tess1
No Longer Posts on the DG
Posts: 228
|
Post by tess1 on Apr 9, 2009 20:22:47 GMT 1
Do you use the clicker when doing every day handling.
Hi Dagbecian
It sounds as though you are having excellent success with your ploy to use food to help your horse understand about coming to the yard and tying up. I have found that lots of horses (and other animals) can be motivated by food – sometimes it just takes a bit of time to find out what type of food they prefer, that’s all! There is a great website called stalecheerios, - so called because that’s what the website owner uses to clicker train her dog!!
Yes, I always use the clicker (tongue click, like Stybba) to teach new behaviours, or improve or reinforce existing ones. But I do separate my horses to have individual clicker sessions with them – I don’t walk around with my treatbag on all the time and reward every behaviour I ask for – mainly because my horses live in a herd, and I do not feel that I am competent enough to click and reward one horse for a specific behaviour while the others are all around, and also because, sadly, there are only a certain number of hours in the day, and some of my horses are clicker trained and some, the old/retired ones aren’t – so it would confuse them as to why I am dishing treats out and they aren’t getting any.
The clicker trained horses don’t get confused when I ask for a behaviour like stepping back without rewarding them, partly because I have those behaviours on what is known as a variable schedule of reinforcement (which means they have learnt they don’t get treated every time) , and also because I have deliberately paired ‘good’ and/or a quick wither scratch or neck stroke with food so many times during training that those kinds of rewards (the stroke or scratch) have a real value for my horses now – so I can get away without clicking and treating all the time. I know there was a discussion on here a little while ago about clicking and not rewarding – I must admit I don’t do that – if I click I treat, but I have made sure that there are other rewards available for my horses that don’t involve food, so I can ‘thank’ them for their good behaviour, even when they are in the herd. But this is not a get-out clause for not using food – food is the primary motivator for my horses in their training.
Now I have blabbed on – as I tend to with clickery stuff – but I guess the answer to your question about using clicker training on a daily basis is ‘no’, not obviously, but ‘yes’ in all my training sessions and ‘yes’ in that the behaviours that aren’t clicked and treated outside of training sessions have all been taught/retaught through the clicker, and regularly get reinforced during clicker sessions!! And I do vary where I train (school, barn, yard, fields, outside on the hill, stable, and frequently with horses just the other side of the fence – so my horses don’t learn – ‘go back in the school, gets you food, go back anywhere else and there’s no food coming’. And if you have read to the end, and followed my random chain of thought – you clearly have not consumed enough wine for this time of the evening!!
Tess x
|
|
|
Post by emmaandstar on Apr 10, 2009 9:00:06 GMT 1
I was really interested in this bit as I have always tried to associate 'good girl' with the treats ever since I started with the hairy one, hoping that one day the verbal praise would be reward in itself, unfortunately madam still looks for a treat when I say it. Maybe I just need to perservere with it. I have to confess I'm a bit like the girl in the youtube clips, squealy and excited when she does good work, that comes from years of working with dogs!
|
|
|
Post by stybba on Apr 10, 2009 11:24:26 GMT 1
I agree Emma, this is an interesting part of the topic. You see I have totally avoided associating physical contact with food, as I guess I use physical contact as an intermediate bridge. So for example in my schooling session this morning, when asking for travers, during our first couple of attempts, I was using a lot of 'good girl' and a stroke on the withers. Then when we managed a particularly good sequence of steps in travers, I clicked and treated. I do feel that the stroking and verbal encouragement had some value, even though it has never been associated with food. Not saying my way is better at all, its just how I prefer to do it. I guess I just believe that the affectionate physical contact I have with my horses (stroking, scratching etc.) has an intrinsic value. Although perhaps I'm flattering myself. I Dunno.
|
|
|
Post by emmaandstar on Apr 10, 2009 12:42:32 GMT 1
Yes, I think stroking and physical contact has a big part to play in your relationship with your horse, it all strengthens the bond. The problem with Star, is she's like a person that can't quite allow themselves to be loved. Which is no surprise given she was just left in a field most of the time and only really handled during school holidays. Food has really been my way in with her, I hope over time she will enjoy the strokes but I may have to accept she won't. There are a couple of scratchy spots she does like me to touch, so I'm working from there. BTW I was serious about William, he is one clever pone and absolute dream to clicker train, very quick to learn. I will have him on a see-saw in no time!
