An index of the changes in responsiveness of octopuses to moving figures can be obtained by studying a set of animals and recording the number of them that come out to attack on each occasion, and the delay. During the period of 1 to 2 h after feeding, untrained octopuses show an increased tendency to attack. This period of increased attacks corresponds to that for which food remains within the crop. However, such attacks do not result in the setting up in the memory of representations that promote later attacks. If food is given after showing a figure the tendency to attack is greatly increased. It remains at a high level for longer than when the food is given before showing, but ultimately declines. However, if the figure is presented again with food within 10 to 24 h the tendency to attack is raised still higher and declines more slowly. After a number of such presentations the animals come out regularly to attack the figure. The changes in tendency to respond after each showing of the figure with food thus provide an estimate of the time course of decay of the increased excitability in certain pathways in the nervous system. The half-life of the change after the first rewarded presentations is 1 to 2 h. Similarly it is possible to map the time course of the reduced tendency to attack that follows the giving of a shock after attack on a moving figure. This time depends on the extent to which the figure had previously been made 'positive' by association with food. By successive feeding and shocking the tendency to attack a given figure by a group of octopuses can thus be raised and lowered. The effects are partly general to all moving figures, but are greater for those that resemble the figure shown when the food (or shock) was administered. Using this effect animals can be trained without shocks to attack certain figures but not others, pathways 'representing' these figures having been specifically facilitated by feeding. In octopuses without vertical lobes the effects of food or shock in increasing or lowering the tendency to attack are similar to those in normal animals, but persist for a shorter time. In these animals lasting representations ensuring or preventing attack are less readily set up. When food is given as reward for attacks at one figure, shocks for a different one, normal octopuses learn within a few trials to attack the one and avoid the other. In animals without vertical lobes the effect of food is to increase the tendency to attack both figures and the effect of shocks is to depress the attacks on both. With alternate trials the tendency to attack thus swings up and down, the animals responding according to the influence of the immediately previous stimulus. The effect of the vertical lobe is therefore to ensure persistence in the memory of appropriate distinct representations of the figures and the associated food or shock, so that there are 'correct' responses after a few trials.