Tolossa Gemechu Bekele, Schneider Aidan M, Dyer Eva, Hengen Keith B
Department of Biology, Washington University in St Louis, St Louis, United States.
Department of Biomedical Engineering, Georgia Institute of Technology, Atlanta, United States.
Elife. 2025 Jun 27;13:RP101506. doi: 10.7554/eLife.101506.
Neurons in the brain are known to encode diverse information through their spiking activity, primarily reflecting external stimuli and internal states. However, whether individual neurons also embed information about their own anatomical location within their spike patterns remains largely unexplored. Here, we show that machine learning models can predict a neuron's anatomical location across multiple brain regions and structures based solely on its spiking activity. Analyzing high-density recordings from thousands of neurons in awake, behaving mice, we demonstrate that anatomical location can be reliably decoded from neuronal activity across various stimulus conditions, including drifting gratings, naturalistic movies, and spontaneous activity. Crucially, anatomical signatures generalize across animals and even across different research laboratories, suggesting a fundamental principle of neural organization. Examination of trained classifiers reveals that anatomical information is enriched in specific interspike intervals as well as responses to stimuli. Within the visual isocortex, anatomical embedding is robust at the level of layers and primary versus secondary but does not robustly separate individual secondary structures. In contrast, structures within the hippocampus and thalamus are robustly separable based on their spike patterns. Our findings reveal a generalizable dimension of the neural code, where anatomical information is multiplexed with the encoding of external stimuli and internal states. This discovery provides new insights into the relationship between brain structure and function, with broad implications for neurodevelopment, multimodal integration, and the interpretation of large-scale neuronal recordings. Computational approximations of anatomy have the potential to support in vivo electrode localization.
已知大脑中的神经元通过其放电活动编码各种信息,主要反映外部刺激和内部状态。然而,单个神经元是否也在其放电模式中嵌入有关自身解剖位置的信息,在很大程度上仍未得到探索。在这里,我们表明机器学习模型仅根据神经元的放电活动就能预测其在多个脑区和结构中的解剖位置。通过分析清醒、行为活跃的小鼠中数千个神经元的高密度记录,我们证明在各种刺激条件下,包括漂移光栅、自然主义电影和自发活动,都可以从神经元活动中可靠地解码解剖位置。至关重要的是,解剖特征在不同动物甚至不同研究实验室之间具有普遍性,这表明了神经组织的一个基本原理。对训练好的分类器的检查表明,解剖信息在特定的峰间期以及对刺激的反应中更为丰富。在视觉等皮质内,解剖嵌入在层水平以及初级与次级水平上是稳健的,但不能稳健地分离各个次级结构。相比之下,海马体和丘脑内的结构基于其放电模式可以稳健地分离。我们的发现揭示了神经编码的一个可推广维度,其中解剖信息与外部刺激和内部状态的编码相互交织。这一发现为脑结构与功能之间的关系提供了新的见解,对神经发育、多模态整合以及大规模神经元记录的解释具有广泛的意义。解剖结构的计算近似有可能支持体内电极定位。