Noise components in Limulus vision.
Noise components in Limulus vision.
复制标题
鲎视觉中的噪声成分。
DOI：
10.1086/bblv187n2p261
复制
发表时间：
1994
期刊：
影响因子：
--
通讯作者：
BarlowJr,RB
中科院分区：
文献类型：
--
作者：
Dodge,FA;Porcello,DM;Dodge,SA;Kaplan,E;BarlowJr,RB
当鲎沿着浅湾底部移动时，它的侧眼不断地将周围环境的图像传输回大脑。这个“视频图片”仅包含大约 40 X 25 像素，由于小眼的宽接收角而变得模糊，并且由于眼睛不均匀地采样视觉空间的方式而扭曲。在眼睛内，两个神经抑制过程增强了对比度，但神经编码中所有步骤的最终结果通常是一个非常嘈杂的图像。尽管图像质量相当差，但鲎大脑中的神经回路可以提取视觉信息来指导动物的运动。在交配季节，单身雄性提供了视觉引导行为的一个特别直接的例子。它们通常会转身并接近任何与伴侣大小相当的物体。正如去年首次报道的（l），我们可以记录当眼睛扫描行为实验中使用的相同目标时大脑接收到的神经信号。对几个此类实验的分析表明，距离超出行为阈值的目标会引起可察觉的尖峰速率调节 (2)。该结果与预期响应的理论计算完全一致 (3)。如果行为阈值不能用光感受器的敏感性来解释，也许我们的实验范式一直低估了动物在自然条件下必须应对的噪音。因此，我们开始研究尖峰率随机波动的统计特性如何取决于不同的视觉环境。眼睛固有的两种噪声源已得到充分表征；即，由光子吸附触发的散粒噪声，它是尖峰发生器的兴奋输入，以及由许多邻居中的尖峰触发的总抑制电位。由于单一 IPSP 的长度超过 0.5 秒，因此它可以平滑其输入中的几乎所有波动，并且实际上对噪声的贡献很小。在昏暗的环境照明下，光转导噪声可能相当大，但随着眼睛适应更高的照明度，光转导噪声会大大降低。该噪声的功率谱显示出相对较宽的通带，与正弦调制光的动态传递函数相匹配 (4)。相当于日光场实验的光照水平下的光转导噪声（加上抑制性突触噪声）会导致尖峰率波动，变异系数 (CV) 为 7% 或更小。
As a horseshoe crab moves along the bottom of a shallow bay, its lateral eye continuously transmits a picture of its environment back to his brain. This “video picture” contains only about 40 X 25 pixels, is blurred by the wide acceptance angle of the ommatidium, and is distorted by the way the eye samples visual space nonuniformly. Within the eye, two neural inhibitory processes enhance contrast, but the net result of all the steps in neural encoding is often a very noisy image. In spite ofthe rather poor image quality, neural circuits in the Limulus brain can extract visual information to guide the animal’s movements. Unattached males during the mating season provide a particularly straightforward example of visually guided behavior; they generally turn and approach any object that is about the size of a mate. As first reported here last year (l), we can record the neural signals that the brain receives as the eye scans the same targets used in the behavioral experiments. Analysis of several such experiments has revealed that targets at a distance beyond that of the behavioral threshold elicit a perceptible modulation of spike rate (2). This result is fully consistent with theoretical computations of the expected responses (3). If the behavioral threshold cannot be explained by the sensitivity of the photoreceptors, perhaps our experimental paradigms have consistently underestimated the noise that the animal must contend with under natural conditions. We have therefore started to study how the statistical properties of the random fluctuations in spike rate depend on differing visual environments. Two sources of noise that are intrinsic to the eye have been fully characterized; namely, the shot noise triggered by photon adsorption that is the excitatory input to the spike generator, and the summed inhibitory potential triggered by spikes in many neighbors. Because the unitary IPSP is longer than 0.5 s, it smooths nearly all fluctuations in its input and, in fact, makes only a minor contribution to the noise. The phototransduction noise can be quite large under dim ambient illumination, but decreases greatly as the eye adapts to higher illuminations. The power spectrum of this noise shows a relatively broad pass-band matching the dynamic transfer-function for sinusoidally modulated light (4). Phototransduction noise at illumination levels equivalent to the daylight field experiments (plus inhibitory synaptic noise) causes spike-rate fluctuations with a coefficient of variation (CV) of 7% or less.