Читаем Shufflebrain полностью

There is an important principle in sensory physiology known by some as the psychophysical law. First suggested in the 1830's by Ernst Weber, it was perfected and verified in 1860 by philosopher-biologist, Gustav Fechner. (Today, some people call it the Weber-Fechner law.) Experimenting, Fechner found that a change in the strength of a stimulus up or down elicits a corresponding increase or decrease in the intensity of a sensation. If H is the change in the sensation, k a constant and S the change in the strength of the stimulus, then H equals k times the logarithm of S (H = klogS). The law may not hold close to threshold or at extremely high intensities, and it's easier to demonstate with some senses than others. But in the range where most sensations occur (with the possible exception of hearing), the law holds remarkably well. There was even speculation that learning fit into the eye-to-one rule. Thus a one-to-one rule ought to exist in changes of sensation to perception to learning--within limits, of course.

Law or no law, Carl insisted, sensory physiology had not settled a fundamental question: given the inherant capabilities of a brain, are the refining constraints imposed out at the sense organs or up in the brain. He'd thought the critical test was impossible until he ran across the eye transplant experiments made famous by Roger Sperry.[4] Adding and subtracting eyes was the approach, Carl thought.

After much procrastination on my part, we launched the study during the following spring. I'd conducted extensive pilot series, and had concluded that the best approach was to mount the extra eye on top of the animal's head just above the pineal body, the vestigial third eye. I'd cut a window in the top of the skill and then aim the stump of the optic nerve directly at the roof of the diencephalon. I called these animals, collectively, Triclops.

Our main controls were animals with an eye transplanted atop the head, like Triclops, but with both natural eyes removed. I tried calling them "monoclops"; then "uniclops"; but (to keep from swallowing my tongue during oral discourse) eventually went with "Cyclops." Cyclops would inform us not only whether but also when: whether the experiment was worth carrying to a conclusion, and, if so, when training could and should begin.

The one-to-one principle applies to increments of change, not static levels of sensation, perception or learning. (Thus, for example, if one pinch begets an ouch, two pinches won't necessarily bring forth two ouches. But the differences between one and two when added onto two should tell you what three pinches will elicit.) We had no way of knowing a priori just what those increments might possibly be in the visual system of the salamander larva. But if one-to-one works, then a normal salamander with two natural eyes would learn faster than a sibling with one eye removed. If the latter held up, the difference would equal our increment. The increment (difference between one- and two-eyed), when added to the two-eyed animals' scores would predict the performance Triclops--if one-to-one was valid. The latter statement became the basic prediction of our study.[5]

Carl, meanwhile, had developed and perfected the training apparatus and the evaluation routines, which he called the "light-shock avoidance test." In principle like the ding of the bell in Pavlov's experiments, a spot light provided the conditioned stimulus (CS is the standard abbreviation). The shock, the unconditioned stimulus (US) was what the animal had to learn to avoid: 10 volts of direct current at 10 Hz for 10 milliseconds. The rig itself was a marvel of simplicity and ingenuity: two low cylindrical dishes, one larger than the other by a little more than the width of anAmblystoma punctatum larva's body, the smaller inverted in the larger, thus creating a circular alley in which the animal would swim. Platinum wires around the two walls served as electrodes, the mediators of the shock. Carl would reposition the light wherever the animal stopped. The circular geometry of the alley meant that every starting point was of the same shape as any other. Animals would not have to be dragged back to a starting point as in a conventional apparatus.

In training, a salamander had 10 seconds to escape from the light before receiving a shock. Carl performed 25 trials one an animal, per session, 2 sessions a day, for 4 days. He randomly varied the intervals between trials from 10 to 25 seconds, to make sure the animals did not cue on the tempo instead of the light .