Despite many people using clicker training successfully on the ground with horses, people often feel confused by how how to apply it once on-board. I've tried a bit of ridden clicker in the past so thought I would share my thoughts and experiences. For me it is all down to two issues - exactly what I want to be rewarding and exactly why I want to be rewarding it. How to do it then rather falls into place a bit more, just by using the same principles that we would use for CT on the ground.
So starting with what to reward...
Firstly you can reward in a very general sense. You ride normally with the usual sort of mild -R that includes various cues and releases (if you're doing it well, anyhow) and then click and treat periodically if you find the general effect pleasing. There are plenty of people who do this and claim that the clicker adds something to the training. I suspect that it does in some cases but only in the sense that the horse gets a bit of a break and a bit of nice experience and so gets a generally positive association with the session. That's all a good thing but it doesn't aid the learning very much because the click isn't actually being paired with a specific behaviour. The potential pit-fall is that some horses will get very worried about trying to work out what behaviour they are being rewarded for and will lose that sense of positive association. Obviously if the pressure used is perceived by the horse to be quite aversive then that too will undermine the limited positive benefits of the rewards. For me, this approach isn't actually clicker training because there are crucial elements missing - learning, choice, a genuine tapping of the brain circuits that govern learning via positive reinforcement and release significant amounts of dopamine... So it's not something I ever do and can'tt see that changing... (But never say never and all that!)
So alternatively we can reward something more specifically, still within a "normal" -R-based riding session. You still have the "generally positive" associations for the horse but with the added benefit that the click is being paired with a particular behaviour that you have chosen to work on. You use shaping in the normal way and the horse has the opportunity to learn the new behaviour, or refinement of an existing behaviour. You still don't get the element of choice in your training and I would say it's debatable how much the horse really feels positively reinforced when it is part of a schooling session in which the horse has not necessarily chosen to participate. But it's a good way of marking exactly what you want so you give the horse a good opportunity to get the right answer quickly, rather than having to work out by more aversive-based techniques. Of course, you can still have a horse who is worried about getting the right answer because of the way you are combining +R and -R - how is the horse supposed to know when he is expected to offer behaviours and when he is supposed to follow cues? But if the pressure is minor (and always was minor, I don't mean starting big and then scaled down so that the high-pressure stuff is always lurking there as a threat...) then most horses should be able to cope. I've used this sort of approach a couple of times with my horse. Once I was trying to get him to speed up a bit instead of our slow dawdly sort of plod which can sometimes take forever to get anywhere. I confused the hell out of him and ended up with tiny steps instead. I was interfering with his natural rhythm and (in hindsight) actually what he needed was regular osteopathy and the freedom to just get on with it instead of being micro-managed. The other times I tried it was out hacking when he would dive for he hedgerow so much we took forever to get anywhere (can you see a pattern emerging...??! It's always down to the impatient human, isn't it?!). I would click after a random number of steps, aiming to get him to eat after a click rather than when it suited him. We'd had a really successful time of doing this in-hand, going across a dairy pasture so I figured I could do it on-board. The main difficulty I had was that hedgerows are not all the same the way a dairy field is. I would click at the wrong moments because he wouldn't want to eat whatever plant was there and he would continue to dive for the plants he wanted, click or no click. Again, I found we were more successful when I just gave up the micro-management and accepted that if I'm going to ride in a bitless bridle (and with half-rubbered reins with the rubber all wearing off so I have NO grip!) then I should take what I get. I probably just need some different reins...
Finally there is the approach where you do the CT properly. You start from scratch and free-shape everything. This might involve following another horse so you can elicit the movement easily or you might use targets or you might genuinely free-shape it and wait for the movement to shape. Then you could incorporate something like David Dodwell's Horse Morse Code where you have a clearly-defined set of cues to pair with the behaviours you have shaped. The more complicated the cues/behaviours the more you might be tempted to revert back to including a bit of -R to help clarify what you want, so if you want to stick to free-shaping you would need lots of imagination and lateral thinking to make it work. I took this approach with Jak for a while. We'd done loads of conventional dressage in the name of trying to keep him supple and hold off his arthritis. It made him miserable which was why I started looking at +R in the first place. So I started thinking how I wanted to start again and free-shape things. I didn't really have anyone to do this with so didn't have the option of following another horse (probably how I would want to do this with a new youngster or abused horses etc) so the way I tried was to elicit forwards motion on the ground by using a series of targets (upturned flower-pots) and then planned to extend this to on-board. It worked at walk but when trying it at trot I only ever seemed to succeed in annoying him. So I tried just sitting on him and aiming to capture forwards motion. There was a hilarious workshop when it tipped it down with rain and I spent 15 minutes or so sitting on Jak, everyone soaked and Jak immobile. There was absolutely nothing we could shape into ridden CT work.
So this all really got me back to why I wanted to do it. By this point, when I took Jak out for a hack he would be enthusiastic, supple, fun and clearly not phased by my occasional use of pressure. I had wanted to re-train dressage for his benefit but we'd kind of moved beyond that point. The hacking we do had him moving much more freely than the dressage ever did because he was much more enthusiastic and self-motivated. Treats weren't going to change that. I could eventually see that the free-shaping dressage was a great clicker challenge for me and I wanted to do it for me. One day with another horse I may still do it but just getting on with it and having fun is right for Jak now.
CT for me is no longer about feeling I have to train everything with CT to make everything positive. It is more about doing enough CT and free-shaping that he can retain his sense of choice and autonomy. And consequently that it is no big deal when I haul his head out of the hedgerow, or any of the other occasions when I resort to pressure. Most of our rides are just point-and-go, rather than planned training sessions. I prefer to retain CT for free-shaping stuff that doesn't matter to me so he can have absolute choice in whether to participate. Anything else I feel dilutes the power of CT. But these thoughts are purely where I am with Jak today. Another day, another horse I may think differently and I think this flexible thinking is really crucial to these sort of discussions so people don't feel there is a set way of doing things. A clicker is only a communication tool, you could use it to mean a smack is coming (please don't!). Its use with horses is still relatively recent and we are all still exploring.
Copyright Catherine Bell 2014