obtain a toy by simply touching it when I presented it to her at the side of the pool. She learned this quickly, and apparently getting the toys themselves was reinforcing.
Then I conducted my first formal experiment: testing if a dolphin could visually discriminate between the white symbols. This was not a trivial question at the time; the everyday world of dolphins was thought to be dominated by sound, not sight, and, as I mentioned before, dolphins' echolocation abilities are exquisite. I therefore couldn't
assume
that dolphins could see well enough and discriminate well enough to use the visual symbols of the keyboard. Back in 1979 when I started with Circe, we did not know if they could. More than a decade earlier, at the Stanford Research Institute in Menlo Park, Winthrop Kellogg and Charles Rice had done some very preliminary studies on this kind of ability in dolphins, but they had gone only so far. Essentially, I was starting at square one. The first part of my doctoral research, therefore, was to discover whether dolphins could visually discriminate among simple shapes.
I built a simple wood-and-metal apparatus that displayed a set of white, wooden visual forms: a circle, a cross, and a triangle. The forms were buoyant and backed with metal studs so that they could float and adhere to the apparatus. The apparatus consisted of a horizontal bar with a vertical bar coming down off each end. This was the setup for a match-to-sample task: one symbol (the sample) was displayed in the center of the horizontal bar, an identical (matching) symbol was on one vertical arm, and a different (nonmatching) symbol was on the other. A symbol was introduced as the sample, and Circe had to look at the sample and pick the correct match. Dolphins are adept at touching things with the tip of their rostrums (the front of their mouths), and Circe was no exception. The symbols were presented in air so she had to use vision, not echolocation, for the task.
Using this setup, I rewarded Circe with a piece of fish when she touched the bottom circle, the correct match to the sample, but not when she chose the triangle. I switched the positions of the circle and triangle at the bottom from time to time, in what we call a pseudo-random pattern, to make sure she wasn't using the symbol's position or any predictable pattern to solve the problem. Other animals are as clever like us and look for useful strategies to solve problems, like choosing what's on the right because that's what worked last time or trying an alternating pattern such as right, then left, then right. Circe was an eager student, and she learned the procedure fairly quickly. So now I knew that, indeed, Circe was able to identify shapes visually and could compare and distinguish between different forms.
The next procedure built on Circe's previous understanding of the match-to-sample task. The conditional-discrimination task was a bit more of a cognitive challenge, one that David Premack had pioneered with his chimpanzee Sarah. The task involved learning to associate a non-iconic visual symbol—that is, a symbol not visually similar to what it will be linked with—with a particular object; for example, a triangle with a toy ball. In this case, using the same match-to-sample apparatus as the first experiment, I put a ball in the top position on the keyboard, the sample position. In the two lower positions I had a triangle and a cross. For this test I rewarded Circe when she touched the triangle key, but not when she touched the cross; I did the same type of thing with the other objects. Again, Circe got it quickly, showing that, indeed, she had learned the concept of conditional discrimination. In the most parsimonious explanation and the driest technical description, what Circe learned was this: If the ball was in the sample position, then she should touch the triangle; if the ring was in the sample position, then she should touch the cross; and if the necklace float was in the
Then I conducted my first formal experiment: testing if a dolphin could visually discriminate between the white symbols. This was not a trivial question at the time; the everyday world of dolphins was thought to be dominated by sound, not sight, and, as I mentioned before, dolphins' echolocation abilities are exquisite. I therefore couldn't
assume
that dolphins could see well enough and discriminate well enough to use the visual symbols of the keyboard. Back in 1979 when I started with Circe, we did not know if they could. More than a decade earlier, at the Stanford Research Institute in Menlo Park, Winthrop Kellogg and Charles Rice had done some very preliminary studies on this kind of ability in dolphins, but they had gone only so far. Essentially, I was starting at square one. The first part of my doctoral research, therefore, was to discover whether dolphins could visually discriminate among simple shapes.
I built a simple wood-and-metal apparatus that displayed a set of white, wooden visual forms: a circle, a cross, and a triangle. The forms were buoyant and backed with metal studs so that they could float and adhere to the apparatus. The apparatus consisted of a horizontal bar with a vertical bar coming down off each end. This was the setup for a match-to-sample task: one symbol (the sample) was displayed in the center of the horizontal bar, an identical (matching) symbol was on one vertical arm, and a different (nonmatching) symbol was on the other. A symbol was introduced as the sample, and Circe had to look at the sample and pick the correct match. Dolphins are adept at touching things with the tip of their rostrums (the front of their mouths), and Circe was no exception. The symbols were presented in air so she had to use vision, not echolocation, for the task.
Using this setup, I rewarded Circe with a piece of fish when she touched the bottom circle, the correct match to the sample, but not when she chose the triangle. I switched the positions of the circle and triangle at the bottom from time to time, in what we call a pseudo-random pattern, to make sure she wasn't using the symbol's position or any predictable pattern to solve the problem. Other animals are as clever like us and look for useful strategies to solve problems, like choosing what's on the right because that's what worked last time or trying an alternating pattern such as right, then left, then right. Circe was an eager student, and she learned the procedure fairly quickly. So now I knew that, indeed, Circe was able to identify shapes visually and could compare and distinguish between different forms.
The next procedure built on Circe's previous understanding of the match-to-sample task. The conditional-discrimination task was a bit more of a cognitive challenge, one that David Premack had pioneered with his chimpanzee Sarah. The task involved learning to associate a non-iconic visual symbol—that is, a symbol not visually similar to what it will be linked with—with a particular object; for example, a triangle with a toy ball. In this case, using the same match-to-sample apparatus as the first experiment, I put a ball in the top position on the keyboard, the sample position. In the two lower positions I had a triangle and a cross. For this test I rewarded Circe when she touched the triangle key, but not when she touched the cross; I did the same type of thing with the other objects. Again, Circe got it quickly, showing that, indeed, she had learned the concept of conditional discrimination. In the most parsimonious explanation and the driest technical description, what Circe learned was this: If the ball was in the sample position, then she should touch the triangle; if the ring was in the sample position, then she should touch the cross; and if the necklace float was in the
Similar Books
