Key points are not available for this paper at this time.
Octopuses were trained to discriminate visually between different pairs of mirror image shapes. The successive method of discrimination training was used. It was found first that octopuses discriminate more readily between up‐down mirror images than between left‐right mirror images; secondly, that octopuses discriminate as readily between mirror image shapes as between non‐mirror image shapes of equivalent complexity. It was also shown that it is possible to train octopuses on a comparatively difficult discrimination by the method of making reward conditional upon an attack upon the positive shape presented upon its own instead of presenting reward with the positive shape every positive trial. Although initial learning by the former method is slower, a smoother learning curve is obtained by this method, and the final level of performance reached is higher.
Neil Sutherland (Mon,) studied this question.