ClickerExpo2004

I just returned from Clicker Expo in Philadelphia. This is a conference devoted to Clicker training (operant conditioning, Skinner's work on training) Mostly this was a conference of dog trainers but there were several trainers from the larger zoos (cheetah, buffalo trainers) and from SeaWorld. (question "What are you training the Buffalo to do?" "I train them to not kill me.") Also there were people who were using this sort of "positive feedback in the moment" to teach developmentally disabled children and gymnastics.

http://www.regonline.com/Checkin.asp?EventId=10367

[If you bug me enough I will write a quick review about how shaping works and put it here. If you really care read the book "Don't Shoot the Dog".]

It was most fascinating. There were lots of interesting presentations which had relevance for training in general and working with people.

Here are some thoughts on the most relevant work.

Karen Pryor mentioned many interesting things about "extinction". She said that it is ALWAYS distressful to the animal and that much new behavioral variety is exhibited in the burst before the behavior is extinguished. This sounds to me quite a bit like chaos followed by a transforming idea.

So her latest approach to putting a behavior on cue is to NOT use extinction of off cue behavior but instead to train a "wait" (starting small and building up) for the the cue. This turns out to be faster and less traumatic to extinguishing off cue behavior by ignoring it.

Part of the job of a trainer is to CREATE reinforcer. Killer Whales that have just been caught have few reinforcer (food) but training with only one reinforcer creates boredom. So trainers create an interest in toys, back rubs, even known tricks are reinforcing and can be used to teach other tricks.

There is new evidence that the RATE of reinforcement is important to the training. More reinforcement will speed training. Now trainers are setting "goals" of 50 reinforcements in 2 minutes to ensure they are working as fast as possible. In fact there is research in TX that food placement is important for this reason. If you throw food the animal must move to get it and this breaks their concentration and creates delays, faster training is possible when food is delivered to directly to the animal.

Kathy Sdao has a great story about training a dolphin in the open ocean. The dolphin not only continued to work when it could have swum away but it worked for frozen mackerel, while the sea was full of fresh mackerel (which the dolphin would sometimes eat). So the animal was not working for the mackerel but instead was working to be reinforced, it was the relationship which was important.

There were several presentations which discussed the important of ignoring behavior that is undesirable. The act of 'punishing' undesired behavior has two consequences, the 'violence' will create an unpredictable animal and the act of calling attention to the act will ensure that it will occur again. There were interesting stories about owners who had very lazy bored behaviors for tricks which were performed correctly but amazingly energetic 'punishing' behaviors for botched tricks. This acted as a reinforcement to the incorrect trick.

Other stories about people who give a signal again if the animal does not do the trick. This is not good training practice since if the animal does the trick on the second cue you will need to reinforce the behavior. You will then be reinforcing BOTH the ignoring of the first cue an the trick on the second cue, far better to ignore the missed trick and move on to something else. This is related to the question of what is the antecedent and what is the consequence. When several commands are given in sequence it is often hard for the trainer to understand that some of his antecedents for the next trick are in fact consequences of the last one. This can create training errors. Animals are amazingly understanding of training errors, there is one dog who will bark and complain of misscalled cues in an agility run, in effect training the trainer.

In short, you should have been there. Perhaps next year I will see more shapers there.

There was an amazing discussion about the merits of a lifemanagment program which is now popular (rough love: nothing in life is free). While most trainers were unfamiliar with the details an interesting consensus came out of a discussion with the few who knew about it. The program is a very detailed lifestyle management program (CMM level 5 for dogs) the successful trainers seemed to all view the program as a means to ensure that the dogs and them had the most fun. This was viewed as a means of ensuring all reinforcer were given with the owner present so that a bond and relationship could be developed. The unsuccessful trainers viewed this more from animal dominance theory. They believed they were the warden and would force the dog prisoner to submit to their will. Listening to these dog trainers sounded much like a discussion on the shape forum. So once again the culture is more important then the process which is being applied.

KenEstes 2004.03.29

Animals are amazingly understanding of training errors, there is one dog who will bark and complain of misscalled cues in an agility run, in effect training the trainer.

My Jack Russell Terrier has me very well trained... and we don't even do agility. Right now he is cueing me to give him supper. :-)

Updated: Monday, March 29, 2004