On positive reinforcement and clicker training

[glenatron]

One of the most endless and repetative discussions that turns up from time to time among Natural Horsemanship types is the conversation about Positive Reinforcement, which is an entirely reward-based way of training animals used very effectively by a lot of animal trainers. I think it's a great way of training animals in general, but not a particularly useful way of training horses specifically. In a reply to a discussion elsewhere I finally managed to put my thoughts on why it's not so great into a coherent order so I thought I'd crosspost here in case I my future self wants to know what my opinion is...

I think maybe if I could achieve something with a clicker that I wanted to achieve and couldn't do with my current approach it might appeal more to me, but when I watch the riders I aspire to be like they're doing fine without the extra stuff and it just seems a little unnecessary really.

It seems like it's very much a tool for a job, but I've never seen it used for a job that wasn't pretty easy to do in another fashion. I mean yes, you can use it effectively for helping a horse to pick up their feet or load into a horsebox or whatever, but actually a lot of people can do all those things anyway.

Once things do get complicated I'm not sure how it helps. For example, one thing I'm working on at the moment is asking for flexion and engagement of the hindquarters to build up a Travers type movement. When I'm working on that I need to be using the reins in both hands to ask for the flexion I want, certainly at first, so I wouldn't really have a hand free to operate a clicker ( sorry bridging folks, I'm not planning on shouting "sexsexsex" at my horse, I get enough funny looks for the zebra-print hat cover ) even if had enough braincells to manage asking for flexion with the rein, asking for engagement with my leg and operating a clicker when I felt the thing I was looking for. And then I'm having to fish around for treats so I lose all the flow of the work where I might have been able to turn it into something else useful. And also I have a "no treats by hand" clause in Zorro's loan agreement so even if I could do all that, my horse would get taken away

I don't find it disagreeable as a training concept and I'm sure it's great for dolphins and dogs and whathaveyou and probably essential for some trick-training type work, but when we start asking a bit more of ourselves in our riding I think we need to be in the moment focussing on the job in hand, rather than reflecting on our shaping plans and I can't see how this approach can avoid creating a layer of mental beaurocracy that could not help but create a layer of remove. I'm thinking that when you watch a really great rider working with their horse the rider's muscle memory and the horse's muscle memory are sort of connected so they flow into each other to a degree. I just can't see where a click-and-treat would fit into that. I'd love to see a proper behaviourist do a badminton run to match Olly Townend, outride William Funnell around Hickstead or turn in a dressage test to match that Andres Helgstrand one, but I'd also be surprised.

I'm not suggesting I can do any of that stuff either , I'm not there yet by any means- indeed I'm much further from there than most people on this forum - but as I begin to glimpse more technical riding the more I question where and whether a +R oriented training approach can be used to enable it.

This is all a bit theoretical for me, obviously, I've only really used a clicker for dog training ( the only time I tried it with my first pony he was scared of the "click" ) but I guess my feeling is that there comes a point where it has to stop coming from the tools and come from you instead and that some tools flow into that better than others.

Comments

[dancing-crow.livejournal.com]

I see completely what you mean. I was using it for some ground work with the mares I work with because I needed a new way to come at the problem.

Kaboose was worried about lumpy but harmless objects, classic horse behaviour but it was making me crazy at how diffuse it was. I would click and reward her for approaching something scary and putting her nose on it. It decreased some of her fears, and now she is much braver about marching up to the (horrifying) mailboxes and garbage cans we meet out on the road.

Penny Would Not Stand, for mounting mostly but for other things either. So she got a lot of clicking for simply standing while I jumped up and down at the mounting block and climbed on and off her. Not that I couldn't have done it without the clicker, but it made it easier.

Ruby needed reassurance for hosing off, fly spray, etc. and the clicker gave me a very simple way to encourage her to hold still when verbal and physical soothing were not working so well.

On the other hand, I find myself in complete agreement about being unable to use it in a coordinated way from their backs, especially to work on dressage issues that are consuming me at the moment.

I think of it as a tool, and a handy one for helping them get past spooky issues. And apparently I don't use it beyond that. huh.

[joycemocha.livejournal.com]

I don't use clickers at all. However, I use food treats on the ground, as well as verbal praise and scratches. The cluck tends to be my standard vocal cue.

If you ever read Mugwump Chronicles, she has some interesting commentary on the whole clicker thing. She's not a NH person (neither am I). However, I do focus on consistency, making the right thing easy and the wrong thing hard, and I think of what I learned in a clinic with Jean-Claude Racinet--he was working on a higher-level movement (tempi changes? something like that) and he would say "He does not understand" when he'd cue and the horse didn't do what he wanted. Then he'd patiently retry it.

I'm working on countercanter with Mocha right now (in Western tack, she gets pretty jazzed up and it's easier to sit), and we're having to overcome a lot of previous conditioning about taking the inside lead. Now I'm really emphasizing "listen to my seat, listen to my leg" (well, mostly leg at this point) and sometimes she finds it hard. But she visibly relaxes when I tell her "good girl!" and drop the rein on her neck.

[penella22.livejournal.com]

Two thoughts;

I've used treats with Thea in some situations simply because it got her to chew, which caused to relax enough to look around and realize she wasn't dying. Also, that new experiences could be fun (not scary). I don't actually have a clicker thingy though; I just used apple treats and no click or bridging sound at all. It's probably not clicker training in fact. But I wanted to point out that when a horse is dissociative in that way, and will be more present for food, I'm okay with using that to build confidence.

