal marker or click the moment the dog includes the handler in the picture .. i.e. probably still at end of lead, but head turn back to handler, and/or lead slackens. Dog is then rewarded preferably with food and words .. I usually say something like .."You noticed I was here .. that's clever" close to the handler, walk resumes .. rinse and repeat. We find in teaching people at dog club that it doesn't take dog and handler long to make the association. Handler timing is important, but coached through it, most get it pretty quickly. No major force involved .. handler just changes direction and keeps walking in new direction (we say to people .. as if you've forgotten something and have to go check).

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share


×
×
  • Create New...