I) Young Sepia, kept in filtered seawater from the moment of hatching until they have absorbed most of the yolk within them, will always attack and eat Mysis when these are presented. 2) When fed daily on Mysis, no change in the movements made in attacking was observed, but the delay before reacting when the mysis was presented decreased from about two minutes on the first occasion to 5-10 seconds on the fifth and subsequent occasions. 3) The length of the delay before attacking was shown to depend upon the number of attacks already made, regardless of the age of the animal or whether it had been allowed to feed when it attacked. 4) Coincident with this increase in speed of attack on mysids. there was an increase in the probability of attack on a variety of other moving objects. ranging from fixed mysids to strips of celluloid, that were at first rarely or never attacked. 5) These results are interpreted as indicating a facilitation process occurring somewhere between the retina and the motor centres, such facilitation leading to a higher probability of attack in a given visual situation and an incidental increase in the range of patterns of stimulation that can evoke attacks. This situation is transient and the animal soon begins to learn to restrict attacks to certain patterns only. 6) The capacity to learn appears to be correlated with the development of the vertical lobe of the brain, a part known to be intimately concerned with learning processes in adult Cephalopods, which is poorly developed in newly hatched Sepia