Lehrer M, Wehner R, Srinivasan M
J Comp Physiol A. 1985 Oct;157(4):405-15. doi: 10.1007/BF00615140.
Freely flying bees were rewarded with sugar solution on a variety of black-and-white shapes as well as on coloured gratings in various training situations. In subsequent dual-choice tests, the bees' discrimination between the various shapes was measured. In addition, the bees were video-filmed while flying in front of the shapes. The scanning patterns thus obtained were then quantified in order to characterize scanning behaviour and its relationship to the geometrical parameters of the scanned shapes, investigate whether scanning plays a role in pattern discrimination and examine the influence of training on the characteristics of scanning. The scanning patterns clearly mirror the contours of the scanned shape in all cases, i.e. the bees fly along the contours contained in the shape. This behaviour does not depend on whether the scanned shape is one that was previously rewarded, or one that is completely novel to the bees. Comparison of the results of quantifying the scanning patterns with the results of dual-choice tests reveals that scanning behaviour is independent of discrimination performance. On the average, horizontal scanning directions occur more often than vertical directions. Variations of the training situation produce measurable differences in scanning behavior. However, except in the case of vertical scanning on a vertical grating, these differences are quite small, indicating that following contours is a largely stereotyped behaviour. Horizontal gratings are very well discriminated from vertical ones even if they offer contrast to only one receptor type, i.e. blue or green, demonstrating that the direction of contours is visible to the pattern recognition system even under these conditions. However, vertical and horizontal coloured gratings offering only blue-contrast do not elicit contour-following, whereas gratings offering only green-contrast do. Thus, the bees' scanning behaviour is colour-blind and most probably governed by the green receptors. We suggest that contour-following is the by-product of a behavioural mode which serves to prevent retinal image movement during flight in front of a contoured visual pattern.
在各种训练情境中，自由飞行的蜜蜂会因落在各种黑白形状以及彩色光栅上而得到糖水奖励。在随后的双选测试中，测量蜜蜂对各种形状的辨别能力。此外，当蜜蜂在形状前飞行时对其进行视频拍摄。然后对由此获得的扫描模式进行量化，以表征扫描行为及其与被扫描形状几何参数的关系，研究扫描是否在图案辨别中起作用，并考察训练对扫描特征的影响。在所有情况下，扫描模式都清晰地反映了被扫描形状的轮廓，即蜜蜂沿着形状中包含的轮廓飞行。这种行为并不取决于被扫描形状是之前得到奖励的形状，还是对蜜蜂来说全新的形状。将扫描模式量化结果与双选测试结果进行比较发现，扫描行为与辨别能力无关。平均而言，水平扫描方向比垂直扫描方向出现得更频繁。训练情境的变化会使扫描行为产生可测量的差异。然而，除了在垂直光栅上进行垂直扫描的情况外，这些差异相当小，这表明沿着轮廓飞行在很大程度上是一种刻板行为。即使水平光栅仅与一种受体类型（即蓝色或绿色）形成对比，它们也能很好地与垂直光栅区分开来，这表明即使在这些条件下，轮廓方向对于图案识别系统也是可见的。然而，仅提供蓝色对比的垂直和水平彩色光栅不会引发沿轮廓飞行，而仅提供绿色对比的光栅则会。因此，蜜蜂的扫描行为是色盲的，很可能由绿色受体控制。我们认为，沿轮廓飞行是一种行为模式的副产品，这种行为模式有助于在面对有轮廓的视觉图案飞行时防止视网膜图像移动。
