3. THE EXPERIMENTAL ANALYSIS OF VISUAL BEHAYIOR
69
cardiac-conditioning simply reflects the advantages of an easier task for
the subject (heart-rate changes are not all-or-none as is the choice demanded by avoidance training) or whether a more interesting physiological explanation is available.
Since each eye of a fish sends visual information exclusively to the
contralateral tectum, we might expect incomplete behavioral equivalence
between opposite eyes. Indeed, the first relevant study by Sperry and
Clark (1949) showed that interocular transfer of a simple up vs. down
problem was poor or absent in the majority of subjects. However, Schulte
(1957) obtained high levels of pattern transfer with carp following extensive training sessions. Schulte found clear transfer failure only when
his fish were confronted via the untrained eye with distorted or rotated
versions of the training stimuli. These subjects could transpose their training experience while using the training eye, however. This delicate ability
to judge similarity between different sets of patterns might have been
disrupted by changing eye covers prior to transfer tests which, at first,
made some fish “neurotic.” This interpretation is made plausible by the
demonstration of McCleary (1960) that goldfish with eye covers may
fail interocular transfer of a simple discrimination that occurs readily
when blinders are not used.
Other studies, however, have demonstrated limits of interocular
transfer that cannot be attributed either to emotional disruption or to
the use of a suboptimal response criterion. For example, Ingle (1965)
trained goldfish to discriminate a striped from a random pattern, where
the stimuli also differed in color (red or green). When these fish were
tested with the same patterns, each appearing in opposite colors, they
could resolve the conflict by responding to either color or pattern differences. Subjects that were tested first via the trained eye responded on
the basis of the (more discriminable) pattern dderences, while those fish
initially tested via the second eye behaved in accord with the color differences. Therefore, one concludes that pattern information transfers less
well than color information in the goldfish. This method of comparing the
“transferability” of various visual discriminations could provide guidelines for eventual recording from units in the various interhemispheric
commissures of the fish brain.
Although the aforementioned study clearly proved the relative failure
of pattern transfer, other studies (Ingle, 1968a) argue that a total failure
can sometimes be obtained. This result might be more encouraging for
the physiologist who prefers all-or-none results. Goldfish trained to
discriminate vertical stripes from those rotated by 23” failed to show any
evidence of transfer, unlike successful controls trained with vertical vs.
52” rotation, although both groups were tested with the same stimuli
differing by 38”. Even transfer of a horizontal-vertical stripe discrimi-
69
cardiac-conditioning simply reflects the advantages of an easier task for
the subject (heart-rate changes are not all-or-none as is the choice demanded by avoidance training) or whether a more interesting physiological explanation is available.
Since each eye of a fish sends visual information exclusively to the
contralateral tectum, we might expect incomplete behavioral equivalence
between opposite eyes. Indeed, the first relevant study by Sperry and
Clark (1949) showed that interocular transfer of a simple up vs. down
problem was poor or absent in the majority of subjects. However, Schulte
(1957) obtained high levels of pattern transfer with carp following extensive training sessions. Schulte found clear transfer failure only when
his fish were confronted via the untrained eye with distorted or rotated
versions of the training stimuli. These subjects could transpose their training experience while using the training eye, however. This delicate ability
to judge similarity between different sets of patterns might have been
disrupted by changing eye covers prior to transfer tests which, at first,
made some fish “neurotic.” This interpretation is made plausible by the
demonstration of McCleary (1960) that goldfish with eye covers may
fail interocular transfer of a simple discrimination that occurs readily
when blinders are not used.
Other studies, however, have demonstrated limits of interocular
transfer that cannot be attributed either to emotional disruption or to
the use of a suboptimal response criterion. For example, Ingle (1965)
trained goldfish to discriminate a striped from a random pattern, where
the stimuli also differed in color (red or green). When these fish were
tested with the same patterns, each appearing in opposite colors, they
could resolve the conflict by responding to either color or pattern differences. Subjects that were tested first via the trained eye responded on
the basis of the (more discriminable) pattern dderences, while those fish
initially tested via the second eye behaved in accord with the color differences. Therefore, one concludes that pattern information transfers less
well than color information in the goldfish. This method of comparing the
“transferability” of various visual discriminations could provide guidelines for eventual recording from units in the various interhemispheric
commissures of the fish brain.
Although the aforementioned study clearly proved the relative failure
of pattern transfer, other studies (Ingle, 1968a) argue that a total failure
can sometimes be obtained. This result might be more encouraging for
the physiologist who prefers all-or-none results. Goldfish trained to
discriminate vertical stripes from those rotated by 23” failed to show any
evidence of transfer, unlike successful controls trained with vertical vs.
52” rotation, although both groups were tested with the same stimuli
differing by 38”. Even transfer of a horizontal-vertical stripe discrimi-