|
Post by ghostrider on Dec 30, 2010 16:57:04 GMT 1
negative reinforcement means the removal of something - its not always aversive! It might be better understood if the terms were clearer - reward reinforcement (+R) and removal reinforcement (-R). Bx Why would you need to remove something to strengthen a behaviour if it wasn't aversive? 'negative' in the context of 'negative reinforcement' means to 'remove' ''reinforcers' strengthen behaviour. Thus negative reinforcement is the removal of an aversive stimulus to strengthen a behaviour. It does not have to be painful or fightening at all - just 'enough' to prompt a behaviour change. Light pressure is not painful or frightening, as long as it is introduced correctly, but horses can learn a lot through the application and removal of light pressure - that's negative reinforcement (they learn a lot more and a lot quicker imo, if you add in salient positive reinforcement, and enjoy it a lot more as well - but that's what we've been discussing all the way through this thread). It's removal is 'relief' (not a reward). A 'neutral' stimulus would not prompt a behaviour change ... therefore would be pointless. Negative reinforcement is NOT the same as positive punishment - although it can escalate into positive punishment. But saying one uses negative reinforcement is nowhere near the same as saying one uses positive punishment. Sorry, but I just don't see the need to re-write learning theory when it has perfectly good terminology which can explain what we are doing when we are training our horses, and why it works (or doesn't) Better to know exactly and take steps to minimise our use of even mild aversives (especially if we would rather think we didn't use them at all) then just use different words to make it sound .... better
|
|
|
Post by mandal on Dec 30, 2010 18:07:20 GMT 1
Sorry, I'm going to be really thick now, what amused you on the cover of the Equid Ethogram? Historical discussion on here about dominance and affiliation theories in horse society. Ethology seems to be viwed differently between myself and a couple of 'learned' posters and the photo of a horse in erm, anguish/extreme emotion/?defensive/non affiliative pose on the cover of an ethogramm amused me that was all. I have a strange sense of humour. I obtained a copy of the book shortly after the discussion... Hartslet there is a thread in the Hall of Fame on negative reinforcement as well as another on CT. I understand the aversive in the theory but like you struggle with all 'asks' being aversive even though I know they must be. I will keep quiet on that subject though to spare regulars.
|
|
|
Post by ghostrider on Dec 30, 2010 18:11:37 GMT 1
Alex K has a saying - "sooner or later, everyone sits down". That's because sooner or later standing up becomes a chore. Waiting out a behaviour can certainly create behaviour change - and I agree entirely that allowing horses to express their feelings is important and worthwhile. And sooner or later the horse will offer another behaviour when he realises his current behaviour is getting him nowhere. Letting them 'work through it' is definitely a great tactic.
I'm slightly confused by your post to be honest though - are you trying to say that because you don't try to "correct" your horse if he's worried that you don't use negative reinforcement ... ? Just holding out and waiting for a behaviour change doesn't exclude the use of negative reinforcement in other aspects of training and handling .... waiting out an unwanted behaviour wouldn't get you from a-b without some other form of directional cues - do you never use (light) pressure on a leadrope or use your body-language to direct your horse somewhere? Do you ride? Do you never ask your horse to step back or step over? And how have you taught those behaviours if not through the use of negative reinforcement? If the horse is being restrained on a leadrope or bridle when he is moving about or showing anxiety how can you avoid negative reinforcement? Perhaps even positive punishment? As soon as we sit on horses, or put a headcollar and leadrope on them we've signed up to negative reinforcement - no way to avoid it. Even loose schooling relies on negative reinforcement.
