DAVID INGLE
68
IV. PERCEPTUAL EQUIVALENCE AND CHANGE IN SPATIAL
POSITION
For all vertebrates the retinofugal fibers project to midbrain or
thalamus in a spatially ordered manner preserving at the tectum or in
the striate cortex a mapping of visual space. Many workers have, therefore, been interested in measuring the extent of an animal's ability to
judge shapes as equivalent when they fall upon disparate parts of the
retina or are viewed by different eyes. Studies of the invariance of shape
recognition with changes in position provide a certain way of excluding
factors of retinal coding from considerations of mechanisms underlying
shape perception. Other possible transformations of a shape-size, orientation, and color-do not necessarily exclude mechanisms of equivalence
at the retinal level.
Cronly-Dillon et al. ( 1966) have demonstrated "intraretinal transfer"
in goldfish by an ingenious method. Since the dorsal and ventral brachia
of the goldfish optic tract diverge to innervate the upper and lower tectum, respectively (i.e., mediate shape discrimination in the upper and
lower halves of the field), cutting one branch restricts retinal input to
half of one tectum. Subjects trained to discriminate vertical from horizontal rectangles after one brachium had been cut could retain this discrimination several weeks later after the severed fibers had regenerated
and the second (formerly used) branch was then cut. This clearly proves
that shape equivalence of at least a simple order exists when different
retinal inputs are employed. However, it is likely that overlap of tectal
receiving neurons occurs at the horizontal margin of these two optic inputs since retinal receptive fields may be as large as 30"-40" of visual
angle.
A study by Ingle ( 1963) indicates that pattern-discrimination transfer may occur when two sets of retinal images are each projected to two
disparate tectal regions. Using a cardiac-conditioning method, goldfish
showed good transfer of a horizontal vs. vertical or horizontal vs. diagonal
stripe discrimination from a temporal training position to a nasal testing
position 120" rostrally. If we allow for the occasional eye movements,
ranging up to 30", the minimal intraretinal distance over which equivalence was demonstrated is a respectable 90". Whether or not more difficult shape discriminations ( e.g., circle vs. square) would transfer over
this distance is a more critical, but unanswered, question. Ingle noted,
rather surprisingly, that intraretinal transfer totally failed with the stripe
discrimination in 12 subjects trained and tested by a conditioned-avoidance method even though a red-green discrimination did transfer from
back to front in the same fish. It is not known whether the success with
68
IV. PERCEPTUAL EQUIVALENCE AND CHANGE IN SPATIAL
POSITION
For all vertebrates the retinofugal fibers project to midbrain or
thalamus in a spatially ordered manner preserving at the tectum or in
the striate cortex a mapping of visual space. Many workers have, therefore, been interested in measuring the extent of an animal's ability to
judge shapes as equivalent when they fall upon disparate parts of the
retina or are viewed by different eyes. Studies of the invariance of shape
recognition with changes in position provide a certain way of excluding
factors of retinal coding from considerations of mechanisms underlying
shape perception. Other possible transformations of a shape-size, orientation, and color-do not necessarily exclude mechanisms of equivalence
at the retinal level.
Cronly-Dillon et al. ( 1966) have demonstrated "intraretinal transfer"
in goldfish by an ingenious method. Since the dorsal and ventral brachia
of the goldfish optic tract diverge to innervate the upper and lower tectum, respectively (i.e., mediate shape discrimination in the upper and
lower halves of the field), cutting one branch restricts retinal input to
half of one tectum. Subjects trained to discriminate vertical from horizontal rectangles after one brachium had been cut could retain this discrimination several weeks later after the severed fibers had regenerated
and the second (formerly used) branch was then cut. This clearly proves
that shape equivalence of at least a simple order exists when different
retinal inputs are employed. However, it is likely that overlap of tectal
receiving neurons occurs at the horizontal margin of these two optic inputs since retinal receptive fields may be as large as 30"-40" of visual
angle.
A study by Ingle ( 1963) indicates that pattern-discrimination transfer may occur when two sets of retinal images are each projected to two
disparate tectal regions. Using a cardiac-conditioning method, goldfish
showed good transfer of a horizontal vs. vertical or horizontal vs. diagonal
stripe discrimination from a temporal training position to a nasal testing
position 120" rostrally. If we allow for the occasional eye movements,
ranging up to 30", the minimal intraretinal distance over which equivalence was demonstrated is a respectable 90". Whether or not more difficult shape discriminations ( e.g., circle vs. square) would transfer over
this distance is a more critical, but unanswered, question. Ingle noted,
rather surprisingly, that intraretinal transfer totally failed with the stripe
discrimination in 12 subjects trained and tested by a conditioned-avoidance method even though a red-green discrimination did transfer from
back to front in the same fish. It is not known whether the success with