Parallel spatial channels in the insect visual system

感觉系统——包括生物的和人工的——处理高度多维的信息。为了提取特定任务的相关特征，他们将复杂的感官输入过滤到平行通道中，并将其分解为可管理的部分。视觉系统以特别高的维度获取有关自然世界的信息。这使得它非常适合控制人类和其他动物的广泛行为，通常作为主导感觉。但信息的广度也增加了在早期阶段对感官过滤和视觉信号分类的需求，以便用有限的神经能力处理复杂的输入。一个重要的例子就是视觉系统中的平行空间通道。它们可以帮助以特定任务的方式解决视觉系统中空间敏锐度和对比敏感度之间的权衡。这种策略的行为和生理证据可以在许多脊椎动物物种中找到。它们的运动视觉路径牺牲了空间分辨率以获得高对比度灵敏度，而模式检测路径以牺牲对比度灵敏度为代价保留了高空间灵敏度。然而，在昆虫中，我们不知道这种平行的空间过滤器是否存在。这是令人惊讶的，因为具有微小大脑的昆虫是有限的神经能力如何控制令人印象深刻的行为复杂性的主要模型-有效的外周过滤是一项至关重要的壮举。在这个项目中，我计划通过描述昆虫的平行空间通道来缩小这一知识差距。最近，我在大舌蛾（Macroglossum stellatarum）身上获得的实验数据首次证明，候选神经元可以支持这些平行空间通道。在这个项目中，我们将从生理学上描述这些神经元，以了解它们如何促进空间信息处理。深入研究不同飞蛾行为的空间特征将揭示这种平行空间通道是否在行为上得到表达。与计算模型一起，这些方法将确定昆虫是否使用与脊椎动物相似的处理策略，通过将视觉输入过滤到平行的空间通道中来减少视觉输入的复杂性。我们希望这个项目的结果能够为昆虫视觉的理解带来范式转变，因为它们可能为迄今为止描述的视觉处理策略增加一个全新的组成部分。这些处理策略也将为人工感觉系统的发展提供有价值的见解，其中昆虫提供了强大的模型系统。此外，该项目还为跨动物门空间加工的比较研究奠定了基础，从而更好地了解视觉信息神经编码的一般策略。

Sensory systems – both biological and artificial ones – process highly multidimensional information. In order to extract relevant features for specific tasks, they filter the complex sensory input into parallel channels and break it down into manageable portions. The visual system acquires information about the natural world with a particularly high dimensionality. This makes it well suited to control a wide range of behaviours in humans and other animals, often as the dominant sense. But the breadth of information also increases the need for sensory filtering and for categorizing visual signals at an early stage, to process the complex input with limited neural capacities. An important example for this are parallel spatial channels in the visual system. They can help to resolve the trade-off between the spatial acuity and the contrast sensitivity in the visual system in a task-specific manner. Behavioural and physiological evidence for this strategy can be found in a number of vertebrate species. Their motion vision pathways sacrifice spatial resolution for high contrast sensitivity, while pattern detection pathways retain high spatial acuity, at the cost of contrast sensitivity. In insects, however, we do not know whether such parallel spatial filters exist. This is surprising, as insects with their tiny brains are the prime model for how limited neural capacity can control impressive behavioural complexity – a feat for which efficient peripheral filtering is crucial. In this project, I plan to close this knowledge gap by characterising parallel spatial channels in insects. Recent pilot data I obtained in the hawkmoth Macroglossum stellatarum provide the first evidence of candidate neurons that could support these parallel spatial channels. In this project, we will characterize these neurons physiologically, to understand how they contribute to spatial information processing. A thorough investigation of the spatial characteristics of different hawkmoth behaviours will reveal whether such parallel spatial channels are expressed behaviourally. Together with computational modelling, these approaches will establish whether insects use similar processing strategies as vertebrates to reduce the complexity of the visual input by filtering it into parallel spatial channels. We expect the results of this project to be paradigm shifting for the understanding of insect vision, as they might add a completely new component to the visual processing strategies described to date. These processing strategies will also provide valuable insights for the development of artificial sensory systems, for which insects provide powerful model systems. In addition, this project also lays the foundation for comparative investigations of spatial processing across animal phyla, and thereby a better understanding for the general strategies underlying neural coding of visual information.