|
|
tess1
No Longer Posts on the DG
Posts: 228
|
Post by tess1 on Apr 11, 2009 9:38:00 GMT 1
Hi Emma I’ve taken the following from Kay Laurence’s book (her website is Learning About Dogs). The book is Clicker Training: The perfect foundation, levels 1 and 2 page 107 – 108 (I hope it is ok for me to post it now that I have referenced it) I go to Kay’s for dog clicker training, and I have to say I have found her courses to be fantastic, and has helped improve my skills enormously with the horses as well – the mechanical skills, observation, timing, ‘the science’ if you like is the same, even though obviously we are doing different things with different species. I think that what she says here is just as relevant to clicker training horses as it is to dogs. Here’s what she says about varying reinforcers and working towards the time when we don’t need to click and reinforce every behaviour. “Through all the practice we are building an expectation of one click and one reward. One day you will have an expectation of not needing to use the clicker or the food. If you remove either and: 1 the behaviour loses its quality 2 hesitation appears 3 the behaviour does not happen at all Then you have removed or lowered the reinforcer too soon. If the dog was expecting chicken and you paid with a breath of hot air “you good dog, good job” and ANY of the above occurred, your hot air was a punisher. This may not be what we consider a punishment, but it must be measured in the evidence of how the behaviour was carried out. Assess that the dog is in their comfort zone, and not under stress or anxiety that may cause the loss in response. (What she is talking about here is negative punishment – with holding the treat from the animal – and the scientific/behavioural definition of punishment as the addition of something that will stop a behaviour, or decrease its strength) Back to Kay: Do not underestimate how many reinforcers a behaviour needs to become fixed. If you need an instant recall from your dog in any situation, then plan to take treats along in all those situations. (I think the first part of the above, about the amount of reinforcers needed, is really important. I have just finished an on-line virtual class with her on reinforcement, and she kept stressing that novice trainers don’t reinforce enough, and that getting the number and strength of reinforcer correct was vital for behaviours that we need to be solid. After following some of the exercises in that course I saw a huge change in the behaviour of my dogs – who were already being clicker trained! She also gave me a really good tip (you’ll find it referred to as the Premack principle – rewarding one behaviour by allowing an animal to do another, preferred behaviour) – basically my horses love their clicker sessions so much I could never get them back into their barn or field at the end – unless I used more negative reinforcement than I was happy with (actually it was positive punishment – bad news mixed with clicker work!! Thereby lies the path to poisoned cues). This was really souring our sessions – Kay told me to take the horse back to the barn, and then to take him straight back out again and do some more clicking – then back in – then back out for more training – the reward for going willingly back to the barn was to be allowed to do more training. First off, poor old Patch didn’t realise this was what I intended and we did have our usual tussle – but within three or four tries (in one session) he was moving more willingly to the barn. Of course, there comes a point he gets left there (with a jackpot of food) but as he gets taken back out more times than he gets left and the session ends, he doesn’t see going back to his barn as such a bad idea. I would add he is not protesting about going into a stable – his barn is big and open, he lives there with six other horses, including his pair bond who he has lived with since they were both very young, and they can leave it whenever they like to roam over about twenty acres – so plenty of reinforcers available there, but I guess that’s the power of clicker training!! (back to Kay) We can begin to use different markers for established behaviours that have been through their strengthening process (Tess – loads of repetitions, using food). You can make eye contact as a marker (Tess – eye contact is one of the first things that gets reinforced with the dogs – I don’t do this with the horses) say the dog’s name, and a word of impressiveness ‘cool’ – the dog will soon learn these sounds. If you are expecting to perform high quality behaviours in a performance or reinforce-less situation, then plan to spend time returning the reinforcers to that behaviour as soon as possible. A behaviour will only tolerate so much lack of reinforcement, unless it is self-rewarding. These are usually behaviours with 1000s of treats, or instinctive behaviours. So that’s Kay on working towards less clicking and treating. She has recently produced a book all about how to use reinforcers during training (with dogs) and gave a talk at the ORCA conference on positive animal training. You can read notes that someone took from the conference here stalecheerios.com/blog/2009/03/orca-part-6-kay-laurence/Remember there are ways to train using less food – but they revolve around things like building duration and back chaining. The way I introduced the idea of not treating every behaviour to my horses was to do the normal routine of cue a behaviour/click/reinforce – cue/click/reinforce – etc – then cue/say good/reinforce – etc, keep mixing it up – then cue/say good/NO reinforcer – and then move straight on to cue/say good/reinforce – then gradually build it up to two cues of non-reinforced behaviour and go back to reinforcing the next few tries, and so on …. – so the idea is introduced gradually. Hope this helps.
