8
M. J. WELLS
visual axes ; but the animals have alternative means of classifylng
shapes when consideration in these terms fails them. Indeed at the
present time it appears possible that octopuses (again like some other
animals, see Sutherland, 1964 ; Mackintosh, 1962, 1965) use a number of
means of classifying the things that they see. It is argued that part of
the problem of learning to make a discrimination is the selection of an
appropriate analysing mechanism from the repertoire available. In
reversal experiments, for example, octopuses, unlike certain other
invertebrates, tend to improve in performance in successive reversals ;
ex-hypothesis, they start relearning with an appropriate analyser
already " switched in." A similar explanation can be given of experiments in which learning one discrimination interferes with subsequent
performance in others that have to be made on a different basis
(Mackintosh and Mackintosh, 1963 ; Mackintosh, 1965).
Muntz (1961 a, b, c) has studied interocular transfer in discrimination experiments and shown that responses learned using one eye are
also made when the animal is obliged to use the other eye. The degree
of transfer depended on the degree to which the octopus had learned to
discriminate before the test; it was almost complete for simple problems (Muntz, 1961%). Transfer is dependent on the integrity of the
vertical lobe and the optic commissures linking the two optic lobes
(Muntz, 1961b-see section B below).
3. Discrimination of size and assessment of distance
Octopuses attack small moving objects and retreat when confronted
with unfamiliar moving objects larger than themselves. They can be
trained to distinguish between geometrically similar objects of different
size, and are not confused when the range of such objects is altered.
Thus, octopuses trained to discriminate between squares of Scm and
lOcm side still discriminated when the 5cm square was shown at half
the usual distance, and when the size of the tank was doubled to alter
the scale of the objects relative to their surroundings (Boycott and
Young, 1956). The fact that transfer to similar shapes of different size
decreases with any increase in the size difference itself implies quite
accurate recognition of particular sizes as properties of objects that
octopuses have learned to recognize (Sutherland and Carr, 1962).
Recognition of size implies an ability to range objects seen, a capacity
that has been independently demonstrated by Maldonado (1964) who
showed that the jump of an octopus at its prey is determined before it
takes off. The animal does not require to readjust its trajeotory
during the approach, and completes its jump accurately even if all light
is extinguished as soon as it begins to swim forward. Decapods such as
M. J. WELLS
visual axes ; but the animals have alternative means of classifylng
shapes when consideration in these terms fails them. Indeed at the
present time it appears possible that octopuses (again like some other
animals, see Sutherland, 1964 ; Mackintosh, 1962, 1965) use a number of
means of classifying the things that they see. It is argued that part of
the problem of learning to make a discrimination is the selection of an
appropriate analysing mechanism from the repertoire available. In
reversal experiments, for example, octopuses, unlike certain other
invertebrates, tend to improve in performance in successive reversals ;
ex-hypothesis, they start relearning with an appropriate analyser
already " switched in." A similar explanation can be given of experiments in which learning one discrimination interferes with subsequent
performance in others that have to be made on a different basis
(Mackintosh and Mackintosh, 1963 ; Mackintosh, 1965).
Muntz (1961 a, b, c) has studied interocular transfer in discrimination experiments and shown that responses learned using one eye are
also made when the animal is obliged to use the other eye. The degree
of transfer depended on the degree to which the octopus had learned to
discriminate before the test; it was almost complete for simple problems (Muntz, 1961%). Transfer is dependent on the integrity of the
vertical lobe and the optic commissures linking the two optic lobes
(Muntz, 1961b-see section B below).
3. Discrimination of size and assessment of distance
Octopuses attack small moving objects and retreat when confronted
with unfamiliar moving objects larger than themselves. They can be
trained to distinguish between geometrically similar objects of different
size, and are not confused when the range of such objects is altered.
Thus, octopuses trained to discriminate between squares of Scm and
lOcm side still discriminated when the 5cm square was shown at half
the usual distance, and when the size of the tank was doubled to alter
the scale of the objects relative to their surroundings (Boycott and
Young, 1956). The fact that transfer to similar shapes of different size
decreases with any increase in the size difference itself implies quite
accurate recognition of particular sizes as properties of objects that
octopuses have learned to recognize (Sutherland and Carr, 1962).
Recognition of size implies an ability to range objects seen, a capacity
that has been independently demonstrated by Maldonado (1964) who
showed that the jump of an octopus at its prey is determined before it
takes off. The animal does not require to readjust its trajeotory
during the approach, and completes its jump accurately even if all light
is extinguished as soon as it begins to swim forward. Decapods such as