Auditory processing: The cellular and synaptic mechanisms of a delay-line and coincidence-detector circuit

与青蛙和许多昆虫一样，不同物种的动物通过由简单重复的声音脉冲模式组成的声音信号进行交流。这种简单的模式识别过程是更复杂的听觉交流的基本要素，包括人类语音和音乐。然而，声音具有非常短暂的性质。神经系统必须能够处理随时间传播的信号序列，以检测它们的时间模式，以便识别含义。在莫尔斯电码中形成字母“H”的四个脉冲，与雌性蟋蟀被雄性叫声吸引时所定向的四个脉冲非常相似。截然不同的神经系统是如何处理和检测这种模式的？不同物种之间的计算挑战非常相似。当不同动物的神经系统执行相似的任务时，神经细胞和神经网络表现出非常相似的适应能力。对于时间声音序列的处理，研究人员分析了不同的动物，得出了非常相似的概念。他们的想法是，延迟线和符合探测器将是解决时间处理问题的理想神经元电路。这是什么意思？让我们来看看莫尔斯电码中字母“H”的4个脉冲，它们之间可能有50毫秒的时间间隔。在神经系统中，听觉信号被分成两条通路。一条路径将信息直接转发到符合探测器，另一条路径使用神经元，对信号施加50毫秒的延迟。结果很简单：当第一个脉冲被处理时，它将到达符合探测器，没有其他匹配的信号；探测器将不被激活。当莫尔斯信号的第二脉冲被处理时，从第一脉冲到第二脉冲的延迟信号和直接信号将在探测器处重合并引起强烈的响应。随着时间的推移，对符合探测器的响应求和，可以驱动具有代表非常特定模式的信号的活动的特征探测器。在像我们这样的复杂大脑中，数千个这样的延迟线和重合探测器很可能被布置来检测不同间隔和模式的声音信号，并形成感知语言和音乐节奏的基础。众所周知，这些神经回路很难研究，而且从一个动物到另一个动物，不可能识别出一个神经元。一些无脊椎动物具有更简单的神经系统，如蟋蟀，它们提供了更深入了解神经处理的机会，因为延迟线和符合探测器电路可以在识别出的神经元的水平上进行分析。因此，我们可以彻底分析棘波和突触的活动及其在许多个体中的网络特性。我们已经描述了板球大脑中电路的关键部件，如延迟线神经元、符合检测器神经元和特征检测器。该电路针对板球歌曲的脉冲间隔进行调谐，但是如何检测脉冲的持续时间以及如何处理歌曲的整体啁啾模式？我们将通过记录回路神经元的活动来解决这些问题，并用我们已经用来描述雌性听觉偏好的特定声音模式来测试它们。如果我们发现与行为偏好相匹配的活动模式，那么这些将使我们能够解释电路对脉冲持续时间和啁啾模式的调节是如何建立的。这将完成我们对在两个非常不同的时间尺度上处理时间模式的理解。然后，我们将探索模式识别电路如何通过识别和分析启动和维持听觉定向行为的下行神经元来控制女性的听觉行为。

Animals across a wide range of species communicate with acoustic signals composed of simple repetitive patterns of sound pulses like frogs and many insects do. Such simple processes of pattern recognition are essential elements to more complex auditory communication including human speech and also music.Sound however, has a very transient nature. The nervous system must be able to process sequences of signals spread across time to detect their temporal pattern in order to identify a meaning. The four pulses, that form the letter "H" in Morse code, are very similar to the 4 pulses that female crickets orient to, when they are attracted by the calling song of a male. How can the very different nervous systems process and detect such patterns? The computational challenge is very similar across species. When nervous systems of different animals perform a similar task, the nerve cells and neural networks show very similar adaptations. For the processing of temporal sound sequences, researchers analysing different animals came up with very similar concepts. Their idea is that delay-lines and coincidence-detectors would be ideal neuronal circuits to address the question of temporal processing. What does this mean? Let's look at the 4 pulses of the letter "H" in Morse code, which may be separated by a time interval of 50 ms. In the nervous system the auditory signal is split into two pathways. One pathway forwards the information directly to a coincidence detector, the other pathway uses a neuron that imposes a delay of 50 ms on the signal. The consequence is very simple: When the first pulse is processed, it will arrive at the coincidence-detector with no other matching signal; the detector will not be activated. When the second pulse of the Morse signal is processed, the delayed signal from the first pulse and the direct signal to the second pulse will coincide at the detector and elicit a strong response. Summing the coincidence-detector responses over time, can drive feature detectors with activity representing signals of a very specific pattern. In complex brains like ours, thousands of such delay-lines and coincidence-detectors are likely arranged to detect sound signals of different intervals and patterns and form the basis for the perception of rhythms in language and music. These neural circuits are notoriously difficult to study, and no one neuron can be identified in one animal to the next. Some invertebrates, with simpler nervous systems such as the cricket, offer the chance to deeper understand neural processing because delay-line and coincidence-detector circuits can be analysed at the level of identified neurons. We can therefore thoroughly analyse spike and synaptic activity and their network properties across many individuals. We have already characterised the key components of the circuit in the cricket brain, like the delay-line neuron, the coincidence-detector neuron and the feature detector. This circuit is tuned towards the pulse intervals of the cricket song, but how is the duration of pulses detected and how is the overall chirp pattern of the song processed? These are questions that we will address by recording the activity of the circuit neurons and testing them with specific sound patterns that we have already used to characterise the auditory preferences of the females. If we find activity patterns that match the behavioural preferences, then these will allow us to explain how the tuning of the circuit to pulse durations and chirp patterns is established. This will complete our understanding of the processing of temporal patterns at two very different time scales. We will then explore how the pattern recognition circuit leads to the control of female auditory behaviour by identifying and analysing descending neurons that initiate and maintain auditory orientation behaviour.