3. THE EXPERIMENTAL ANALYSIS OF VISUAL BEHAVIOR
61
and blues from one another where reliance on brightness cues was excluded. Furthermore, Yager (1967, 1968) demonstrated the ability of
goldfish to detect additions of monochromatic light of any hue to pure
white light. The saturation functions so determined in three subjects approximated quantitative predictions that Yager had derived from an
"opponent color theory" model, itself based upon studies of human color
vision.
These lines of research-acuity and color measurements-off er twin
rewards: detailed interspecies comparisons of the psychophysical laws of
visual function and fruitful correlation with anatomy, biochemistry, and
electrophysiology. These studies help to confirm a faith in the existence
of discrete units that underly visual behavior. Yet further complexities
have simply been avoided thus far by psychologists: measurement of
"color constancy" or the interaction of color with form vision. In the
following sections we shall review other dimensions of fish vision for
which our image of discrete visual analyzers offers a still tenuous and
unfulfilled hypothesis.
11. RELATIVE DISCRIMINATION WEAKNESSES
The search for relative weaknesses in shape recognition that might be
attributed to limits of peripheral analysis of the visual image is exemplified by Mackintosh and Sutherland (1963), who found that goldfish took
longer to acquire a discrimination between 45" and 135" oblique rectangles than to distinguish equivalent shapes set horizontally and vertically (Fig. 1). Sutherland has argued ( 1968) that the ability to distinguish orientations of contours might require orientation-specific visual
units-somewhere within the afferent system-such as the units described
by Hubel and Wiesel (1962) in the cat visual cortex. Indeed, Westerman (1965) has reported that elongated receptive fields of units recorded
in the goldfish tectum do seem to fall more often along horizontal or vertical axes than along oblique axes. If goldfish have a paucity of units
sensitive to oblique contours, they may nevertheless notice the horizontal
or vertical extents of oblique edges. In fact, fish trained by Mackintosh
and Sutherland to discriminate vertical from horizontal continued during
transfer tests to distinguish shapes oriented at 30" from those set at 60"
If goldfish pay more attention to horizontal and vertical components, even
of such tilted shapes, it would be important to see whether fish initially
trained on the 30"/60" problem improve their performance when tested
with the 0"/90" pair.
61
and blues from one another where reliance on brightness cues was excluded. Furthermore, Yager (1967, 1968) demonstrated the ability of
goldfish to detect additions of monochromatic light of any hue to pure
white light. The saturation functions so determined in three subjects approximated quantitative predictions that Yager had derived from an
"opponent color theory" model, itself based upon studies of human color
vision.
These lines of research-acuity and color measurements-off er twin
rewards: detailed interspecies comparisons of the psychophysical laws of
visual function and fruitful correlation with anatomy, biochemistry, and
electrophysiology. These studies help to confirm a faith in the existence
of discrete units that underly visual behavior. Yet further complexities
have simply been avoided thus far by psychologists: measurement of
"color constancy" or the interaction of color with form vision. In the
following sections we shall review other dimensions of fish vision for
which our image of discrete visual analyzers offers a still tenuous and
unfulfilled hypothesis.
11. RELATIVE DISCRIMINATION WEAKNESSES
The search for relative weaknesses in shape recognition that might be
attributed to limits of peripheral analysis of the visual image is exemplified by Mackintosh and Sutherland (1963), who found that goldfish took
longer to acquire a discrimination between 45" and 135" oblique rectangles than to distinguish equivalent shapes set horizontally and vertically (Fig. 1). Sutherland has argued ( 1968) that the ability to distinguish orientations of contours might require orientation-specific visual
units-somewhere within the afferent system-such as the units described
by Hubel and Wiesel (1962) in the cat visual cortex. Indeed, Westerman (1965) has reported that elongated receptive fields of units recorded
in the goldfish tectum do seem to fall more often along horizontal or vertical axes than along oblique axes. If goldfish have a paucity of units
sensitive to oblique contours, they may nevertheless notice the horizontal
or vertical extents of oblique edges. In fact, fish trained by Mackintosh
and Sutherland to discriminate vertical from horizontal continued during
transfer tests to distinguish shapes oriented at 30" from those set at 60"
If goldfish pay more attention to horizontal and vertical components, even
of such tilted shapes, it would be important to see whether fish initially
trained on the 30"/60" problem improve their performance when tested
with the 0"/90" pair.