|
|
|
Post by emmaandstar on Apr 11, 2009 12:09:37 GMT 1
Thanks very much Tess, I'll take time to go through this and have a good look at it. I need to get to grips more with the behavioural science bit, will have to look for a course to do. Did mention to OH about me doing an MSc in equine behaviour, but finances won't allow at the moment
|
|
tess1
No Longer Posts on the DG
Posts: 228
|
Post by tess1 on Apr 11, 2009 22:41:18 GMT 1
Yes, I agree with you both, I think the bond/relationship we develop with our horses also impacts enormously on the training - one of the little exercises we had to do on the reinforcement course was change the way we delivered the food to the dogs - just tossing it on the floor, or handing it to he dogs in an off-hand manner, as opposed to giving the dogs attention/eye contact when we give the treat. Kay told us to watch what would happen to the behaviour - sure enough I got a few surprised looks, quickly followed by hesitations - I stopped that exercise pretty quickly - there's obviously a lot more going on than just dishing out food to the dogs/horses.
I do think the physcial contact with horses is so important - I was thinking about this today - had a lovely day down in the fields with the horses, just brushing them all over in turn while they were wandering around grazing - not a click or treat in sight but they were all obviously really enjoying the contact and I didn't need to put a headcollar on anyone.
Yup, agree with you Stybba, about your stroking and verbal praise being valuable to Tilly, and acting as an intermediate bridge which probably gives her a lot of information. I think there are so many parts to this training - the bond/relationship; the communication - be it verbal, physical or through a clicker, and the science behind it all - it's just such a fascinating mix - I love it!!
Emma, I'm sorry my posts get wordy sometimes and difficult to follow - I hope you can get some sense out of my ramblings occasionally!! It helps me lots to write things down though, and also to exchange ideas with other people.
Tess x
|
|
|
Post by stybba on Apr 12, 2009 8:00:45 GMT 1
Right, here's another little issue to throw into the mix. As you know I am teaching Tommy to ground tie.
During yesterday's session, Tommy started exhibiting some mildly aggressive behaviour. Oviously i am building up the duration of my requests to ask him to stand still. I am trying to keep the steps tiny, and am offering lots of intermediate bridges in the form of verbal encouragement (not pats, as I have deliberately moved a little distance away from him for this exercise).
Now Tommy has learned that mugging, ground pawing etc. does not get rrewarded, it gets ignored. However, he ha started 'making himself big' and flattening his ears and making faces at me. i.e. he is piling on the visual pressure.
My response to this was to stand my ground and do the same in return (the visual pressure, not the faces!) by holding up my hands and clapping them. This worked in the sense that he shifted his body weight back (not stepped back), and looked suitably chastened.
It only happened once more before the end of the session, with the same result, and we were able to end the session on a really good note.
Coincidentally, I happened to be at the part of ben Hart's book where he mentions this, and i think he suggests that negative reinforcement + clicker is not a good mix.
