This discussion was split from Titum's diary.Donald Redux wrote:
Do you behavior mark, that is do you click or make some other identifying sound that identifies for the horse the exact behavior increment you are working to capture?
Not very consistently. That is, if I want them to do a very specific thing, I do use short words ("goed", "jip" and so on) as reward signals, hopefully with some precision, and sometimes I even click with my tongue. However, during our usual interaction and play, I mosty don't. Of course I praise the horses a lot when they do something I like, but I guess I can't really call this a marker.
I have often asked myself why I don't use reward signals more precisely, knowing about their benefits. But somehow I was feeling that it didn't really seem to fit, it just didn't come to me naturally. I guess there are several reasons behind this intuitive aversion (a much too strong word, but I hope you know what I mean).
One of them is that often I am not after specific behaviours, but a more general state of the horse. For example when I reward them for being careful with me, it's not this one step of moving backwards, or that tiny bit less physical pressure of their muzzle on my hand when taking the treat. It's their general attitude, which is of course reflected in these behaviours to some extent, but I feel that when I am marking the behaviour more specifically, I am also shifting their focus away from what actually matters to me.
Another reason is that often there is no "correct" behaviour, but communication with answers that depend on the course of the conversation. For example during our body language experimentation it is nice when the horses react to my movement in the way I had planned, but if they do so, that means that my cue was good enough. If they react differently, that reaction generally is not less correct. It is consequential. As long as I am not able to evaluate my own body language cues in an objective way (and I am afraid there is no such way), I can't judge the correctness of their response, so I prefer not to try rewarding correctness but hand out the reward in a less specific way and adapt my own behaviour according to their response.
Another aspect of non-existing correct responses is that many of my actions are suggestions and not cues in a strict sense. That is, if I am asking Titum to rear and he does not, this is not necessarily less correct that when he does. It is just his reply, which in turn determines the course of the subsequent conversation. Perhaps that's a bit hard to explain in horse training terms, so I will try it with a human example. Imagine you are shopping groceries with Kate. You ask her if she wants to buy tomatoes for dinner, and she says she prefers spinach. Was this more or less correct than if she had said yes? I often find myself in a similar situation with my horses. I can still reward them, because after all they were answering and this is keeping up the conversation, but I would have some difficulties if I should specifically mark the "correct" aspect of their behaviour.
Finally, there is a reason we have already talked about before. Not sure if you remember though, it has been some years ago. Personally I am not able to adequately show my enthusiasm when I click. Theoretically I know that I could first click and then still praise them verbally, just as much and with just as much emotional differentiation as I do without a click. But often I don't, as if I had already "done my duty" with the click. So not only my expression of happiness gets lost but also the differentiation between the degrees of fantastic.
And then there are other situations where I am after a very specific behaviour, and where the situation also determines that there is a single correct behaviour, with the horse's response being more or less close to it. In these situations I usually do mark the behaviour, but in the whole of our interaction they are rather rare.