The neural mechanisms that lead to sparse coding within the midbrain of weakly electric fish

当你阅读这篇文章时，你大脑中的神经元允许你区分组成单词和句子的各种字母，从而形成一个随着大脑区域的不同而变化的神经表示。虽然更多外围区域(例如视网膜)的神经元倾向于对感觉输入的物理属性(例如图像的亮度和对比度)做出反应，但大脑更中央区域(例如视觉皮质)的神经元倾向于对抽象特征(例如字母“d”而不是字母“b”)做出更多反应，尽管有明显的相似之处(例如，一个是另一个的镜像)。这被称为“反应选择性”，已经在从昆虫到人类的各种系统(视觉、听觉、体感、嗅觉)和物种中观察到。与此同时，大脑更中央区域的神经元对感觉输入的身份保持转换变得更耐受(例如，神经元对字母“d”的反应与大小或字体无关)。这被称为不变性，也可以在系统和物种中观察到。也许不变性最著名的例子是所谓的“珍妮弗·安妮斯顿”神经元，它对女演员的照片做出反应，而不是对另一个人的照片做出反应(例如，哈莉·贝瑞)。因此，需要在选择性和不变性之间取得平衡，选择性的目的是限制引起神经反应的刺激，不变性的目的是扩大引起神经反应的刺激。这样的平衡是绝对必要的，这样我们才能识别环境中的其他人和物体，从而使我们能够成功地与环境互动。然而，导致选择性和不变性的机制在脊椎动物中普遍知之甚少，对于两者之间如何实现平衡更是知之甚少。在这里，我建议使用神经记录和行为方法相结合的方法来研究这些问题，使用的是弱电鱼类的电感系统。这个模型系统与我们的听觉和视觉系统有许多相似之处，并具有良好的解剖学和电路特征的优势。此外，感官刺激很容易在实验室中被模仿，并将产生可靠的行为反应。使用允许我们从大型神经集合进行记录的新技术，我们将能够理解网络(即大脑中的神经电路是如何组织的)和内在(即单个神经元的各种组件如何)如何相互作用，以实现选择性和不变性之间的权衡。这项研究的结果有望进一步加深我们对大脑基本功能的理解，并可能被用来理解高等脊椎动物是如何在选择性和不变性之间实现权衡的。

As you are reading this text, neurons in your brain allow you to distinguish the various letters forming words and sentences, thereby forming a neural representation that changes from brain area to brain area. While neurons in more peripheral areas (e.g., your retina) tend to respond to the physical attributes of the sensory input (e.g., the luminance and contrast of the image), neurons in more central brain areas (e.g., your visual cortices) instead tend to respond more to abstract features (e.g., the letter “d” but not the letter “b”), despite obvious similarities (e.g., one is the mirror image of the other). This is known as “response selectivity” and has been observed across systems (visual, auditory, somatosensory, olfactory) and species ranging from insects to us. At the same time, neurons in more central brain areas become more tolerant to identity preserving transformations of sensory input (e.g., a neuron would respond to the letter “d” irrespective of size or font). This is known as invariance and has also been observed across systems and species. Perhaps the best-known example of invariance is the so-called “Jennifer Anniston” neuron that respond to pictures of the actress irrespective of viewpoint but not of another person (e.g., Halle Berry). There thus needs to be a balance between selectivity, which aims at restricting the stimuli that give rise to neural responses and invariance, which instead aims at expanding the stimuli that give rise to neural responses. Such a balance is absolutely essential in order for us to recognize other people and objects in our environment, thereby allowing us to successfully interact with our environment. However, the mechanisms that lead to selectivity and invariance are poorly understood in general in vertebrates, and even less is known about how a balance between both is achieved. Here I propose to study these using a combination of neural recordings and behavioral approaches using the electrosensory system of weakly electric fish. This model system shares many similarities with our auditory and visual systems and has the advantage of well-characterized anatomy and circuitry. Moreover, sensory stimuli can easily be mimicked in the laboratory and will give rise to reliable behavioral responses. Using new technology that allows us to record from large neural ensembles, we will be able to understand how network (i.e., how neural circuits are organized in the brain) versus intrinsic (i.e., how various components of the single neuron) interact to achieve a tradeoff between selectivity and invariance. The results of this research are expected to further our understanding of basic brain function and could potentially be used to understand how the tradeoff between selectivity and invariance is achieved in higher vertebrates.