1. Octopuses from which the vertical lobe had been removed attacked a vertical rectangle more readily than a horizontal one when both shapes were moved vertically.
2. During discrimination training the results differed markedly according to which figure was rewarded with food.
3. When a vertical rectangle was positive the excess of attacks at this figure increased both within and between sessions, the animals showing considerable ability to learn.
4. When a horizontal rectangle was the positive figure the responses to it were at first ‘paradoxical’, the negative shape being attacked more often. There was then improvement within each session, but little if any between the sessions.
5. Octopuses without the vertical lobe quickly ceased to attack crabs shown at 4 min. intervals when shocks were given for attacks, and the effect could still be detected 4 hr. later.
6. When crabs were shown at 8 min. intervals, and shocks given for attacks, there was no decline in attacks over fourteen trials. Fifteen hours later, however, some residual signs of the shocks were evident when trials were resumed at 8 min. intervals. Although all of the animals at first attacked they soon stopped doing so. After a further 15 hr. all attacked again but soon ceased to do so after receiving shocks.
7. Normal animals, by contrast, all ceased to attack the crabs shown at 8 min. intervals after receiving four or five shocks. The memory is much more retentive in normal animals, so that 2 days after receiving shocks the octopuses seldom attacked the crabs.