This got me wondering if I am being unfair on Tommy by......
a) asking him to do something which encourages him to produce this behaviour in the first place.
or if it is a good thing because.......
b) This is exactly the behaviour that I am trying to encourage him not to do, so it needs to be brought up and tackled....
or....
c) yes bring it up, and tackle it, but with the clicker (i.e. train a different behaviour in its place)
But this is a level of visual pressure that I am not prepared to tolerate from him, even long enough to train something else in its place, which I guess is what I am trying to achieve with the ground tying.
There is no good reason for this aggression, he was not hungry etc. He was just frustrated. This is him all over, he is a very big Haflinger who has used his body and size to intimidate his previous owners, and has become spoiled and wilfull in the process. However, he is not a 'dangerous' horse. I do not think he would ever follow through his threats, but at the same time he must learn....and quickly.... that these threats will not be tolerated, much less pandered to.
So, thoughts anyone?
|
|
|
Post by emmaandstar on Apr 12, 2009 11:24:45 GMT 1
Yes, I have had this agressive behaviour during sessions as well. ground tying was particularly frustrating for her, stillness is not a Star attribute! With her I just had to ignore it, it never got rewarded and I would just move onto a different exercise like head lowering, that's a good one as they can't be big and intimidating with nose on the ground! And just gradually intersperse bits of the standing in with that so it didn't get too boring.
She then started it outside of sessions and I have to admit worried me more than once, so I kept a rattle bottle in my pocket. I only used it once, when I asked her to back off and she didn't. It is something I have to constantly be aware of, it disappears for a bit then she'll have a go again. But it's never rewarded, I then start making her back up away from me.
Does Ben Hart say not to use -re with clicker?? I know definitely he says not to use punishment in any way, but I was always of the understanding gentle -re rewarded with +re and release of pressure can make the behaviour twice as strong. You need to be absolutely spot on with your timing though.
Tess your posts aren't wordy they're very informative, I just have the internet police tutting I'm on here too long some days!
|
|
|
Post by stybba on Apr 12, 2009 11:53:56 GMT 1
Does Ben Hart say not to use -re with clicker?? I know definitely he says not to use punishment in any way, but I was always of the understanding gentle -re rewarded with +re and release of pressure can make the behaviour twice as strong. This is the bit that concerns me...... "If a trainer uses clicker training for problems or new behaviours, yet uses punishment or heavily negative reinforcement based methods during training for groundwork or riding, the animal may have to deal with a conflict between wanting to offer behaviour and being worried by the possible consequences of that behaviour." Now I wouldn't describe my actions as being 'heavily negative reinforcement based', but it was negative reinforcement. Hart also talks about teaching patience, by waiting for a few seconds between the click and the treat. This is no problem for Tommy, once he has heard the click, he will wait patiently for the treat that he knows will come. Tommy also knows perfectly well not to mug, this wasn't mugging, this was intimidation based on frustration. This frustration would come with head lowering, cone touching, or whatever. Now I am happy with what I did, and will do it again if needs be, but I am just hoping that it won't cause any major conflict with the training itself.
|
|
|
Post by mandal on Apr 12, 2009 11:54:30 GMT 1
What an interesting thread, thanks guys.
|
|
tess1
No Longer Posts on the DG
Posts: 228
|
Post by tess1 on Apr 12, 2009 12:15:12 GMT 1
Hi Stybba
Thinking right off the top of my head now.....
I don't think you used negative reinforcement with Tommy by clapping at him - you used positive punishment - in that it stopped his behaviour (punishers decrease, reinforcers increase) (Punishment used in the scientific sense here - not in any way a criticism of your training).
If you have managed to extinguish mugging and pawing through negative punishment, I don't see this as being any different - safety first - work behind a barrier if you have any doubt - also means you can walk away and give him a time out if he really crosses the line with his threats.