I do think it's important to stress that people can be 'kind' trainers and still use negative reinforcement - in fact I think if people knew how to use negative reinforcement correctly instead of shying away from it or getting over-emotional about it, horses would be a lot better off generally. I'm afraid it's a bit of a bugbear of mine that people try to deny the use of negative reinforcement when handling and training horses so I'm sorry to be a bit of a pain on this one :-)
|
|
|
Post by ghostrider on Dec 30, 2010 18:21:46 GMT 1
Sorry, I'm going to be really thick now, what amused you on the cover of the Equid Ethogram? Historical discussion on here about dominance and affiliation theories in horse society. Ethology seems to be viwed differently between myself and a couple of 'learned' posters and the photo of a horse in erm, anguish/extreme emotion/?defensive/non affiliative pose on the cover of an ethogramm amused me that was all. I have a strange sense of humour. I obtained a copy of the book shortly after the discussion... Hartslet there is a thread in the Hall of Fame on negative reinforcement as well as another on CT. I understand the aversive in the theory but like you struggle with all 'asks' being aversive even though I know they must be. I will keep quiet on that subject though to spare regulars. oh, I thought he was having a roll!! ;D lol, I see what you mean if that is the case! The 'asks' are only aversive enough to prompt a change - an itchy foot prompts a good scratch, a draft prompts closing the window, Eastenders prompts finding the remote control (for me) (apologies to anyone who likes Eastenders), an empty glass prompts the search for the bottle of wine ... they're not the end of the world, (well, no wine might be ) just minor irritations that prompt a behaviour change to deal with them
|
|
|
Post by mandal on Dec 30, 2010 18:22:00 GMT 1
I use negative reinforcement all the time. ;D I don't see how we can avoid it. Even if we are only ever free shaping we wouldn't be able to do anything with a horse that it hadn't been trained to do first by FS. The chances of even the best trainer in the world being able to function at such a detailed level and hope in the meantime a situation doesn't crop up where -R is needed is imo impossible. Like CT it is all in the timing and the ability of the handler to be light and able to read every little 'try' to release the stimulus at the right moment to say 'yes that's what I want'. Non free shaped CT then goes on to say, 'yay, now have this lovely reward for doing that'. I hope I've not simplified that to be meaningless or wrong GR. Mta as crossed posted. oh, I thought he was having a roll!! ;D lol, I see what you mean if that is the case! Lol, he probably is now you say that. It's just the way I was 'tuned' at the time.
|
|
|
Post by june on Dec 30, 2010 20:07:17 GMT 1
I found a youtube extract where Alex encouraged the house to lean on its haunches, I understood this to be teaching the horse to use his behind. I think I know the clip you mean. I find that one really uncomfortable to watch as it looks so similar to the laminitis stance. To shift the weight onto the haunches the horse also needs to flex all the joints in the hind end. If you look at the Spanish Riding School horses doing piaffe or pesade then you'll see the difference in the way the joints all flex to take the weight rather than the horses just rocking backwards. I want to incorporate CT into Nancy's school work to make being in the school pleasurable, as we have had some horrible times in there and she associates the school with pain (she had grade 3 ulcers and i was trying to work her through it, resulting in me ending up on the floor), so what I am wanting to achieve is her acceptance of work and wanting to make it fun for her. I use CT quite a lot with ridden work. I use a specific tongue click rather than the clicker itself as it is too difficult to deal with the clicker and the reins at the same time! As ghostrider says the downside with clicker under saddle is that you get a halt after a click. I don't mind that and have found clicker a fantastic way to motivate horses that don't enjoy or see the point of schooling. It makes things really clear to the horse and the click pinpoints the exact moment the horse is offering the desired behaviour, which is what makes it a better training tool than simple praise or scratches. It also focuses the rider on exactly what they are asking and how, and notice when they've got what they wanted, so it trains the trainer too IMO. I don't reward on every click as I don't want her to expect a treat on every click. My deal with the horse is I always reward after a click. When the horse is learning something new I will click every small try but as the behaviour is learned then I only click the best tries and eventually fade out the click altogether or click exceptional tries. I also use the word "good" as a bridge to tell the horse they are on the right lines and to keep going and there will be a click coming.
