LEARNING BY MARINE INVERTEBRATES
11
appropriate left-right choices on reaching the end of the corridor. Most
individuals had to be shown a crab and led round the maze on one or
more occasions before they did it spontaneously, but once having made
a successful detour, an octopus rarely failed. Thus eleven animals, each
tested for twenty trials a t four runs per day starting with the trial at
which it first made a successful spontaneous run, together failed to
detour or went into the wrong compartment only five times in 220
trials. Both the time spent trying to get directly through the glass, and
the number of abortive entries into the corridor (animal returns to the
attack without completing the detour) became less as training continued
(Fig. 6). Having learned to detour to one side, the animals performed
I20
90
60
30
to
I
2
3
4 I k y s
Fro. 6. A summary of the performance of an octopus in successive trials of a detour
experiment in the apparatus shown in Fig. 6. Both the time taken to enter the
passage and the number of abortive entries into the passage declined as tasts
proceeded. (From Wells, 1964b.)
faultlessly first time when tested in detours towards the other. There
is evidence that the orientation of the animal towards its prey is
maintained entirely visually. Thus unilateral blinding produces systematic errors, the operated octopuses going into the wrong compartment
when required to detour towards the unblinded side, despite the positional disorientations that this involves (Fig. 6). Removal of the
statocysts does not prevent successful detouring (Wells, 1964b).
6. Learned and innate components i n visual responses
It seems to be particularly difficult to teach octopuses to ignore
crabs, and certain brain lesions that leave the animals still capable of
11
appropriate left-right choices on reaching the end of the corridor. Most
individuals had to be shown a crab and led round the maze on one or
more occasions before they did it spontaneously, but once having made
a successful detour, an octopus rarely failed. Thus eleven animals, each
tested for twenty trials a t four runs per day starting with the trial at
which it first made a successful spontaneous run, together failed to
detour or went into the wrong compartment only five times in 220
trials. Both the time spent trying to get directly through the glass, and
the number of abortive entries into the corridor (animal returns to the
attack without completing the detour) became less as training continued
(Fig. 6). Having learned to detour to one side, the animals performed
I20
90
60
30
to
I
2
3
4 I k y s
Fro. 6. A summary of the performance of an octopus in successive trials of a detour
experiment in the apparatus shown in Fig. 6. Both the time taken to enter the
passage and the number of abortive entries into the passage declined as tasts
proceeded. (From Wells, 1964b.)
faultlessly first time when tested in detours towards the other. There
is evidence that the orientation of the animal towards its prey is
maintained entirely visually. Thus unilateral blinding produces systematic errors, the operated octopuses going into the wrong compartment
when required to detour towards the unblinded side, despite the positional disorientations that this involves (Fig. 6). Removal of the
statocysts does not prevent successful detouring (Wells, 1964b).
6. Learned and innate components i n visual responses
It seems to be particularly difficult to teach octopuses to ignore
crabs, and certain brain lesions that leave the animals still capable of