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

动物必须有效地处理行为相关的刺激，以成功地与环境互动。神经元如何解码然后编码感觉信息，以及神经编码策略如何在连续的大脑区域中变化，仍然是神经科学的中心问题。跨感觉系统的研究表明，高阶大脑区域的表征更有效，因为单个神经元选择性地响应感觉输入的特定特征（即稀疏编码）。然而，与此同时，对感官输入的给定特征（例如，特定对象、面部）的鲁棒识别要求神经响应对应用于该特征的变换（例如，从不同视点观看对象所导致的旋转/平移）变得鲁棒。这种特征不变性在某种程度上与选择性相反，因为它要求神经元对属于同一“类别”的具有非常不同的时空特征的刺激（例如从不同的视角看一辆汽车）做出类似的反应。然而，在脊椎动物中，介导反应选择性和特征不变性之间权衡的细胞和网络机制知之甚少。我的建议的重点是阐明的机制，使电感觉中脑神经元在gymnotiform波型弱电鱼选择性地作出反应，但在一个不变的方式，以通信呼叫发生在同种之间的社会互动。这些鱼在它们的身体周围产生一个准正弦电场，并通过散布在它们皮肤上的一系列电感受器神经元来监测这个电场的扰动。在竞争或求偶行为期间，雄鱼倾向于发射其电场的短暂扰动，其由电场频率的瞬时增加组成，并且通常被称为“啁啾”。以前的研究表明，鱼类将稳健地响应这些啁啾，即使他们的刺激波形显示大的异质性。根据电生理学研究，该结果甚至更令人惊讶，所述电生理学研究表明，外周电感觉神经元对由啁啾产生的不同刺激波形的反应相当不同。因此，一个重要的问题是：动物如何将这些不同的波形识别为属于同一类别（即啁啾）？我们的初步数据表明，一些中脑电感觉神经元对啁啾产生的刺激波形有类似的反应，但对其他类别的刺激波形根本没有反应。使用在体内膜片钳记录，药理学和数学建模的组合，我建议揭示的细胞和网络机制，使这些神经元显示功能不变，但在同一时间选择性响应啁啾刺激。由于电感觉系统与其他第八神经系统（例如前庭/听觉）有许多解剖/生理相似之处，因此这项研究的结果很可能适用于其他系统。

Animals must efficiently process behaviorally relevant stimuli in order to successfully interact with their environment. How neurons decode and then encode sensory information, and the ways in which neural strategies for coding change across successive brain areas, remains a central problem in neuroscience. Studies across sensory systems have shown that representations in higher order brain areas are more efficient because individual neurons respond selectively to specific features of sensory input (i.e. sparse coding). Yet, at the same time, robust recognition of a given feature of sensory input (e.g. a particular object, face) requires that neural responses become robust to transformations applied to that feature (e.g. rotation/translation resulting from looking at the object from different viewpoints). Such feature invariance somewhat opposes selectivity as it requires that neurons respond similarly to stimuli with very different spatiotemporal characteristics that belong to the same “category” (e.g. a car looked at from different viewpoints). The cellular and network mechanisms that mediate the tradeoff between response selectivity and feature invariance are however poorly understood in vertebrates. My proposal focuses on elucidating the mechanisms that enable electrosensory midbrain neurons in gymnotiform wave-type weakly electric fish to respond selectively but in an invariant manner to communication calls that occur during social interactions between conspecifics. These fish generate a quasi-sinusoidal electric field around their body and monitor perturbations of this field through an array of electroreceptor neurons scattered on their skin. During either agonistic or courtship behaviors, male fish tend to emit brief perturbations of their electric fields that consist of a transient increase in the electric field frequency and are commonly referred to as “chirps”. Previous studies have shown that fish will robustly respond to these chirps even when their stimulus waveforms display large heterogeneity. This result is even more surprising in light of electrophysiological studies showing that peripheral electrosensory neurons respond quite differentially to different stimulus waveforms resulting from chirps. Thus, an important question then is: how does the animal recognize these different waveforms as belonging to the same category (i.e. chirps)? Our preliminary data shows that some electrosensory midbrain neurons respond similarly to stimulus waveforms resulting from chirps but not at all to stimulus waveforms from other categories. Using a combination of in vivo patch clamp recordings, pharmacology, and mathematical modeling, I propose to uncover the cellular and network mechanisms that enable these neurons to display feature invariant yet at the same time selective responses to chirp stimuli. Because the electrosensory system shares many anatomical/physiological similarities with other eighth nerve systems (e.g. vestibular/auditory), it is very likely that the results from this research will be applicable to other systems.