LEARNING BY MARINE INVERTEBRATES
7
These findings are particularly interesting in view of the anatomical
studies of Young (1960a, 1962, 1964b) who has shown that both the
retinal elements and the dendritic fields in the superficial layers of the
optic lobes are arranged mainly horizontally and vertically. Moreover,
that the number of vertically oriented dendritic fields is substantially
less than those oriented horizontally, an observation clearly compatible
with the more accurate estimation of horizontal extents. It has,
moreover, been shown that octopus eyes are held in a fixed orientation
with respect to gravity, so that the retinal and dendritic arrays indeed
remain in a constant orientation with respect to things seen. When
retinal orientation is upset by removal of the statocysts, visual disFigure
Horizontal
Verticol
Proportion
woiection
proiection of correct
. .
responses
1
0 0 %
H n n
FIG. 4. Pairs of figures that octopuses can be trained to distinguish despite their
having identical projections. The first pair are as readily distinguished as simple
horizontal and vertical rectangles. Percentages show the proportion of correct
responses i n the &st 160 (pair 1) or 240 (pair 2) trials. (From Sutherland, 1969,
and Sutherland eL al., 1963.)
crimination of orientation fails, the animals continuing to behave as
though the retinae were still held correctly (Wells, 1960). The implication is that for once the basis of an internal sensory classifying
mechanism is both visible and to some extent comprehensible.
Perhaps unfortunately, it has since become apparent that classification in terms of horizontal and vertical extents is by no means the whole
story, for octopuses can be trained (albeit sometimes with difficulty) to
make a number of discriminations between objects that yield identical
horizontal and vertical projections (Fig. 4). The horizontal and
vertical axes in octopuses, as in most other animals (Sutherland,
196313, 1964), are evidently important compared with other possible
7
These findings are particularly interesting in view of the anatomical
studies of Young (1960a, 1962, 1964b) who has shown that both the
retinal elements and the dendritic fields in the superficial layers of the
optic lobes are arranged mainly horizontally and vertically. Moreover,
that the number of vertically oriented dendritic fields is substantially
less than those oriented horizontally, an observation clearly compatible
with the more accurate estimation of horizontal extents. It has,
moreover, been shown that octopus eyes are held in a fixed orientation
with respect to gravity, so that the retinal and dendritic arrays indeed
remain in a constant orientation with respect to things seen. When
retinal orientation is upset by removal of the statocysts, visual disFigure
Horizontal
Verticol
Proportion
woiection
proiection of correct
. .
responses
1
0 0 %
H n n
FIG. 4. Pairs of figures that octopuses can be trained to distinguish despite their
having identical projections. The first pair are as readily distinguished as simple
horizontal and vertical rectangles. Percentages show the proportion of correct
responses i n the &st 160 (pair 1) or 240 (pair 2) trials. (From Sutherland, 1969,
and Sutherland eL al., 1963.)
crimination of orientation fails, the animals continuing to behave as
though the retinae were still held correctly (Wells, 1960). The implication is that for once the basis of an internal sensory classifying
mechanism is both visible and to some extent comprehensible.
Perhaps unfortunately, it has since become apparent that classification in terms of horizontal and vertical extents is by no means the whole
story, for octopuses can be trained (albeit sometimes with difficulty) to
make a number of discriminations between objects that yield identical
horizontal and vertical projections (Fig. 4). The horizontal and
vertical axes in octopuses, as in most other animals (Sutherland,
196313, 1964), are evidently important compared with other possible