Scholes Chris, Coombes Stephen, Palmer Alan R, Rhode William S, Mill Rob, Sumner Christian J
School of Psychology, University of Nottingham, Nottingham, United Kingdom.
School of Mathematical Sciences, University of Nottingham, Nottingham, United Kingdom.
PLoS Biol. 2025 Jun 16;23(6):e3003213. doi: 10.1371/journal.pbio.3003213. eCollection 2025 Jun.
The dynamic fluctuations in the amplitude of sound, known as sound envelopes, are ubiquitous in natural sounds and convey information critical for the recognition of speech, and of sounds generally. We are perceptually most sensitive to slow modulations which are most common. However, previous studies of envelope coding in the brainstem found an under-representation of these slow, low-frequency, modulations. Specifically, the synchronization of spike times to the envelope was enhanced in some neuron types, forming channels specialized for envelope processing but tuned to a restricted range of fast, high-frequency, envelopes (200-500 Hz). Here, we show using a historical dataset from cats that previous analyses, which made strong assumptions about the neural code, underestimated the encoding of low-frequency envelopes. While some neurons encode envelope better than others, most encode a wide range of envelope frequencies, and represent slower envelope fluctuations most accurately in their precise patterns of spike times. Identification of envelope frequency from spike-timing was linked to reliability, and to the way that dynamics of spiking interacted with the time-varying envelope. In some of the best-performing neurons, temporally complex "mode-locked" spike patterns served to enhance envelope coding. A second long-standing contradiction was that neural envelope coding is degraded at high sound levels, whilst the perception of envelope is robust at a wide range of sound levels. We find that spike-time encoding of envelope shape becomes level-robust for small populations of neurons. These findings argue against feature-specific coding of envelopes in the brainstem, and for a distributed population spike-time code for which synchrony to the envelope is an incomplete description. This code is accurate for slow fluctuations and robust across sound level. Thus, precise spike-timing information in the brainstem is after-all aligned with the needs of communication and the perception of environmental sounds.
声音幅度的动态波动，即所谓的声包络，在自然声音中无处不在，并传递着对于语音识别以及一般声音识别至关重要的信息。我们在感知上对最常见的缓慢调制最为敏感。然而，先前对脑干中包络编码的研究发现，这些缓慢的低频调制表现不足。具体而言，在某些神经元类型中，动作电位时间与包络的同步性增强，形成了专门用于包络处理的通道，但这些通道被调谐到一个有限的快速高频包络范围（200 - 500赫兹）。在这里，我们使用来自猫的历史数据集表明，先前那些对神经编码做出强烈假设的分析低估了低频包络的编码。虽然一些神经元比其他神经元能更好地编码包络，但大多数神经元能编码广泛的包络频率范围，并在其精确的动作电位时间模式中最准确地表示较慢的包络波动。从动作电位时间识别包络频率与可靠性以及动作电位发放动态与随时间变化的包络相互作用的方式有关。在一些表现最佳的神经元中，时间上复杂的“锁相”动作电位模式有助于增强包络编码。另一个长期存在的矛盾是，神经包络编码在高声级下会退化，而包络感知在广泛的声级范围内却很稳健。我们发现，对于少量神经元群体，包络形状的动作电位时间编码变得对声级具有稳健性。这些发现反对脑干中包络的特征特异性编码，并支持一种分布式群体动作电位时间编码，对于这种编码，与包络的同步只是一种不完整的描述。这种编码对于缓慢波动是准确的，并且在声级范围内具有稳健性。因此，脑干中精确的动作电位时间信息终究与通信需求和环境声音感知相契合。
