Thursday, March 31, 2016

Clicker Expo Cincinnati 2016 (part 2)

As I mentioned in my previous post, much of  my time at Clicker Expo was spent on talks which weren't directly about training but there were a few! The first was "The Sound of Silence: Accessing the Power of a Withheld Click" with Michele Pouliot. 

I like the term withheld click because it describes a moment when you could have clicked but didn't. We aren't talking about missed clicks here, but when one intentionally chooses not to click. In the talk she discussed guidelines for how to do this effectively and what to watch out for. 

Various reasons people withhold a click include shaping successive approximations, extinction, differential reinforcement, inducing resurgence and others. Using it inappropriately or without developed skill can certainly lead to problems. 

Withholding a click can lead to frustration if we don't, at the same time, set it up so that the animal makes a good guess as to what we're looking for in more behavior. It can also lead to an animal giving up because nothing seems to be working or they just can't get it right. More intense individuals may start throwing other behaviors at you in an attempt to get you to click something. For these reasons, we need to think carefully about how we are going to do this.

A key aspect to when you can withhold is the history of that behavior. If there is too little history, you have nothing to fall back on if the animal doesn’t understand. If you have spent a lot of time clicking for a behavior, then the animal thinks that behavior is good just as it is and also won’t understand if you stop clicking for it. Michele mentioned that handlers tend to fall into either side of this challenge: raising the criteria too fast, or staying at the same criteria too long. I know which I do and when!

She did say that if you use a well conceived training plan initially, and you want to change an established behavior, you can go back to an earlier stage and reestablish a history of clicks before withholding to ask for variety. 

One thing I see people do fairly frequently is to build duration into a behavior, and then if the behavior falls apart for some reason (possibly due to poor information during the shaping process), they try withholding a click, but the animal simply thinks it’s about duration so they don’t offer any changes. This leads to frustration on the parts of both learner and trainer.

In addition to the animal being sufficiently prepared to benefit from withholding, the trainer must have a plan. She listed four pieces. The first is anticipation of what the change will be (you should have your training session and environment set up so that you are pretty sure that your learner will offer what you want, rather than something in the other direction). You must then be prepared to click that! If you don’t really know what your learner will offer, it’s hard to catch the right effort that first time. You must also be ready to to back to clicking previous criteria if things start to go awry or else your moment of silence will extend into more than a moment and the accompanying frustration or giving up. And finally, know yourself. Is your tendency to ask for too much too soon? Or is it to allow the animal to get stuck by not inviting variety?

Something else she mentioned which I feel requires time for a trainer to develop is how to assess the animal’s response to withheld click. She said there can be “thinking moments” and she looks for the animal to remain engaged mentally, even if physically still. So if you withhold a click and your animal then looks off into the distance or frantically starts trying things, those aren't thinking moments. Then when your animal does offer something and you click it, does he offer that new variety again and does he offer it quickly? If so, he probably made a conscious decision about what to do and understood the new click. If not, it may have been a random movement and the click not understood by the animal. 

Saturday, March 26, 2016

Clicker Expo 2016 (Cincinnati)

my own neckpiece for the weekend included several species!
Clicker Expo can be many things to many people. Billed as the world's most innovative training conference, this year included training topics from many species. The main focus of the event is dog training, and dogs are the sole specie one is allowed to actually bring to the event. This year, for the first time, there was also a track devoted strictly to horse training. For horse people among us, this was long awaited and well attended. 
The opening talk by Ken Ramirez was about training butterflies, of all things. Watch the BBC for upcoming video of this amazing project.

Several of the talks which drew me in this year were actually directed at working with people: everything from Sustaining a Competitive Advantage in business (given by Aaron Clayton, president of Karen Pryor Clicker Training) to Critical Client Conversation Skills (given by the always entertaining and enlightening Dr. Susan Friedman of Behavior Works). I also attended one of several TAGteach presentations by Theresa McKeon on how to teach our human learners using the same basic principals we use teaching our animal learners.  Yet another human based talk I attended was "The Future is Now: Creating Powerful Trainer/Vet Client Teams" by Debbie Martin. All these just go to show that we can't help our animal learners without including the people associated with them.

Clicker Expo also advertises "training skills with cutting edge science" and if you want science, talks by Dr. Jesús Rosales Ruiz are the ones to attend! In the past, his talks have made my head hurt. He has a knack for putting behavior into algebraic equations and I struggled to keep up. This year I felt I was able to keep up with the entire talk, even when attending two of his back to back. "The Quadrant Quandary: Clarity and Perspective on an Icon" was the name of the first. In this talk he shared history of how the terms punishment and reinforcement, both positive and negative, came about. He spoke to whether these terms were defined in relation to process, procedure or stimuli. Looking up process vs procedure took me right down a rabbit hole but I think the point he was trying to make is that we need to be careful when we talk about the trainer "adding" or "removing" something compared to the behavior of the animal causing something to be added or removed. Additionally the notion that something is labeled as "aversive" or "appetitive" is completely dependent on multiple variables such as the learner, the environment, and the history. A longe whip might have an aversive effect on one individual and appetitive on another (such as my horse who loves the opportunity to take it out of my hands and play with it). For these reasons (among others), one needs to be very careful when labeling a trainer based on a photo or a video or a paragraph or an experience. Going back to Susan Friedman, it's best just to drop the labels altogether unless you have a very clear definition understood by all involved.

And that was just the first five slides of his presentation.

The second talk of Dr. Rosales-Ruiz which I attended was titled ""Fast Learner" The interplay between reinforcement rate and criteria". Again, he took some standard catch phrases and helped to me see them a little more carefully. "Raise the rate of reinforcement" is a common solution I give when people are having trouble training. It's a simple mantra, and often pertinent to people I am coaching who are new to training. But it is possible to bore a learner, to build in superstitious behaviors and to just get stuck by going too slowly. The takeaway here for me is to carefully assess criteria I am after and observe what is offered.   

More tomorrow.