I think I might experiment with a Keep Going Signal (KGS). Someone wrote a response to the blog on a list I am on and asked why I wasn't using one. She explained how she uses it with duration behaviors such as going over trotting poles and also for something in a series of behaviors like a canter depart or flying change. The reason for using it with going over fences or a flying change is that you don't want the horse to slam on the brakes when he hears the click. Many times that is acceptable because you then can ask again, giving the horse several repetitions of successful behavior. But slamming on the brakes when landing from a jump is not good for the structure or performance...so a KGS would mean "good job but keep cantering until I ask you to stop or you hear a click".
To use it to keep the trot going, I would first condition a KGS by using it right before the click and primary reinforcer (treat). Ande would then learn that the KGS (I'm thinking of using the verbal "gooooood" which just kind of oozes out of me when a horse is trying hard) means that a click and treat is coming and he needs to continue what he is doing to get it. That is why it's called a "keep going" signal. Rather than getting frustrated and quit because he isn't reinforced, he would know to keep trying.
I looked up KGS on Karen Pryor's website and this is what it said:
A signal, verbal or otherwise, given in the middle of a behavior to tell the dog he is doing the behavior correctly and should keep doing what he’s doing. Keep-going signals add an unnecessary level of complexity in training.
So, obviously some consider KGS to be unnecessary and complicating. I think the reasoning is that simply withholding the click has the same affect. Has anyone else experimented with them and found them to be helpful or complicating?
3 comments:
When I started with clicker I used a method developed by a trainer called Ben Hart which is similar, although not quite the same as a KGS. The clicker is used as an intermediate bridging signal, and when you reach the hand to get the treat, that serves as the terminal stimulus and ends the behaviour. So you give the horse multiple, treatless clicks before ending the behaviour.
I found that although this method does help with some duration behaviours I had a lot of problems trying to use it for all training. My horse showed high levels of frustration, that occasionally became agression and nipping. I found it very hard to get clean loops as essentially the behaviour at the click has to be the same as the behaviour at the terminal stimulus. I was inadvertantly chaining in behaviours between clicks and the TS, so getting a whole lot of stuff I didn't want, and got in a bit of a mess with it to be honest. We've got on a whole lot better going to the simpler method as Alex Kurland uses.
Obviously I know this isn't exactly the same as what you are talking about and I'm sure some horses get on great with it, I just found it complicated matters too much for me and the horse.
I think that's probably the exact sort of thing that becomes a problem. It's amazing how many variations of "clicker training" there are. I"m not sure I've ever heard Alex speak on KGS but certainly 1 click = 1 treat. I'm glad to know things have gotten better. I think I'll have to be very careful how I condition the KGS so it does not lead to frustration later on. Good point! Thanks.
I've actually never heard of the KGS before (until I just read it on your blog).
But now that I know what it is, I realize I use it all the time with my ponies. Especially Minnow.
For example when teaching my ponies to paint. Once they've got the concept of target brush to paper. I use a "good boy" to get them to keep targeting the paper and continue on with strokes until I ask for the brush back and click/reward them.
I also use it in my mounted games training. a good boy for a really nice spin around the end pole, but I want you to keep going until you finish the race and I can give you your treat.
Its been working really well for all of them...although I guess I didn't realize until now I was doing it. I think it developed as I was teaching the initial clicker training with all of them. a valiant effort got a "good boy" click and treat all at the same time. over time I guess I started to space them out. now when they hear good boy they all get more animated about their current effort because they know their reward will be coming shortly.
Post a Comment