An Information Theoretic Approach to Stimulus Processing in the Olfactory System II

神经生物学中的一个基本问题是，外部刺激是如何由大脑中神经元的活动编码的。一个新出现的解决这个问题的原型模型系统是在小鼠OB的神经元网络中。一般来说，对于神经元来说，刺激的大小为每毫秒毫伏，动作电位(棘波)是编码和传播信息的唯一二进制手段。然而，刺激信息究竟是如何在输出神经元的尖峰序列中转换和包含的，仍然存在争议。OB的相对解剖学简单性，OB网络内感觉和状态依赖活动的组合计算，以及它绕过感觉丘脑与皮质的直接(和部分相互的)连接，使OB成为研究感觉信息处理基本原理的一个有吸引力的模型。作为我们先前提案中详细介绍的正在进行的研究的必然结果，我们现在的目标是扩展和完善信息论模型的范围，以更紧密地匹配/描述小鼠OB模型中的神经元通信和网络处理。基于扩展通道模型的特定预测/假设随后将在神经生理学实验中被系统地研究，反过来，模型参数将根据实验结果进行调整。虽然我们有意将自己局限于小鼠嗅觉系统，但我们预计所获得的结果可以在很大程度上推广到各种(神经)生物通讯系统。遵循当前资助期的研究逻辑，我们提出了一种多学科战略，将重点放在两个主要目标上：a.将为神经生物学家提供分析模型，以在纯数字的基础上模拟神经信息处理--这是一种创新的方法，我们预计将有助于对管理大脑感觉编码的基本规则获得新的见解。b.神经信息编码和传播是健壮的、能量高效的和高度容错的。尽管如此，神经网络是快速而高效的--这是技术通信系统中追求的功能。利用我们合作努力产生的模型和分析，我们将把生物启发的设计原则应用于通信领域，这在过去已被证明是非常富有成效的各种应用。我们坚信，理论和实验方法的协同将极大地促进我们对脑计算的理解，同时，为通信理论提供新的想法。

A fundamental question in neurobiology is how external stimuli are coded by the activity of neurons in the brain. An emerging prototypical model system to address this question is in the neuronal network of the mouse OB. For neurons in general, stimuli have dimensions of millivolts per millisecond and action potentials (spikes) are the sole binary means of both encoding and propagating information. How exactly stimulus information is transformed and contained in spike trains of output neurons, however, remains controversial. The relative anatomical simplicity of the OB, the combined computation of sensory and state-dependent activity within the OB network, and its direct (and partly reciprocal) connection to the cortex, bypassing the sensory thalamus, render the OB an attractive model to study the basic principles of sensory information processing.As a corollary to the on-going research detailed in our previous proposal, we now aim to extend and refine the scope of information theoretic models to more closely match / describe neuronal communication and network processing in the mouse OB model. Specific predictions / hypotheses based on extended channel models will then be systematically investigated in neurophysiological experiments and, in turn, model parameters will be adapted according to experimental findings. While we intentionally confine ourselves to the mouse olfactory system, we expect that the obtained results can be largely generalized to a variety of (neuro)biological communication systems. Following the research logic of the current funding period, we propose a multi-disciplinary strategy that will focus on two major goals:A. Neurobiologists will be provided with analytical models to simulate neural information processing on a purely numerical basis - an innovative approach that we expect to be instrumental in gaining novel insights into the principle rules that govern sensory coding in the brain.B. Neural information coding and propagation is robust, energy efficient and highly error tolerant. Nonetheless, neural networks are fast and efficient - features sought after in technical communication systems. Using models and analyses that emerge from our collaborative efforts, we will adapt bio-inspired design principles to the field of communications, which in the past has turned out very fruitful for a variety of applications.We firmly believe that synergy from theoretical and experimental approaches will substantially advance our understanding of brain computations and, in parallel, fuel novel ideas in communication theory.