Tactile Discrimination of Surface Curvature and Shape by the Octopus

Octopuses can readily be taught to distinguish between objects that differ in surface texture, or in taste. But they cannot be taught to recognize the distribution of irregularities on the surface of objects, or to discriminate between objects that differ only in weight (Wells, 1961a, 19636, Wells & Wells, 1956, 1957). In visual experiments the animals can recognize the orientation of things that they see, provided that the statocysts are intact. If these are removed, a reflex mechanism governing the orientation of the retina is upset, and discrimination fails (Wells, i960). In experiments with a maze, in which octopuses were required to learn to make a detour in order to get food, correct orientation was found to depend upon the maintenance of visual contact with the partitions separating the octopus from its goal. The animals performed satisfactorily after removal of both statocysts but unilateral blinding led to systematic errors in which the octopuses regularly turned through 180 without learning that this invariably led to a failure to collect the reward (Wells, 19646). These results all suggest that octopuses are unable to use information about the position of parts of their own bodies (or their own orientation in space) as a basis for learned responses. It is arguable that this state of affairs is not altogether surprising in view of the extreme flexibility of the animal; that in a very flexible animal motor control must of necessity be hierarchical and proprioceptive information must be utilized locally, a condition perhaps common to all animals which lack jointed skeletons. In such creatures proprioceptive inputs giving details of bodily position probably never penetrate to levels of the central nervous system concerned with learned responses (Wells, 19616, 1963a). In apparent contradiction of these results it was found that octopuses can distinguish by touch between a sphere and a cube, a performance most readily explained by supposing that the animals take into account the relative positions of the arms or suckers used to grasp the objects. Transfer experiments, however, imply that this is not the basis on which the octopus makes the discrimination. Instead it seems that the animal must learn to recognize the presence of corners on the cube, not as right-angle changes in plane, but as textural irregularities. Thus a narrow rod, presented to animals already trained to distinguish between a sphere and a cube, was treated as equivalent to the cube. Indeed it appeared from the experiments that the narrow rod used was actually a ' better' cube, so far as the octopuses were concerned, since the proportion of errors made by the trained animals fell when it replaced the cube in a series of discrimination experiments (Wells, 1964a).