Fig. 4.4. Percentage of choices for the rewarding standard color stimulus in tests against the background alone ("Back.") and against each of two alternative color stimuli ("Alt. 1" and "Alt. 2"). The dashed line at 50% indicates random choice. Values in parentheses depict the total number of choices recorded in each test situation. The visual angle subtended by the stimuli was (a) 30° or (b) 6.5°. With either visual angle, bees discriminated between the arm containing the standard colored disc and the arm containing the background alone. However, discrimination of the standard from each of the two alternatives, 1 and 2, differed significantly between the two experiments.
eye. Nor can we conclude that green contrast is used exclusively at small visual angles. Before such conclusions can be drawn, one must investigate the independent roles of chromatic and achromatic cues at small as well as at large visual angles.
To this end, individual bees were rewarded at a colored disc (henceforth termed "standard") that produced both chromatic and achromatic (green) contrast against the gray background (Giurfa et al. 1997). Bees were first trained to discriminate the standard from the background alone, as described above. The standard was then presented alternately against two different stimuli. Alternative 1 differed from the standard in color contrast, but it contained a similar amount of green contrast. Alternative 2 differed in the amount of green contrast, but not in the chromatic contrast that it produced against the background (for details of the colors used, see Giurfa et al. 1997). The visual angle subtended by the stimuli was 30° in one experiment and 6.5° in another.
Bees trained and tested with stimuli subtending 30° detected the standard against the background alone, and chose it correctly when it was presented against Alternative 1, i.e., they used chromatic cues (Fig. 4.4a). At the same visual angle, bees chose the standard at random when it was tested against Alternative 2, i.e., the amount of green contrast played no role.
When the visual angle subtended by the stimuli was 6.5°, bees still detected the standard against the background alone (Fig. 4.4b). In the tests against the alternative stimuli, however, their performance was reversed compared to the previous experiment. They now discriminated the standard from Alternative 2, which differed from it in green contrast but not in chromatic contrast, and were incapable of discriminating the standard from Alternative 1, which differed from it in chromatic contrast, but not in green contrast.
Thus, bees use either chromatic or achromatic cues, depending on the visual angle subtended by the stimuli. At small visual angles, i.e., at a distance at which detection first occurs, bees rely on the achromatic signal provided by the green contrast, and not on chromatic information. At large visual angles, they rely on chromatic information. At these large angles, achromatic cues are ignored, even when they are, in principle, available to the visual system. These conclusions are in accordance with the results of earlier studies on color discrimination in bees (for selected references, see e.g., Menzel & Backhaus 1991), all of which were conducted using colored stimuli viewed at a very close range and thus subtending large angles at the bee's eye.
Bees discriminate among different amounts of green contrast
Most natural flowers produce green contrast against the background, because all natural backgrounds (foliage, soil, rocks) are green or gray, whereas animal-adapted flowers are hardly ever green. If the bee's performance in the detection phase were to depend only on the presence of green contrast, then bees arriving at the feeding site would be expected to steer towards any flower. Only an ability to learn a particular amount of green contrast would ensure that the bee steers towards the particular flower that it has memorized on a previous visit.
The question of whether or not bees possess this ability was investigated by training bees to a standard stimulus that was later tested against three alternative stimuli, all of which differed from it in the amount of color contrast (Giurfa et al. 1997). At large visual angles, bees preferred the standard against each of the three alternatives, i.e., their choices were based on color discrimination. At small visual angles, however, the bees discriminated the standard from the two alternatives that presented either a higher or a lower amount of green contrast, but not from the alternative that matched the standard in the amount of green contrast. Therefore, when a rewarding stimulus containing green contrast is viewed at a small visual angle, bees not only detect it, they also recognize the particular amount of green contrast contained in it. Bees can thus discriminate the learned stimulus from other stimuli as soon as they can detect it.
The use of achromatic spatial cues contained in black-and-white stimuli
Black, white, and gray stimuli evoke equal amounts of excitation in all three spectral types of photoreceptor. The comparison among the three photoreceptor excitations (see section "The bee's color vision system," above) thus renders no differences among them, and therefore no color is perceived. Such achromatic stimuli can be discriminated only on the basis of intensity differences (intensity defined as the sum of the three excitations).
