Neural limits of visual discrimination

描述（由申请人提供）：神经信号传导采用许多随机过程（递质释放、受体结合、通道开放），这些过程可能会增加噪声，从而丢失信息。损失已经通过测量神经效率、人类和“理想观察者”的灵敏度的比率来评估，“理想观察者”是包括人类前神经损失（模糊、光子噪声等）的计算模型。但没有神经损伤神经效率似乎最多达到约0.5，这意味着相当大的损失（50%）;然而，神经效率可能要高得多（约1），并被低估，因为测试刺激不匹配的神经通路，介导的歧视。 为了测试这些替代假设，AIM 1将测量“活跃”神经节细胞的神经效率，这些神经节细胞通过精确的尖峰定时进行编码并为膝状体-纹状体系统提供能量。我们记录细胞对最佳刺激的反应（即，匹配的时空脉冲响应），然后我们比较细胞的敏感性，一个理想的观察者只受前神经损失。神经节细胞效率（这些测量的比率）很可能>0.5。我们还将记录相邻的活跃神经节细胞对，并使用它们的组合响应来计算整体视网膜效率，期望值>0.50。最后，我们将测试对类似刺激的心理物理敏感性，这可能表明大脑和视网膜同样有效。 大约一半的神经节细胞使用不同的编码策略。它们“缓慢地”（更少的尖峰，更不精确的定时）发射信号来表示视觉场景的复杂特征，例如局部边缘。我们假设，一个“缓慢的代码”，虽然不如“轻快的代码”在传输高时间频率的有效性，是更有效的代谢能量和线量。为了评估这一假设，AIM 2将比较编码特性，能量预算，和线体积的轻快持续和“局部边缘”的细胞，具有相似的感受野大小，是最多的轻快和迟缓的类型。 最后，AIM 3将研究可能为两种编码提供效率的局部电路。某些双极末梢以“爆发”的方式释放谷氨酸量子，其时间类似于活跃的神经节细胞中的尖峰爆发。我们推测，爆发是由抑制性无长突反馈到终端的轻快，但不迟缓的双极细胞。我们将通过识别接触活跃和迟缓细胞的双极类型（形态和功能）并重建其局部无长突回路来测试这一点。拟议的研究解决了视网膜功能的基本机制，对人类视觉的早期阶段至关重要。

DESCRIPTION (provided by applicant): Neural signaling employs many stochastic processes (transmitter release, receptor binding, channel opening) that might add noise and thus lose information. Loss has been assessed by measuring neural efficiency, the ratio of sensitivities for a human and an "ideal observer", which is a computational model that includes the human preneural losses (blur, photon noise, etc.) but no neural loss. Neural efficiency has seemed to reach at most about 0.5, implying considerable loss (50%); however, neural efficiency might be much higher (about 1) and have been underestimated because test stimuli were ill-matched to the neural pathway that mediates the discrimination. To test these alternative hypotheses, AIM 1 will measure neural efficiency for "brisk" ganglion cells that encode by precise spike timing and supply the geniculo-striate system. We record a cell's response to an optimal stimulus (i.e., matched to the spatiotemporal impulse response); then we compare the cell's sensitivity to that of an ideal observer subject only to preneural loss. Ganglion cell efficiency (the ratio of these measures) may well be >0.5. We will also record from pairs of adjacent brisk ganglion cells and use their combined responses to compute overall retinal efficiency, expecting values >0.50. Finally, we will test psychophysical sensitivity to similar stimuli, which might show that the brain and retina are equally efficient. Roughly half of all ganglion cells use a different coding strategy. They fire "sluggishly" (fewer spikes, less precise timing) to signal complex features of the visual scene, such as local edges. We hypothesize that a "sluggish code", although less effective than the "brisk code" at transmitting high temporal frequencies, is more efficient with respect to metabolic energy and wire volume. To evaluate this hypothesis AIM 2 will compare the coding properties, energy budgets, and wire volumes for brisk-sustained and "local-edge" cells, which have similar receptive field size and are the most numerous of the brisk and sluggish types. Finally, AIM 3 will investigate local circuits that may lend efficiency to both types of coding. Certain bipolar terminals release glutamate quanta in "bursts" whose timing resembles the spike bursts in brisk ganglion cells. We hypothesize that the bursts are caused by inhibitory amacrine feedback onto the terminals of brisk but not sluggish bipolar cells. We will test this by identifying the bipolar types that contact brisk and sluggish cells (morphology and function) and reconstructing their local amacrine circuits. The proposed studies address fundamental mechanisms of retinal function critical to early stages of human vision.