I think emotional control is vital in any horse - so this is definitely something to work through. But, as you have already said, he does find this behaviour tough - and I wasn't kidding when I said one of the hardest behaviours to teach is 'do nothing'! So, although you are taking it really slowly, maybe you need to break it down even more for him. So, for example, are you increasing duration and distance together (even if only incrementally) - remember they are two different criteria, work on one, then the other. Make sure you look at the whole picture when you click - not just standing still, but standing still with ears forward (more work on happy faces, maybe?) and with a relaxed body posture. Agree with Emma here, teaching him to drop his head would be a great thing - although I'm not completely convinced of the science behind it, it does seem to be that head lowering is a calm down cue.
re: negative reinforcement, Alex Kurland bases all her work on negative reinforcement, and I think it is practically impossible to train horses without it. So long as we don't go too far down the spectrum into positive punishment, I think it is fine - but using clicker means you should never need to get to that stage. How can we ride our horses unless we use negative reinforcement? Does Ben Hart transfer the clicker to ridden work, or even in hand work? In my understanding, positive punishment is the thing to avoid at all costs - which is why negative reinforcement needs to be used with care.
There is a great (but deep, sorry guys...) article here - equineclickertraining.com - scroll down to 'the four quadrants of operant conditioning and how they apply to clicker training'.
Ok - just one last thing jumped into my head - then I must go and plant my onions - take care that you are not teaching a behaviour chain - if he stand there all big with his ears back, and suddenly pops his ears forwards - don't click him then - you risk reinforcing the unwanted behaviour along with the wanted behaviour. Quick example to try to make this clearer - dog jumps up - ignore it - dog puts four paws on the floor - trainer clicks - big risk of training a sequence - jump up/jump down to get the reinforcer - does that make sense? Important thing is to not get the unwanted behaivour in the first place - by training incompatible behaviours, taking things slower, using negative punishment effectively, whatever, rather than risk trying to select for one good behaviour when he is throwing a few unwanted things at you as well.
Sorry guys - it's long again!!!
happy clicking Tess
|
|
|
Post by stybba on Apr 12, 2009 12:21:55 GMT 1
Actually Emma, sorry, this is the bit that confused me..... "On balance, the click has to be more rewarding and motivating to the equine than the aversive stimulus is uncomfortable or frightening......................Clicker training is very effective and very motivating for the animal, so there is no need to try to improve it by adding a negative or aversive stimulus because witholding the click is sufficient." My problem is that in this case, witholding the click was not only not sufficient, it led to the frustration that caused the behaviour. I felt that the resulting behaviour was potentially too serious to be dealt with in any other way than by using an 'aversive method' Just to reiterate what I said before, sensing Tommy's frustration prior to this happening, I had gone back to taking tiny tiny steps to reinforce the duration that we had built up before, so it wasn't like I had kept him hanging about for an unreasonable length of time, quite the contrary, I had made it just a little easier for him to be rewarded. Or maybe it was just the wrong time, on the wrong day to be doing what I was doing? But one can only come to that conclusion with the benefit of hindsight.
|
|
|
Post by mandal on Apr 12, 2009 12:24:09 GMT 1
Ok - just one last thing jumped into my head - then I must go and plant my onions - take care that you are not teaching a behaviour chain - if he stand there all big with his ears back, and suddenly pops his ears forwards - don't click him then - you risk reinforcing the unwanted behaviour along with the wanted behaviour. Quick example to try to make this clearer - dog jumps up - ignore it - dog puts four paws on the floor - trainer clicks - big risk of training a sequence - jump up/jump down to get the reinforcer - does that make sense? This is my biggest worry about me doing clicker work!
|
|
|
Post by stybba on Apr 12, 2009 12:30:11 GMT 1
Thanks Tess, cross posted there! I was nodding vigorously through most of that, (only read it hurredly though as the 'internet policeman' is out fiddling with the tractor so I have to be quick!). I am very careful not to click when Tommy is 'being big' etc. so I have never ever rewarded him for this behaviour, in fact yesterday, and only on those two instances were the only times he has ever done it. I do teach head lowering and cone touching etc. and he is good at those things, but as I say, this frustration is bubbling under the surface, but until yesterday has never been quite so explicit. I know Alex Kurland teaches her horses to recognise 'No' as a signal that an offered behaviour is undesireable, so I am considering using that. Yes Ben Hart does talk about riding with the clicker, but I haven't got to that chapter yet! The 'internet police' are back in the room, so I have to go. Thank you so much for your input Tess, lots to think about there.
|
|