Abstract
An experiment was conducted to test a theory of shape discrimination in the octopus: the theory predicts that shapes differing only in their horizontal extents will be more readily discriminable than shapes differing only in their vertical extents. Sixteen octopuses were trained to discriminate between pairs of shapes which satisfied these conditions but whose members were not mirror-images of one another. The prediction was not confirmed. It was found that the direction in which shapes were moved did not affect performance except that where both shapes were moved in the direction of their points performance was better than where both shapes were moved against the direction of their points. In an attempt to investigate the properties on the basis of which shapes were discriminated, 20 different transfer tests with new shapes were carried out. The results strongly suggest that Ss had been discriminating the original shapes on the basis of the difference between the ratios length of contour/square root of area for the two shapes. It is shown that this hypothesis is consistent with earlier results, and is suggested that there is a second shape-analyzing mechanism of this kind at work in the brain of Octopus. (PsycINFO Database Record (c) 2006 APA, all rights reserved)