14
M. J. WELLS
diameter; these were treated as being of the diameter of their components. I n some transfer experiments it was found that textural
differences could be substituted for diameter differences. Thus grooved
cylinders were taken and smooth ones rejected where octopuses had
previously been trained to take the smaller of two smooth cylinders,
(0)
'b)
Compound
Single
0
0
6mm
8 mm
8 mm
) .
Random errors
c
0
L
-
0
.
75%correcl
0
Ln
0
c -
0\9
+ + + +
$ @ 0 $
,." + +
e *
O+q
++
-f
I
1
1
0 -
e * @ * * 0 ( 0
( 0 ( 0
(0
0
1 0
20
30
Difference in curvature meosured a5 l/r x 100
FIQ. 8. Showing that the proportion of errors made varies with the difference in surface
curvature of objects to be dktinguished Each point plotted shows the result of an
experiment in which an octopus waa trained to discriminate between two cylinders
of different diameter. The diameters of the pairs of cylinders are given (in m m )
along the abscissa, which also shows the difference between the cylinders in terms
of surface curvature. Plots
and 0 respectively show the results of experiments
with simple and compound cylinders ((a) and (b) above show the relative sizes of
the 6 and 8 mm cylinders of the two types), the stated diameter in each case being
the unit and not the overall diameter, which would appear to be irrelevant. (From
Wells, 1964a.)
which implies that the receptors used for diameter assessment and
textural detection are the same (Wells, 1964a). The anatomy of the
receptors in the suckers has been described by Graziadei (1962, 1964)
and a preliminary attempt to link these anatomical results with the
experimental results has been made in Wells (1964a).
Octopuses cannot be taught to distinguish between similar objects