3. THE EXPERIMENTAL ANALYSIS OF VISUAL BEHAVIOR
63
bowed-out base from a circle but easily discriminated this pie-wedge
shape from a triangle.
Early workers also noted directional biases in fish vision, but they
seldom used enough subjects to allow firm conclusions. Yet, in the light
of subsequent knowledge, it is interesting that Hager (1938) found minnows discriminating one vs. two stripes, or three vs. four stripes, more
readily when these patterns were oriented vertically. As Trevarthen
( 1968a) has emphasized, such frontally approached stimuli generate
expanding retinal images, which could be analyzed by retinal units sensitive to image motion. Two studies (Jacobson and Gaze, 1964; CronlyDillon, 1964) have revealed that goldfish ganglion cells sensitive to direction of motion are most often sensitive to horizontal movement of small
objects or edges. This mechanism might apply as well to Meesters' finding
(1940) that a single fish trained to distinguish large and small squares
transferred best to long vs. short rectangles when these shapes were horizontally oriented. That is to say, Meesters' subject was more sensitive
to image-expansion (measuring the distance between the two edges)
when the differences lay along the horizontal axis. An observation by
Saxena (1966) suggests that the trout may-in distinction-pay more
attention to size differences along the vertical axis. Her subjects failed
to transfer a size discrimination involving outline squares when either the
top or bottom side was removed (as did a subject tested in the same way
by Meesters, 1940). Unfortunately, control experiments involving removal
of a vertical side were not reported. It might be useful to pin down such
possible species differences in shape recognition, since we assume that
visual analyzing mechanisms are variously adapted to the ecology or
social behavior of the species. The trout, for example, must execute h e
distance judgments within the upper sagittal plane preparatory to jumping for an insect and might well profit from a mechanism sensitive to
distances along the vertical axis,
An electrophysiological study by Jacobson and Gaze (1964) provides
further information on directional bias within the goldfish visual system.
Retinal units were more often sensitive to nasalward-as opposed to
temporalward-movement of spots within the visual fields. This built-in
asymmetry might explain the observation by Harden Jones (1963) that
several species of fish would follow nasalward but not temporalward
rotation of a surrounding striped drum. Furthermore, Ingle (1967) has
demonstrated a behavioral correlate of this directional bias using a cardiac-conditioning method (Fig. 2 ) . When a small 2" spot-moving at
12" 1 sec in the lateral field-served as a conditioned stimulus paired with
shock, all subjects responded with a stronger cardiac deceleration during
nasalward motion of the stimulus. This stimulus was comparable to that
63
bowed-out base from a circle but easily discriminated this pie-wedge
shape from a triangle.
Early workers also noted directional biases in fish vision, but they
seldom used enough subjects to allow firm conclusions. Yet, in the light
of subsequent knowledge, it is interesting that Hager (1938) found minnows discriminating one vs. two stripes, or three vs. four stripes, more
readily when these patterns were oriented vertically. As Trevarthen
( 1968a) has emphasized, such frontally approached stimuli generate
expanding retinal images, which could be analyzed by retinal units sensitive to image motion. Two studies (Jacobson and Gaze, 1964; CronlyDillon, 1964) have revealed that goldfish ganglion cells sensitive to direction of motion are most often sensitive to horizontal movement of small
objects or edges. This mechanism might apply as well to Meesters' finding
(1940) that a single fish trained to distinguish large and small squares
transferred best to long vs. short rectangles when these shapes were horizontally oriented. That is to say, Meesters' subject was more sensitive
to image-expansion (measuring the distance between the two edges)
when the differences lay along the horizontal axis. An observation by
Saxena (1966) suggests that the trout may-in distinction-pay more
attention to size differences along the vertical axis. Her subjects failed
to transfer a size discrimination involving outline squares when either the
top or bottom side was removed (as did a subject tested in the same way
by Meesters, 1940). Unfortunately, control experiments involving removal
of a vertical side were not reported. It might be useful to pin down such
possible species differences in shape recognition, since we assume that
visual analyzing mechanisms are variously adapted to the ecology or
social behavior of the species. The trout, for example, must execute h e
distance judgments within the upper sagittal plane preparatory to jumping for an insect and might well profit from a mechanism sensitive to
distances along the vertical axis,
An electrophysiological study by Jacobson and Gaze (1964) provides
further information on directional bias within the goldfish visual system.
Retinal units were more often sensitive to nasalward-as opposed to
temporalward-movement of spots within the visual fields. This built-in
asymmetry might explain the observation by Harden Jones (1963) that
several species of fish would follow nasalward but not temporalward
rotation of a surrounding striped drum. Furthermore, Ingle (1967) has
demonstrated a behavioral correlate of this directional bias using a cardiac-conditioning method (Fig. 2 ) . When a small 2" spot-moving at
12" 1 sec in the lateral field-served as a conditioned stimulus paired with
shock, all subjects responded with a stronger cardiac deceleration during
nasalward motion of the stimulus. This stimulus was comparable to that