Beyond the single neuron: computational principles of distributed neural coding in the age of large-scale recordings

Neural coding is the fundamental language of the brain, the set of rules by which stimulus features and internal states are transformed into electrical signals. Historically, research focused on the properties of single neurons, emphasizing firing rates and receptive field tuning. However, a major paradigm shift is underway, moving toward the study of distributed representations where information is encoded in the collective activity of large-scale neural ensembles. From the precision of spike-timing, to the robustness of population vectors and high-dimensional manifolds, the principles governing neural representation are remarkably diverse.The central problem this Research Topic addresses is the absence of a unified theoretical framework bridging microscopic neural activity and macroscopic functional representations. While detailed models of single-cell biophysics and high-level descriptions of population dynamics exist, the "translation rules" between these scales remain elusive. Specifically, it is unclear how local computations at the synapse and individual neuron level aggregate into distributed codes that are both flexible enough for learning and stable enough for long-term memory. This gap has become increasingly pressing as modern recording technologies, Neuropixels probes, widefield calcium imaging, and large-scale multi-region arrays, now allow us to monitor thousands of neurons simultaneously across multiple brain areas, generating datasets that far outpace our current theoretical frameworks.To address this, the Research Topic brings together interdisciplinary work utilizing tools from information theory, Bayesian inference, manifold learning, and network modeling. We seek to identify the core computational principles that allow neural codes to optimize for energy efficiency, speed, and accuracy. By synthesizing these multi-scale perspectives, we aim to uncover how the brain maintains robust representations across different modalities and how these codes are reconfigured during various cognitive and motor tasks.We are particularly interested in original research articles, theoretical and computational studies, and focused review articles, however, invite authors to submit all articles that investigate the mechanisms of neural coding across different levels of organization. We particularly welcome work that proposes novel mathematical frameworks or uses computational models to provide mechanistic explanations for experimental data.Specific themes of interest are organized around three areas:Coding Mechanisms- The balance between temporal coding and rate coding in different physiological contexts- Sparse coding principles and the impact of energetic constraints on information processing- The role of neural synchrony and oscillations in coordinating distributed codesGeometric and Theoretical Frameworks- Geometry and topology of neural manifolds in population-level representations- Information-theoretic measures for quantifying neural representation and communication- Bayesian and normative approaches to understanding population codesDynamics Across Learning and the Hierarchy- Stability and plasticity of neural codes during lifelong learning and environmental changes- Hierarchical coding: transformations from primary sensory inputs to high-level executive representations- Reconfiguration of neural codes during cognitive and motor task demands Topic editors: Daya Shankar Gupta, Yunpeng Cai, Fernando Montani Source: https://www.frontiersin.org/research-topics/81171/beyond-the-single-neuron-computational-principles-of-distributed-neural-coding-in-the-age-of-large-scale-recordings