I disagree with using food routinely in training though. Because I think ultimately if you're working with your horse right, then the way you both feel while doing the work is reward enough in itself. If the horse doesn't feel that way, then maybe that's feedback that you have something more to work on.

I DON'T use clicker training with Sage, or really even treats at all now because he's so food oriented anyways, and it can be hard to get his mind back on what we're doing. He just becomes a "food? food? for me? Now? how bout now?" monster and loses the whole train of thought and will start trying a million and one things in a frenzied way in an attempt to get the treat.

But I am using treats for Sage around the mounting block right now, to overcome his negative associations with it in the past.

Clicker training, btw, doesn't have to involve food at all. My understanding is that you could do any kind of positive reinforcer, from patting your horse's neck to itching their bugbites, to letting them gallop to letting them stop. ANYTHING that is a reward to them can be used, right?

To leap onto another tangent...what I dislike most about clicker training is how the whole thing makes me feel like a parent paying their child to get good grades. I'd rather my kid got good grades for the rewards inherently involved in good grades. I dislike any kind of training that feels like bribery, and much prefer training that says "here are your choices..."

I think clicker training CAN be like that. But in practice, most of the time, it's done with loads of treats and far less meaningful questions.

[glenatron.livejournal.com]

Mugwump is as NH as I am by this point- she learned with Ray Hunt and a few of the other people who influence the people I learn with - I like a lot of what she has to say. I'd do some things differently to her recommendation but her general approach is very sound indeed. I haven't seen those posts, but I'll look them out.

The timing and release things, those are a part of riding already and it seems more logical to me to build up around those rather than putting something in you'll need to take out later.

[glenatron.livejournal.com]

The thing about non-food rewards is that you have to have a horse that wants them- most horses want food most of the time, whereas sometimes they are itchy in one place, sometimes another- Zorro's strongest non-food motivator is scratching his arse-cleavage ( to the point he will practically sit on me in the stable sometimes ) and I'm just not sure how I could use that from the saddle and I'm pretty sure I don't want to.

I don't mind using food rewards at all - it's how sleepsy_mouse has trained Small to be perfect at the mounting block - he goes and waits there and doesn't go until he's had a snackie - but the people who bother me are the ones who will only use positive reinforcement for training. They tend to get very religious about it, often very rude to anyone who doesn't do things their way ( for people who need to be positive there is a whole lot of negativity they carry around with them ) and generally irritate me. The post on The Dog Whisperer from this page (http://www.markrashid.com/blog_archive.html) is quite relevant.

[glenatron.livejournal.com]

That's how I see it too- it's a tool you can use for some discrete jobs and it works fine for that. I have other tools to do the same job, but they're no better or worse really, they just fit in with everything else I do.

The people who bother me are the ones who tell me that to use anything other than Positive Reinforcement is basically the same as cruelty and how you can solve any problem through pure +R training. I think they are wrong and that they are -at best- unlikely to ever be able to get much done with their horses. You can't manage life forever so that a horse is never afraid, not if that horse is ever going to have a job to do, but that approach basically can't work in that environment. I heard of a clicker training demo where they got some nice work but the demonstrator nearly got herself flattened at the end because her horse was scared by the applause and they had nothing in place to handle that.

It all gets a bit One True Way for me.

[fleefloodle.livejournal.com]

Personally, I couldn't see clicker training working with either of ours. Willow tends to have a meltdown about noises she doesn't understand (ie phones, speakers etc) and I suspect a clicker would have a similar effect. Ranger is too food orientated, and any training method involving treats (even if only in the early stages) would likely result in a very heavy, very strong pony attempting to obtain food from you in a rather indelicate way.

However, generally I'd agree that positive reinforcment is the way forward. Willow is terribly people orientated, so for her a pat/scratch/vocal praise is a reward to her so she knows she's done well. An easy example is if we've been having trouble with a particular jump, once we get over it I'll praise her vocally, give her a quick stroke on the neck and her ears will go up and her whole body language will become more positive (of course this may also be because she's pleased with herself for getting over it...).

Positive reinforcement is fantastic, and in the long run much more effective than negative I would say, but a lot of people don't realise that it's a lot more difficult giving a horse food rewards than it is to say a dog. Dog's are generally smaller if nothing else. What this all seems to come down to is the fact that a lot of NH users seem to assume that one thing will work for every single animal in the world ever, which most people of course realise is complete nonsense. I'm quite sure clicker training is very effective for some people, but for me and my family's horses, probably not.

[glenatron.livejournal.com]

I think positive reinforcement is pretty handy for animals in general and I certainly use rubs and scratchies as a reward with my horse ( but also just when we're sharing some undemanding time- a training regime that suggested one should only use those as training aids would not work for me at all ) but I see it as a nice thing to do, but not core to anything.

Negative reinforcement is how horses communicate among themselves and it's the approach that I use in most training- any time you apply your leg to ask the horse forward, put pressure on the rein to ask for a bend or anything else along those lines that is negative reinforcement after all - and I honestly can't imagine getting close to the results I am getting if I wasn't using it.

I think the source of that "one thing" idea is that you see a really good trainer and you think that the thing they do must work for every horse and it does work for them with every horse and it's not until you've got a reasonably clear understanding of what they are actually doing that you can see how much they adjust their approach to each different horse.