The role of intensity contrast in resolution and discrimination of achromatic stimuli has been demonstrated in many studies (see Wehner 1981 for review). Even in object-detection tasks, intensity contrast is crucial when achromatic stimuli are used (Lehrer & Bischof 1995). Most of the studies on pattern discrimination in the honeybee were conducted using black-and-white stimuli, because these stimuli produce the highest possible amount of intensity contrast. We shall restrict our attention, as we did in our first section, to studies that compare performance at different visual angles. Experimental results allow such a comparison with respect to three spatial parameters: (1) pattern disruption (spatial frequency); (2) contour orientation (spatial alignment); and (3) symmetry.
Pattern disruption and its role in pattern-
Natural flowers differ in their degree of outline disruption, i.e., in the amount of edge per unit area (also referred to as "contour density"). This parameter is a good candidate to serve the bee in pattern-discrimination tasks. For example, a flower with six petals will appear more disrupted to the eye of a flying bee than a flower of the same area with only three petals. Indeed, all of the earlier workers on bee spatial vision have agreed that the degree of disruption constitutes the main spatial parameter that bees use for pattern discrimination (Hertz 1929, 1933; Zerrahn 1933; Wolf & Zerrahn-Wolf 1935; Free 1970; Anderson 1977). It was suggested that the bee measures contour density in terms of the average frequency of intensity fluctuations (on- and off-stimulation, also termed "flicker") that the photoreceptors experience: more disrupted patterns produce a higher temporal frequency of intensity fluctuations than do less disrupted patterns.
In most of the early studies, investigators presented the patterns on a horizontal plane, recording the bee's choices at a very close distance from the patterns. When given a choice among novel patterns, bees preferred highly disrupted patterns over less disrupted ones (for references, see von Frisch 1967, Wehner 1981). However, patterns presented on a horizontal plane may change their appearance from one visit to the next, depending on the direction of the bee's approach. In this situation, contour density, a cue that is largely independent of the bee's direction of approach, is likely to be more reliable than any other spatial parameter.
To demonstrate the bee's use of spatial parameters other than disruption it was therefore necessary to present patterns on vertical planes, thus forcing the bees to arrive at the pattern from a constant direction. Even in the vertical plane, though, contour density was found to constitute an effective discrimination cue. Bees previously trained to a sectored disc of a particular spatial frequency discriminated that disc, in subsequent tests, from each of a series of novel discs that differed from it in spatial frequency. The greater the difference in disruption between the trained disc and the novel disc, the better the discrimination (Wehner 1981).
Presenting patterns in a vertical plane allows an experimenter to examine the bee's choice behavior towards disrupted patterns even at some distance from the patterns. To this end, Lehrer et al. (1995) used an experimental set-up consisting of 12 identical arms opening into a central arena from which bees had access to any of the arms (Fig. 4.5a). During training, six checkerboard patterns (Fig. 4.5b) were presented, one at a time in a quasi-random sequence, on the back wall of one the arms, the access to the reward box being through a tube penetrating the center of the pattern. The six checkerboards were randomized with respect to contour density, so bees could not memorize any particular spatial frequency. In subsequent tests, each of the 12 arms had a novel pattern placed on its back wall. In each test, a set of four patterns was used, each pattern being repeated in three different arms. The criterion for a choice was a bee entering an arm, i.e., 30 cm from the pattern. Regardless of the type of pattern, the lowest spatial frequency was the most attractive one (Fig. 4.5c, d), which is in contrast to the results of the earlier workers, who found a spontaneous preference for high-frequency patterns (see above).
The results indicate that, when patterns are viewed at a close range, bees prefer high spatial frequency, regardless of whether the stimuli are presented on a horizontal or vertical plane. At greater ranges, however, global cues - those provided by the pattern as a whole - dominate (e.g., Zhang et al. 1992; Lehrer et al. 1995; Horridge 1997). In this case, low spatial frequency imparts more accurate shape information than does high frequency.
Although it makes no difference whether patterns are presented on a horizontal or a vertical plane for discrimination of pattern disruption (see previous section), the mode of presentation is crucial when cues other than disruption are involved. Using patterns presented on a vertical plane, Wehner (1972) trained bees to a half-white and half-black disc, with the edge between the white and the black halves oriented at 45° with respect to the vertical. In subsequent dual-choice tests, bees discriminated
Was this article helpful?