Espinosa Irene, Cuthill Innes C
School of Biological Sciences, University of Bristol, Bristol, United Kingdom.
PLoS One. 2014 Jan 22;9(1):e87153. doi: 10.1371/journal.pone.0087153. eCollection 2014.
Camouflage is the primary defence of many animals and includes multiple strategies that interfere with figure-ground segmentation and object recognition. While matching background colours and textures is widespread and conceptually straightforward, less well explored are the optical 'tricks', collectively called disruptive colouration, that exploit perceptual grouping mechanisms. Adjacent high contrast colours create false edges, but this is not sufficient for an object's shape to be broken up; some colours must blend with the background. We test the novel hypothesis that this will be particularly effective when the colour patches on the animal appear to belong to, not merely different background colours, but different background objects. We used computer-based experiments where human participants had to find cryptic targets on artificial backgrounds. Creating what appeared to be bi-coloured foreground objects on bi-coloured backgrounds, we generated colour boundaries that had identical local contrast but either lay within or between (illusory) objects. As predicted, error rates for targets matching what appeared to be different background objects were higher than for targets which had otherwise identical local contrast to the background but appeared to belong to single background objects. This provides evidence for disruptive colouration interfering with higher-level feature integration in addition to previously demonstrated low-level effects involving contour detection. In addition, detection was impeded in treatments where targets were on or in close proximity to multiple background colour or tone boundaries. This is consistent with other studies which show a deleterious influence of visual 'clutter' or background complexity on search.
伪装是许多动物的主要防御手段,包括多种干扰图形-背景分割和物体识别的策略。虽然匹配背景颜色和纹理很常见且概念上很简单,但利用感知分组机制的光学“技巧”(统称为破坏色)却较少被深入研究。相邻的高对比度颜色会产生虚假边缘,但这不足以破坏物体的形状;一些颜色必须与背景融合。我们测试了一个新假设,即当动物身上的色斑看起来不仅属于不同的背景颜色,而且属于不同的背景物体时,这种伪装会特别有效。我们使用基于计算机的实验,让人类参与者在人造背景上寻找隐藏的目标。在双色背景上创建看似双色的前景物体,我们生成了具有相同局部对比度但位于(虚幻)物体内部或之间的颜色边界。正如预测的那样,与看似不同背景物体匹配的目标的错误率高于与背景具有相同局部对比度但看似属于单个背景物体的目标。这为破坏色除了之前证明的涉及轮廓检测的低级效应外,还会干扰高级特征整合提供了证据。此外,在目标位于多个背景颜色或色调边界上或附近的处理中,检测受到阻碍。这与其他研究一致,这些研究表明视觉“杂乱”或背景复杂性对搜索有有害影响。