Wednesday, March 21, 2012

What are my cues....really?

This weekend Eloise (dog) and I head down for our second KPA workshop. As I had hoped, this course is improving my skills and depth of knowledge in training. Whenever the exercises are appropriate, I try to do them with a horse as well as, or instead of, a dog. And when they aren't compatible, I stretch my brain to see what correlations I can draw.

My biggest stumbling block (so far!) is stimulus control. I have been whining to anyone who will listen (or read) how frustrating I find it. All trained behaviors should be under stimulus control. This is defined by:
  1. The animal should perform the behavior every time it perceives the cue (can be verbal, visual or other...thus "perceives")
  2. The animal should not perform this behavior when it hears a different cue
  3. The animal should not perform a different behavior when it perceives the cue
  4. The animal should not perform the behavior in the absence of a cue
The first one is easy. The second one is pretty easy. The third one is less so. The fourth one is impossible.

My problem is that my animals offer behaviors all the time. They love training. They love to work for me. They love to show off, interact with people and get attention. This is the result of Clicker Training- animals who throw behaviors at you in order to engage you. It's far better than animals who do obnoxious things like bang on stall doors or paw on the cross ties. Instead, they offer lovely Pilates positions, target anything which might be accepted as a target and grab clothing they think you might have dropped and hand it to you helpfully. How lovely! So I reward. But did I ask for these behaviors? No. So technically they aren't under stimulus control.

I thought it would be doable to correct this. I had read my lessons, watched the videos, understood the concepts and related it all to things I've done before. How wrong I was. It seems like everything in the environment is functioning as a cue. Percy thinks a target just needs to be touched. "What do you mean, wait until you say touch? Why? I know what you want. Give me the bloody thing so I can touch it!"

One of the reasons I did not predict such a problem is that horses who are not under stimulus control can be dangerous (as opposed to Eloise who sees me pick up the clicker and starts popping and backing and downing and spinning helter skelter but does no damage). The crutch I have used, however, is asking for an incompatible behavior. If I don't want Percy to touch a target, I ask for Grownups. He happily complies. But that's different from an animal who simply waits to see what exactly you want. I should be able to put out a target and choose to have him pick it up or touch it. If I speak fast enough, I can do that...but if I don't immediately tell him what to do with it, then he decides. I should be able to put down a mat and ask for him to touch it rather than step on it. Not a chance. Mats are for stepping on! (he actually tried to step on it when I held it up to try to get him to wait...yup, dangerous).

Now it's not that Percy doesn't understand the cues. It's Jane that doesn't understand the cues. As Alexandra Kurland says, objects are cues. People are cues. So he knows what to do in the presence of all these things (put anything on the floor these days and Eloise has the same opinion only she lies on it, rather than just stepping on it....her mat work is excellent. But not under stimulus control. Sigh). So if the object is the cue, then neither Eloise nor Percy are wrong. They are not performing the behavior in the absence of the cue...the mat is present. Cue perceived, behavior performed.

Wow, I've been struggling with this for over a month and there it is, right there. Alex says cues evolve out of the shaping process. So for Percy, the presence of an object evolved into a cue. I did not train for it differently. I'm taking well-known behaviors with well-known cues and changing the rules. It's a wonder he hasn't refused to work with me at all. Just like we need to generalize cues to new environments (Eloise believes strongly that the chair I am sitting in, the room we're in or whether I'm standing or sitting are the cues), we also need to introduce the idea that an object, a place, a body position is not a cue fairly early on in the shaping process. Unless of course we want the object to be a cue.

In hindsight that all seems so obvious. I've got work to do in the next 2 days....wish me luck!


Lottie Eriksson said...

Certainly am wishing you the best of luck! This is an issue I'm constantly struggling with as well. Did you see the dog video that someone, I think it was Alexandra Schreiber, posted on Peggy's site recently? Where the trainer was teaching cue discrimination by giving a "nonsense cue" for a default behaviour? Having such a "stand still and wait" behaviour as a default might be a good way to teach the cues we really want to use.

Bookends Farm said...

Thanks Lottie!
The tricky part of reinforcing "wait" is that I end up creating ghost behaviors. Whatever they happen to be doing when I click they will repeat. It does work if you can click something different each time, but it seems there has to be a certain amount of extinction as well. I was trying to avoid that but perhaps it's not possible.