|
|
|
Post by 2 bays & a grey:D on Dec 30, 2010 20:21:47 GMT 1
June I get exactly where you are coming from with the 'pose' stance, and now you mention it I can see why it looks disturbing . I don't give treats on every click as I am teaching very basic things at the moment, ie target training, and as she is now doing it so well I guess I am filtering out the treats. I am not knowledgeable on CT to extend to anything else at the moment as I don't want to start confusing them. I can't wait to start with new things though as Nancy is loving her new CT games
|
|
|
Post by june on Dec 30, 2010 20:30:19 GMT 1
I'd filter out the clicks rather than the treats. I'd do things like make it more specific, so touch a certain place on the target, click and treat, touch the target for longer, click and treat, touch the target twice, click and treat etc. I'd work on those things separately rather than all at once though, as all at once would be quite confusing to the horse. The general rule is to change one thing at a time. So, maybe go for touching a specific place on the target, click and treat, and when the horse is getting that right 8 out of 10 times, go for extending duration so touch the target and hold the nose there for a fraction longer, click and treat. Once you've got the horse touching the target for a few seconds you can start moving it around so they follow the target, click and treat, etc.
|
|
|
Post by ghostrider on Dec 30, 2010 20:44:07 GMT 1
Of course I give my horses cue's, and yes I do ride. As I have explained though, I don't consider light short cue's to be 'negative reinforcement'. Whether you consider them to be or not, technically speaking, that’s what they are. If my horse doesn't respond to a light cue then I'm not giving the cue correctly, I change tact and try something else. I wouldn't stand there keep giving the same cue (as your hand on shoulder analogy) until it got uncomfortable.How do you change your cue – what else do you do? If I ask my horse to move forward into walk, I am doing just that, asking him to move forward into walk. I'm not making standing a negative experience so he has to move away from it.How are you asking him to walk forward? And if you are using a verbal cue, how did you teach that cue? There have to be cues, communications, asks, requests and relationship building, and on both sides of the partnership. They don't have to be pressured. Agreed. If a horse indicates to me that he has pain in his back by pulling faces when saddled, I don't consider the horse to be using negative reinforcement to condition my behaviour to stop saddling him. I take that as an indication of a problem.Agreed – a horse indicating back pain has nothing to do with negative reinforcement. If my horse is indicating he doesn't understand I have to respond to that, if he's indicating that he's anxious about something, I respond to that.Agreed The idea of pressure and release is just that, pressure and release, not pressure, more pressure, still giving pressure, I'm stilling hanging on with pressure, and release.If you ‘release’ when the horse is not producing the required behavior you have reinforced an unwanted response … and potentially caused confusion for the horse. The horse learns mainly through the release, not the pressure – the timing of the release is crucial to clear communication. The pressure can/should be light, and the try from the horse can be very small – but it is the timing that matters. Obvious I don't actually know you, and I'm new to posting on this forum, but your earlier post indicated that if you don't use enough positive reinforcement you can end up "relying on significant amounts of negative reinforcement - which is one of the things CT'ers try to avoid/minimise."Yes – the majority of behaviours we teach our horses is either taught through positive reinforcement or negative reinforcement, or most likely (hopefully) a combination thereof. Traditionally, in horse training, there has been far more reliance on negative than positive reinforcement during training (along with positive punishment) – but hopefully things are shifting towards more positive/reward based training techniques. I can agree and appreciate that, my point is that whilst I respect what CT does for some people, there is more than one way to skin a cat, if you'll pardon the horrid expression. Yes, many ways to skin a cat, also many roads to Rome and CT is most certainly not for everyone. There are plenty of successful approaches to horse training; classical, conventional/traditional, NH, and within each of those there are many different schools of thought – but they all involve negative reinforcement to a greater or lesser degree. And even if someone decides to use clicker training they have to ‘adapt’ it to their chosen school of training – and so (light) negative reinforcement is inevitable even in clicker training, and definitely when it is used with riding. But that is fine – as long as it’s not mixed with heavy negative reinforcement or positive punishment – that’s when problems may arise, as mentioned earlier in this thread.
