That was the same summer that Dad and I went to an “Analysis of Behavior” conference in Chicago. Dad still subscribed to their journals and noticed that Karen Pryor, the famous marine mammal biologist and behavioral psychologist would be speaking there. Karen was a pioneer in developing and applying purely positive classical and operant conditioning methods to train dolphins in captivity. Dad had given me her book, Lads before the Wind: Diary of a Dolphin Trainer, a few years before. In it, Karen chronicles her experiences as the head trainer at Sea Life Park oceanarium in Hawaii. Sea Life Park was the first of its kind, opening eight years before the first SeaWorld.
We were the first people to find seats in Karen’s conference room. By the time she started, the room was at capacity. She introduced herself and said she wanted to give us a demonstration of a very common training method called “shaping,” which was used frequently with animals when you wanted them to perform a complex behavior on cue, such as a very high jump out of the water. The probability of a dolphin ever performing that behavior on its own is very low. So, a trainer would have to begin with an existing response and then differentially reinforce closer and closer approximations to the target response. For instance, a dolphin might be reinforced first for a small jump out of the water. Once that behavior increased in frequency, the trainer would put the animal on a “variable ratio schedule of reinforcement”—in other words, the dolphin wouldn’t receive reinforcement anymore EVERY time he made the jump, he would receive it every few times he made the jump. This had to be done so the dolphin would not come to expect the reinforcer every time and quit behaving when it didn’t come.
Once the behavior was stable on a variable reinforcement schedule, the trainer would begin to only reinforce jumps that were higher than the first ones, and withhold reinforcement for the old behavior. And because of the schedule, the dolphins would persist! Then once the dolphin was performing that jump regularly, the trainer would begin to only reinforce jumps that were higher than the second ones, and then only ones higher than the third, and so on, until the dolphin is jumping very high out of the water. Delivering reinforcement at just the right time obviously took great timing on the part of the trainer. It seemed like a challenge, considering the dolphin is swimming around quickly in a tank of water sometimes meters away from the trainer. I doubted that throwing a fish across the pool at a dolphin at the precise time was the answer.
That’s when Karen explained how they must first “whistle-train” the animal using classical or Pavlovian conditioning. They simply pair the whistle sound with receiving food over and over—not requiring any behavior on the part of the animal—until just hearing the whistle signals to the dolphin that food is coming. Then, they stretch out the interval from the time the dolphin hears the whistle until he receives the food. That’s necessary because often the dolphin performs a trick far away from the trainer and must swim to the trainer after the trick is completed for his reward. After classical conditioning, the whistle can be blown at the precise moment the dolphin performs the desired behavior and the dolphin will associate the reward of food with whatever he was doing at that moment.
I absolutely loved this stuff. So, when Karen was ready for her demonstration and asked for two volunteers, I stuck my hand up fast. She picked me and an older man and told us both to go out in the hall while she explained what she was going to do to the rest of the audience. The man was called in first and after about fifteen minutes I heard a round of applause. Then I was called in. Karen told me to pretend that I was already “clicker-trained” because she was going to use her clicker to reinforce me every time I performed the right behavior. She said she had a task in mind she wanted to “train” me to do and that she was going to use shaping to get me to do it. She then told me to start behaving.
From the back of the room, I quickly looked around and wondered what she might want me to do. Turn the lights off? Shake someone’s hand? Go sit in a chair or on the ground? Do jumping jacks? Rather than guessing, I decided to do something small and see what happened. I cautiously took a step forward. CLICK. I took another step forward. CLICK. I took several steps forward. CLICK. I was now at the front of the room looking at the conference table in front of me. I looked to the left. NO-CLICK. I looked to the right. NO-CLICK. I looked at the pitcher of water and empty glass sitting in the middle of the table. CLICK. I picked up the pitcher. CLICK. I poured a glass of water. The room erupted into applause. It all happened so fast, in less than two minutes, I hadn’t even thought about what I was doing. I sat down a little dizzy.
Karen then thanked us both for demonstrating one of the main points she was trying to make, that not all animals—or humans in this case—are equally easy to train. The first volunteer completed the task in fifteen minutes. He had lost some time trying out a variety of behaviors that were in the wrong direction and sometimes emitting behaviors in the right direction, only to then veer off-course and cause Karen to have to start the shaping process over and break it down into smaller increments. As the second volunteer, I’d completed the task more quickly, perhaps because I was more sensitive to the reinforcer and let myself be shaped. Dad and I couldn’t help laughing on the way home about that one. At least I’m highly trainable.