The role of colour in insect pattern vision

一个世纪前，卡尔·冯·弗里施提高了我们对动物感官世界的理解，在经典实验中证明了蜜蜂能看到颜色。许多昆虫和大多数其他动物一样，具有色觉，对颜色的感知不需要复杂的高级大脑功能。昆虫体型较小，并配备了复眼，可以模糊地观察到空间上复杂的世界。这些景观有多鲜艳，昆虫如何在视觉场景中识别五颜六色的物体、图案和形状？这些问题可以在蜜蜂身上得到很好的研究，蜜蜂是生态学、行为学和生理学研究中重要的昆虫模型系统。觅食蜜蜂和其他授粉昆虫一样，通常寻找特定的花朵来有效地收集花蜜和花粉。它们会在寻找过程中丢弃许多其他的花，不会接近或降落在这些花上。这种花朵的恒定性需要蜜蜂学习和识别花卉的特征，如颜色、图案和气味。这种学习能力可以被利用来很容易地训练蜜蜂使用人造食物来源，就像冯·弗里施所做的那样，不同的颜色或图案表示奖励的存在。使用这样的实验方法和新的工具来测量和量化蜜蜂感知的颜色，我将调查图案在蜜蜂看来是什么样子，如果它们是五颜六色的，蜜蜂是否以及何时看到或独立于图案学习颜色。我将测试的一个关键假设是，蜜蜂视觉系统中的图案和颜色视觉是否分离，例如，昆虫的图案视觉是否可能是色盲，以及在多大程度上可能是色盲，就像过去提出的那样。我还将通过蜜蜂的眼睛和大脑，使用光学和感官模型来观察花朵，并预测蜜蜂的花卉颜色和图案有多容易察觉和辨别。如果一朵花距离很远，需要从背景中挑选出来，那么检测是主要的视觉任务。在第一组实验中，我将训练蜜蜂识别放置在两个Y迷宫手臂之一的后壁上的图案。另一只胳膊是空的。在迷宫中，蜜蜂将不得不决定它在哪只手臂上看到了图案。到图案的距离决定了它投射到蜜蜂眼上的大小。只要蜜蜂能够检测到一种模式，它就会进入正确的手臂，并找到蔗糖奖励。将图案移得更远，当蜜蜂随机选择双臂时，就会达到检测极限。将确定各种双色图案的检测下限，这些图案由中央圆盘和周围的环或分为两半的圆盘组成，类似于通过蜜蜂低分辨率眼睛看到的主要类型的花卉展示。颜色将被系统地改变，以了解颜色色调和相对于背景的颜色对比度如何影响图案的可识别性。这些参数将被用于分析完整野花展示中的颜色和图案，我将使用紫外线敏感的多光谱成像来记录这些颜色和图案。图案元素在形状和相对大小上也将变化，以识别蜜蜂视觉系统中可能的过滤机制。在第二组实验中，蜜蜂将辨别这些模式。只有一个模式会得到奖励。如果蜜蜂能够区分它们，它们会表现出对奖励模式的明显偏好。这些结果将告诉我们，蜜蜂如何感知图案中的空间和颜色线索，以及哪些线索在某些图案中可能更重要。了解了蜜蜂如何进行模式检测和辨别，我将能够解释蜜蜂如何观看和记忆复杂的五颜六色的花卉展示。这项研究将提供新的见解蜜蜂的感知能力可能影响了花中复杂和五颜六色的展示的进化，旨在有效地向蜜蜂和其他具有良好色彩视觉和学习能力的昆虫传粉者宣传自己。

A century ago Karl von Frisch advanced our understanding of animal sensory worlds demonstrating in classic experiments that bees see colours. Many insects as most other animals have colour vision and the perception of colour does not require complex higher brain functions. Insects are small-sized and equipped with compound eyes that give blurred views of a spatially complex world. How colourful are these views and how does an insect recognise colourful objects, patterns and shapes in a visual scene? These questions can be well studied in bees, an important insect model system in ecological, behavioural and physiological research. Foraging bees, as other pollinating insects, usually look for particular flowers to efficiently collect nectar and pollen. They will dismiss many other flowers during their search and not approach or land on them. Such flower constancy requires that floral features, e.g. colours, patterns and odours are learned and recognised by bees. This learning ability can be exploited to easily train bees to artificial food sources, as done by von Frisch, where different colours or patterns signal the presence of reward. Using such experimental methods and new tools to measure and quantify colours as they are perceived by bees, I shall investigate how patterns look to a bee if they are colourful, whether and when a bee sees or learns colours independently from patterns. A crucial hypothesis that I will test is whether pattern and colour vision are segregated in the bee visual system, e.g. whether and to which extent pattern vision in insects may be colour-blind, as has been suggested in the past. I shall also look at flowers through bee eyes and brain using optical and sensory models and make predictions of how detectable and disciminable floral colours and patterns are for bees. Detection is the main visual task if a flower is at a large distance and needs to be picked out from the background. In the first set of experiments I will train bees to detect a pattern which are placed on the back wall of one of two Y-maze arms. The other arm is empty. In the maze the bee will have to decide in which arm it sees the pattern. The distance to the pattern determines the size with which it is projected onto the bee eye. As long as the bee can detect a pattern it will enter the correct arm and find a sucrose reward. Moving the pattern further away the detection limit will be reached when the bee will choose both arms randomly. Detection limits will be determined for a variety of two-coloured patterns that consist of a central disc and a surrounding ring or a disc divided in two halfs, resembling the major types of flower displays as seen through the low-resolution eyes of bees. Colours will be varied systematically to understand how colour hue and colour contrast against the background influence pattern detectability. These parameters will be used to analyse colours and patterns in displays of intact wild flowers that I will record using UV-sensitive multispectral imaging. Pattern elements will be also varied in shape and relative size for identifying possible filter mechanisms in the visual system of the bee. In a second set of experiments bees will discriminate between these patterns. Only one pattern will be rewarded. If bees can distinguish them, they will show a clear preference for the rewarded pattern. These results will tell us, how spatial and colour cues in a pattern are perceived by the bee and which ones may be more important in some patterns. Knowing how bees perform pattern detection and discrimination, I will be able to explain how bees see and memorise complex colourful flower displays. The study will provide new insights the bees' perceptual abilities may have influenced the evolution of complex and colourful displays in flowers that aim to advertise themselves effectively to bees and other insect pollinators with excellent colour vision and learning capacity.