A Theory That Missed the Mark

Although many scientists must narrowly fail to make an important discovery, it is hard not to feel guilty for not having pushed oneself just that little bit harder. Early in my career as a psychologist, I began to study vision in the octopus. I chose this strange beast because it was an invertebrate; hence its visual system, though highly developed, has evolved from structures very different from that of vertebrates. I believed (perhaps rather naively) that by finding the differences between the ways in which octopuses and vertebrates classify and recognize two-dimensional forms, it would be possible to correlate them with differences in the neuroanatomy of the two kinds of visual systems and hence throw light on both. Prof. J.Z. Young of University College, London, who originally prompted me to work on octopuses, was at the time studying their neuroanatomy and behavior. I must admit to having other reasons for choosing the octopus. The only institute in Europe where it was possible to carry out behavioral experiments on them was the celebrated Stazione Zoologica on the hay of Naples, and that farseeing body the U.S. Navy was prepared to subsidize my annual pilgrimages there. Unless upset by fear or indigestion, the octopuses spent most of their time beneath a pile of bricks at one end of their individual tanks. Using a method devised by Young, I trained them by presenting a moving shape at the other end of the tank. I used two shapes. If an animal came out and grabbed one (the positive), it was rewarded with a small crab; if it took the other (negative), it received an electric shock. The number of trials needed on a given pair of shapes before the subjects came consistently to take the positive and avoid the negative formed a measure of how similar the shapes appeared to the octopus visual system. Further light could be thrown on how the animals were perceiving the shapes by running transfer tests at the end of training. For example, if animals trained to discriminate between a square and a triangle are then shown a diamond, they treat it in the way they have learned to treat the triangle—attacking it if they have been trained to attack the triangle and avoiding it if they have been trained to take the square. means that for the octopus the diamond resembles a triangle more than a square, although it is in fact a 45-degree rotation of a square. Octopuses are not the easiest of species to work with. Though the lids of their tanks were roped down, they often would succeed in climbing out and would slither away over the floor. If not retrieved within a few minutes, they had a disconcerting habit of dying before the tests were finished. They were sensitive to water temperature, becoming too dozy for behavioral experiments if it was too cold and too restless if it was too hot. A liter of water per minute had to be squirted into their tanks to keep the water oxygenated. When the water jets popped out from holes in the lids, they would spray all and sundry, sometimes short-circuiting the wiring needed to deliver shocks. It was necessary to lift the lids of the octopus tanks with great caution when about to perform an experiment, since they took delight in squirting seawater with unerring accuracy at the experimenter’s eye. And of course any close contact with them was likely to result in ajet of ink. I was comparatively penurious in those days and did not have sufficient changes of clothes to cope with this habit; indeed, as I wandered through Naples in the evening I was several times mistaken for a particularly dirty tramp. The generosity of Neapolitans might have provided a useful supplement to my U.S. Navy grants, but I meekly tried to hand the proffered money back, though not always successfully, for those offering it, however poor themselves, would argue that I needed it more than they did. In my first experiment I trained separate groups of animals to discriminate between a vertical and a horizontal rectangle, a horizontal and an oblique rectangle, and two oblique rectangles at 90 degrees to one another. The subjects readily mastered the discrimination between horizontal and vertical, but never learned to discriminate between the two opposite obliques. Their performance on the other two pairs was in between. On the basis of this and another experiment I put forward, with the temerity of youth, a theory of octopus vision. I suggested that each column of cells in the octopus retina was wired to a single cell in the brain and similarly each row to a single cell. These central cells would therefore count the vertical extent of a shape at each point along its horizontal axis, and its horizontal extent along its vertical axis. The theory accounted nicely for the failure to discriminate between two opposite oblique rectangles since the central cells give identical responses to these shapes. Moreover, it correctly predicted the results of a number of further experiments that I undertook. In two ways, however, the theory was seriously wanting. First, it gave no account of computations made on the output from the central units detecting vertical and horizontal extents. Second, it permitted the vertical extents of two separate shapes one above the other to be combined in the output (and similarly the horizontal extents). But for all its naivete the theory had one great merit. Soon after it was put forward, David Hubel and Torsten Wiesel published the first of their papers on single cells in the visual cortex. They had discovered cells that respond maximally to a bar or an edge. Just as in my own theory, they found that different cells responded to stimuli in different orientations. Now, if only I had been just a little bit brighter, I would have realized that it is not the extent of the interior of a figure that defines its shape, but rather the orientation of its contours at every point on its boundary. If I had thought that little bit harder, I could have inferred from behavioral data the existence in the brain of bar and edge detectors. This was one of the most important neurophysiological discoveries of recent times, and one subsequently marked by the award of a Nobel prize to Hubel and Wiesel. Without wishing in any way to detract from their achievement (which was followed up by a great deal of elegant and important work), it would have been nice if psychology on this occasion had beaten neurophysiology to the post. In fact I subsequently went on to disprove my own theory. I also found to my dismay that when I replicated my many octopus experiments on vertebrates such as rats and goldfish, the vertebrates classified shapes in. exactly the same way as the octopus. Given the differences in the origins of the two visual systems, this suggests that convergent evolution was at work and that there is a best way of classifying shapes. But 30 years later we are still ignorant of the mechanisms used by either group of animals. Professor of experimental psychology at the University of Sussex since 1965, Sutherland is the central subject of Simon Gray’s play Melon, which opened this year in London’s West End. it is based on Sutherland’s book Breakdown: A Personal Crisis and a Medical Dilemma, published in 1976, in which he recounted as both patient and professional observer the near-madness triggered by jealousy over his unfaithful wife. He attributes his recovery from manic-depressive illness to lithium rather than to the psychotherapy he received. Sutherland has published widely on perception, cognition and learning, and has been a visiting professor at MIT.

A Theory That Missed the Mark

Interested in reading more?

Become a Member of