ClickerSolutions Training Articles

Learning About Learning:
The Great Box Experiment

Having some free time on my hands last week, and having access to 5 different dogs to train, I decided to try an experiment. There had been some discussion on click-l (the internet list) about an exercise called '101 things to do with a box'. This exercise was first described by Karen Pryor and is a great way to get your dog to start offering behaviors.

You start with a box (actually, any object will do), a dog, your clicker, and some treats. Then you click and treat the dog (CT) for any interaction with the box. I tried a couple of variations on this exercise with the dogs I had available and saw some interesting individual differences in behavior.

First, let me describe the dogs: Katie is a 7 year old Black Lab who is a crossover dog and who has weathered most of my training experiments. She is semi-retired from competition due to health problems with one Open leg and two Novice Agility legs. Katie is the 'perfect' companion and pet.

She has an air of self-confidence about her. Sully is a 3 year old Golden who has always been clicker trained. Sully is goofy and silly and eager and overall, lots of fun. He's been trained for most of the obedience exercises, but not shown yet.

Java is a 2 1/2 year old Bernese Mountain Dog with a CDX who is working on Utility exercises. Java is calm, reserved, quiet, and has been trained with positive but traditional methods. He did not have any CT experience.

Buddy is a Pembroke Corgi who is also 2 1/2 years old. Buddy is bold and outgoing and funny. He has been trained for most obedience exercises, but not shown yet. His training has been similar to Java's.

Daisy is also a Pembroke Corgi. Daisy is 4 years old. Daisy has never been asked to do much of anything in her life. She's spoiled and extremely intelligent. She's very pushy with dogs and people, but can be loving when it suits her purposes. Daisy had a small amount of previous clicker experience as I taught her, on a dare, to retrieve in a couple of hours using the clicker. The challenge was "If you can teach Daisy, you can teach anyone."

I worked with each dog separately. The box I used was about 2 ft. by 2 ft. (big enough for a Corgi to fit into). I went into each shaping session being completely open about what would get reinforced. I decided to watch the dogs, see what they might offer, and go with the flow.

Katie had the most interesting responses. She's fairly used to my strange methods by now, and has learned to try things until she gets a click. In very short order (5 minutes) she was doing a wide variety of behaviors. Every interaction with the box that was different from the last one was reinforced.

She sniffed the box, put her head in the box, pushed the box around with her nose, put her paw on the box (first one, then the other, then both), put her paws in the box, picked up the box with her mouth and dropped it, picked up the box and threw it at me, rolled the box, stood in the box, etc. etc. She was very animated and seemed to be having a great time.

I realized that basically, I had just trained my dog to tear up a box! Boy, what a brilliant trainer. She reminded me of that scene in the movie "Ace Ventura, Pet Detective" where he kicks, drops, throws, etc. the box he delivers.

Sully started offering responses similar to Katie's very early on as well. In fact, he got so involved in box destruction, that he completely missed some of his clicks. He just kept jumping on the box, pushing it around, throwing it, etc. That must have been very self- reinforcing.

I worked for a while refining some of his responses and got him pushing the box all over the house and standing with his head hanging down into the box like he was looking for something. Duration is very hard to get with Sully, so it took a while to increase the time he would stand still and keep his head in the box. If I tried to increase duration too quickly, he would try another behavior.

Java was slow to start and the progress was much slower. I expected that due to his personality and the fact that he didn't really understand the process of shaping. It took about three 30 minute sessions to get the behavior of him putting his head into the box. I stuck with shaping one behavior because Java was very unsure of himself, and expecting creativity this early seemed too much for him.

He did learn the CT connection very quickly. He laid on the floor in front of the box and would put his nose in, then his muzzle, then his forehead, etc. Each step was slow going, but he seemed very willing to continue working. Duration was very easy to increase with Java, as compared to Sully. He's very patient and would wait quite nicely for his click.

By the end of the first session he'd put half his head in the box, hold perfectly still, then snap it out when he heard the click. I got worried he'd give himself whiplash! Towards the end of our sessions he started experimenting by putting first one foot, then the other, into the box as well. I reinforced that.

Even though Buddy didn't have any CT history, his personality was so bold and outgoing that he caught on much quicker than Java had. I started by reinforcing him for putting first his head, then his entire body, into the box. He quickly learned to dart in and out of the box.

Then I switched and started reinforcing him for pushing the box around with his nose. I saw a little extinction burst, but he was fast to figure out the new contingencies. While he wasn't as physical as Katie or Sully, he did learn to push the box all around the living room.

Daisy was the last dog in the experiment. Like Buddy, she quickly learned to go in and out of the box for CT. Then I decided to shape her to put her paws up on the top of the box (the box was on its side) and stand there. Daisy was a very amusing worker.

She seemed to have a sense of when reinforcement should occur, and she let me know with a pointed stare and grumble when she thought I was wrong. I've never had that experience before. I believe that if Daisy had thumbs she would have taken the clicker from me and clicked herself!

At one point I discovered that Daisy worked quite well for the chance to lick a tiny bit of whipped cream off my finger. While shaping the stand a friend suggested that I lure her instead. I tried to explain why I like shaping better, but went ahead and tried the luring in order to demonstrate.

When I lured, of course, I got the behavior to occur very quickly. However, Daisy wasn't actively learning, she was just following. She made no attempts to duplicate the behavior without the lure. I think that the luring actually inhibited the active learning process that occurs during shaping. I went back to shaping and within 10 minutes Daisy was standing with her little feet on the box.

So, what did I demonstrate with this experiment? Several of my friends have laughed about the usefulness of teaching dogs to put their heads in boxes. However, the actual behavior wasn't the point.

To me, the point was the learning process and the different ways it worked with different dogs. The box exercise was really a completely different exercise for each subject. I noticed myself becoming a different trainer with each dog. My methods, timing, and shifts in criteria all changed as the dogs changed. I love this type of training because it is so interactive.

The trainer and the dog create something together that requires total concentration from both parties. I believe that it's very important to focus not only on the goal behaviors, but on the training journey that leads to those behaviors. How you get there is as important as getting there.

Deb Jones, Ph.D
copyright 1999 Deborah Jones


| Training Articles Contents || Site Home |

List and Site Owner: Melissa Alexander, mca @