1. A vertical longitudinal cut through the supraoesophageal part of the brain does not prevent discrimination between rough and smooth cylinders.
2. It does, however, prevent correct response by the opposite side after training on one side of the animal. Octopuses with intact supraoesophageal lobes, trained on one side, discriminate when tested with the same objects on the other side; split-brain animals do not.
3. In this situation, the performance of one side of the octopus can be used as a control for the effect of a brain lesion in the other.
4. Removal of large parts from the supraoesophageal brain does not prevent touch-learning. Discrimination continues until the lesion includes the whole of the subfrontal lobe on the side concerned.
5. Some of the small amacrine cells of the subfrontal/posterior buccal region must remain if learning is to occur. There are normally about 2.5x106 of these in each side of the subfrontal lobe; animals continued to discriminate between rough and smooth cylinders with as few as 10,000 of these cells remaining, but failed when none were left.
6. Octopuses with the brain intact or with the vertical lobes removed do not show marked rough/smooth preferences. In animals with large lesions to the inferior frontal system, however, there is a preference for rough, so that rough+/smooth- is an easier discrimination than vice versa.