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

了解自然刺激如何被外周感觉神经元编码，并随后被更高级的结构（神经编码）解码，是神经科学的一个核心问题。该研究计划的长期目标是了解使感觉神经元在对自然感觉输入的反应中更具选择性的细胞和网络机制。人们普遍认为，外周感觉神经元使用密集的神经编码（即神经元对广泛的刺激作出反应的编码），而高级区域的神经元使用稀疏的神经编码（即神经元对自然刺激中稀疏分布的信息成分作出选择性反应的编码）。这些稀疏编码现已在动物分类群中被广泛观察到，包括人类的海马、猴子的视觉皮层、蝗虫的嗅觉系统和鸣禽的前脑。有人认为，这种稀疏编码更有效，因为单个神经元检测到感觉输入的特定特征，并且代谢成本低于密集编码。同时，识别一个给定的刺激特征（例如，一个特定的物体，人脸）需要神经反应对与该特征相关的有时非常不同的输入（例如，从不同的角度观察到的同一个物体）变得强大。

Understanding how natural stimuli are encoded by peripheral sensory neurons and subsequently decoded by higher structures, the neural code, is a central problem in neuroscience. The long-term goals of the proposed research program are to understand the cellular and network mechanisms that make sensory neurons more selective in their responses to natural sensory input. It is generally agreed that peripheral sensory neurons employ dense neural codes (i.e. codes in which neurons respond to a wide range of stimuli) and that neurons in higher areas employ sparse neural codes (i.e. codes in which neurons respond selectively to the sparsely distributed informative components of natural stimuli). These sparse codes have now been widely observed across animal taxa, including the human hippocampus, monkey visual cortex, locust olfactory system, and songbird forebrain. It is argued that such sparse codes are more efficient because individual neurons detect specific features of sensory input and are metabolically less costly than dense codes. At the same time, recognition of a given stimulus feature (e.g. a particular object, face) requires that neural responses become robust to the sometimes very different inputs associated with this feature (e.g. the same object observed from different viewpoints).