|
|
|
Post by Yann on Dec 30, 2010 20:48:57 GMT 1
I maybe need to read the book again but I'm sure Alex K recommended moving on to intermittent treating as part of a variable reward schedule, however this obviously isn't quite the same thing that Ben Hart does. There are seemingly as many different styles with CT as with any other type of training I can see the difficulty with the pose thing too, you clearly do have to be careful what you teach and there are many different ways of creating a braced horse. I do wonder if teaching a horse to go correctly is best done using a bit and at least some negative reinforcement. I've had good experiences using clicker under saddle and found my horse's atttitude to schooling, which was already OK, suddenly got quite a bit better. Little niggling things I hadn't been able to make much progress on, like staying soft in halt and halt walk suddenly got better, as did my brakes, I am now able to go down from canter through trot to halt on a voice command alone a lot of the time, something that would have been unthinkable 6 months ago. She is softer and more attentive. I do seem able to use the clicker and hold the rein without too much trouble.
|
|
|
Post by ghostrider on Dec 30, 2010 20:56:11 GMT 1
Can't remember what AK says on it tbh, but personally I'm with June in gradually reducing the amount of clicking, but treating when you do click. Eventually I fade the clicker altogether and just give praise and scratches and the occasional food reward. It seems to work OK.
I've not seen the video on the 'laminitis stance' - but I have seen a picture in one of her books ... :-( I'm afraid it's another one that set my alarm bells ringing.
|
|
|
Post by ghostrider on Dec 30, 2010 21:34:38 GMT 1
I'm afraid I would like a bit more clarification than simply "I change it" to be able to respond appropriately. You say you follow through with a "slightly different cue" but what cue(s)? I can appreciate different horses and situations require different approaches, but some examples would help clarify what you are saying.
When I teach verbal cues using the clicker I use targeting, negative reinforcement, and free-shaping depending on what I'm teaching and the horse. If it is possible to have the horse free when I'm teaching him, I'll do so. I'm asking you (because I'm genuinely interested) how you teach them without using any negative reinforcement.
If an owner mis-interprets a horse's pain as bad behavior then it is mis-interpretation, not negative reinforcement. The horse is not making a conscious decision to 'condition' the owner, only trying to express his concern in the only way he can.
|
|
|
Post by Yann on Dec 30, 2010 21:50:25 GMT 1
Hartslet, Ghostrider is correct in what they are saying in regard to negative reinforcement. It's a technical term, pure and simple, describing how a given behaviour is reinforced by something being removed. It's not uncommon for people to misunderstand this and give the term emotional connotations it shouldn't have. Any pressure and release, however gentle or tactful is negative reinforcement, it doesn't automatically mean it's bad in any way, or necessarily bad at all.
|
|
|
Post by ghostrider on Dec 30, 2010 22:19:54 GMT 1
yes, that is clear, and perfectly understandable, and yes I do similar myself with youngsters. However, even after training that sort of thing positively I know there is quite likely to come a point when, for some reason, the horse may have a difference of opinion to me, or simply spook or become excited and rush to the end of his lead rope and feel pressure on his head, to me that is negative reinforcement. Even in a flat headcollar as opposed to a pressure halter, and attached to a bungee instead of a leadrope we simply can't avoid these kinds of things happening completely. Likewise when riding, I accept that pressure from a weight shift, a leg aid or a touch on the rein is also negative reinforcement. I don't see these things as 'bad' necessarily - negative reinforcement doesn't mean 'negative' in the bad context - they just are a part of communicating our wishes to horses. I like training horses loose because that does at least minimise the amount of negative reinforcement - and the horse can leave if s/he wants to - which is always valuable information in it's own right.
|
|
|
Post by Yann on Dec 30, 2010 22:24:58 GMT 1
There isn't a 'CT' definition, it's the behaviourist definition that's pretty much universally accepted - it can be found in the back of Kelly's Perfect Manners book. No, clicker isn't the only way to work with low or no pressure by any means. The description you give of teaching a foal to lead is by means of positive reinforcement too. However if at some point the foal has a difference of opinion on where you are both going and you gently correct it using the lead rope, then that would be negative reinforcement. We all probably use a mix of the two in varying amounts, along with a bit of positive punishment now and again. Like I said they're just technical definitions of what's going on, not a moral judgment on that training
|
|