From sensory ecology to cognition - the multi-sensory perception and representation of objects in fish

大多数动物使用来自多种感官的信息来分析其环境的特征并指导行为。感觉生物学中的一个基本问题是多个感觉系统如何一起工作以增加整体感知的可靠性。在我们之前的项目中，我们已经证明了弱电鱼类Gnathonemus petersii能够进行跨模态对象识别，即它们可以跨模态地识别对象，而无需事先训练。这表明，在学习之后，包含物体特征的表征存储在大脑中，随后可以被其他感官访问。此外，我们表明，多感官对象感知的影响，由一个动态加权的感官输入，即鱼的重量输入从单一的感官形式，根据其可靠性，以尽量减少不确定性。这种感觉输入的权重也可以导致在特定情况或任务中某种感觉的主导地位（感觉捕获）。 在这个项目中，我们想研究脊椎动物多感官知觉的基本过程。利用G.作为一个模型，我们将探讨规则和控制跨模态物体识别和感觉输入加权的神经基质。我们将使用特别设计的物体，这些物体在通过视觉或电感觉感知时提供不同的信息。在行为的一部分，我们将调查是否Gnathonemus整合信息在一个最佳的方式，即是否感官加权依赖于特定的感官模态的可靠性。通过改变训练条件，我们将研究多感官输入的动态加权是否受先前经验的影响，或者是感知系统的固有属性。此外，我们将探讨如何感官信息加权，如果不同的感官提供冲突的信息在某些定义的可靠性水平。 在神经解剖学部分，我们的目标是寻找跨模态物体识别和感觉加权的神经基质。首先，我们将选择性示踪剂注入视觉和电敏感的感觉脑区，以确定上行感觉系统，并能够解释单感觉和多感觉会聚。在物体检测过程中，相关大脑区域的主要候选者是小脑、视顶盖（中脑）和软腭（前脑）的部分。除了追踪研究之外，我们还将进行旨在揭示多感觉学习和跨模态迁移过程中神经元激活过程的实验。这是通过可视化鱼脑中激活标志物（磷酸化外部信号相关激酶，pERK）的产生来实现的。我们将定义在多感官物体检查期间活跃的单感官和多感官部位。这可能最终告诉我们，感觉加权在G.彼得西。

Most animals use information from multiple senses to analyse features of their environment and to guide behaviour. A fundamental question in sensory biology is how multiple sensory systems operate together to increase the reliability of the overall perception. In our previous project we have shown that the weakly electric fish Gnathonemus petersii is capable of cross-modal object recognition, i.e. they can recognise objects cross-modally without prior training with the sense being tested. This suggests that after learning, a representation containing object features is stored in the brain and can subsequently be accessed by other senses. Furthermore we showed that multi-sensory object perception is influenced by a dynamic weighting of sensory inputs, i.e. fish weight input from single sensory modalities according to their reliability to minimise uncertainty. This weighting of sensory inputs can also lead to the dominance of a certain sense during specific situations or tasks (sensory capture). In this project we want to investigate the fundamental processes that underpin multi-sensory perception in vertebrates. Using G. petersii as a model, we will explore the rules and the neural substrates governing cross-modal object recognition and weighting of sensory input. We will use specially designed objects, which provide different information when sensed either through vision or the electric sense. In the behavioural part, we will investigate whether Gnathonemus integrates information in an optimal fashion, i.e. whether sensory weighting depends on the reliability of the particular sensory modality. By varying the training conditions we will examine whether dynamic weighing of multisensory input is influenced by prior experience or is an innate property of the perceptual system. In addition we will explore how sensory information is weighted if the different senses provide conflicting information at certain defined reliability levels. In the neuroanatomical part, we aim at finding the neural substrates for cross-modal object recognition and sensory weighting. First, we will inject selective tracers into visual and electro-sensitive sensory brain areas to define the ascending sensory systems and to be able to interpret unisensory and multisensory convergence. Prime candidates for relevant brain areas during object detection are parts of the Cerebellum, the Optic Tectum (midbrain) and the Pallium (forebrain). In addition to the tracing studies, we will perform experiments aimed at revealing neuronal activation processes during multi-sensory learning and cross-modal transfer. This is achieved by visualizing the production of activation markers (phosphorylated external signal related kinase, pERK) in the fish brain. We will define the uni- and multi-sensory sites which are active during multi-sensory object inspection. This might ultimately tell us, where and how sensory weighting takes place in the brain of G. petersii.