This is a common question for beginning clicker trainers. First we learn the rudiments of marker training: that if we mark and reinforce the behavior we want, it will increase until we can eventually put it on cue.
But at first most of us feel like we are doing only half our job. We have been telling the dog when she does something right. That’s great! Hardly anybody disagrees with doing that. But it seems to us that we also need to explain to her what is wrong.
How can she really understand what’s right if she doesn’t know what’s wrong? Whether it is purely cultural or a primate thing, as some have suggested, it is deeply imprinted in us humans that pointing out mistakes, which can have a punishing effect, is essential to teaching and learning.
First of all, many studies have shown that it is not necessary for an animal to make mistakes, much less be told about them, to learn a behavior to fluency and extreme reliability. Also, I have already written about the research that shows that mixing reinforcement and punishment is actually detrimental to learning. (You can read Jean Donaldson’s magnificent rant on the subject here.)
But my point in this post is much more “nut and bolts.” Let’s take Josephine the Average Trainer, who may be a beginner and in any case is not conducting her training in a Skinner box (an environment that can block out any extraneous environmental events and help make training very “pure”). Josephine’s dog is probably making a fair number of errors. So the question is, why not punish them somehow? The punishment doesn’t have to be harsh, and clicker trainers generally support the thoughtful use of negative punishment, where something the dog wants is taken away after they perform an undesired behavior. Wouldn’t that speed things up by making the process ultra clear? Increase the right stuff and decrease the wrong stuff, all at the same time?
Watch this video and come back. Pay special attention to what happens at 0:45 – 0:48 and especially 0:56 – 1:01.
First, wasn’t that phenomenal? That is Sue Ailsby and her young Portuguese Water Dog Syn. Sue is the author of Training Levels: Steps to Success, in my opinion the best book available for someone training a dog on their own. In the space of one short training session, Sue teaches Syn to stand, to hold the stand for a few seconds’ duration, and to maintain the stand while being touched. Sue shared with me that the session before editing was about 10 minutes long, with some of that time being spent on Syn’s learning process and part of it dealing with the issues that always come up when you try to film something.
Sue made the video for the Training Levels Yahoo group. Some members had questions about teaching a stand. One of the things Sue is showing is that you don’t have to wait and capture the action of standing up; you can click while the dog is already standing and they will quickly learn that that position pays off.
But what I love about the video is that it shows a clicker-savvy dog learning from the click and from the absence of the click what pays off. That’s right. Absence of a click was all Syn needed to understand right away that downs and sits (and putting paws on stools) were not what Sue wanted in this session. Nothing else was necessary.
Absence of a click is generally agreed to employ extinction, another way animals and humans can learn. Extinction means not reinforcing a behavior that has been previously reinforced. It too can be an aversive process. But I think most of us would agree that Sue’s just waiting for Syn to try something else (especially since trying stuff is itself a behavior that has been reinforced) is a pretty darn painless response to Syn’s mistake. For most dogs with most trainers it would be the least aversive response by the human.
The video helped me put something together in my mind. Trainers who train primarily with positive reinforcement often say that most trainers who rely on punishment are announcing their lack of skill. Folks, this is not just rhetoric, and this video with a highly skilled trainer demonstrates that. Syn can learn from the lack of click so quickly because Sue’s timing is good. If instead of Sue we had someone a lot less skilled (like me), they would likely have clicked a handful of behaviors they didn’t want, and perhaps failed to click some really good stands. That would mean that the percentage of good information the dog was getting would be much lower. She would then be much more likely to choose the “wrong” behavior. And when that happens is when most of us are tempted to use punishment.
I don’t know about you all out there, but this video inspires me to work harder on my timing. And in the meantime, remind myself that even with errors on my part, just marking what is right, to the best of my ability, will work. We just may not be as fast as Syn and Sue.
Thanks for reading! Stay tuned for: