Core–periphery organization reveals functional heterogeneity in spatial coding ensembles

Head-direction cells (HDCs) and grid cells (GCs) are fundamental to spatial cognition, and their identification has traditionally relied on spatial tuning profiles measured under controlled conditions. Although widely used, this approach may not fully capture neuronal function and can be difficult to apply in complex or naturalistic settings. Moving beyond single-neuron tuning, we show that ensembles containing spatially tuned neurons exhibit a continuous core–periphery (CP) organization in their activity-correlation patterns. By computing pairwise correlations from neuronal activity sequences, we assigned each neuron a core score and used these scores to define core and peripheral groups for subsequent analyses. Core neurons exhibited stronger population coupling, higher spatial information measures, and greater tuning stability than peripheral neurons. Across multiple open-access datasets, this CP-based partition showed substantial but incomplete concordance with conventional labels, exhibiting higher alignment with HDC/non-HDC than with GC/non-GC classifications; notably, it distinguished differences in spatial coding that were not fully resolved by conventional classification. Moreover, pairwise tuning offsets could be inferred from activity-correlation patterns and integrated to reconstruct the population-wide configuration of preferred head directions and grid phases, without relying on individual tuning profiles. Together, this work establishes the CP framework as a simple, data-driven population-level approach to characterizing spatial coding beyond what is captured by conventional tuning-based classification.