Neuronal population inference reveals how neuromodulation reshapes conductance space

Neuromodulators such as dopamine, serotonin, and histamine continuously reshape neuronal excitability by acting on ion channel densities and kinetics. While this process is central to brain function, it remains difficult to characterize experimentally: metabotropic signaling cascades make the mapping from neuromodulator to ion channel nontrivial, and direct measurements of all affected conductances are rarely feasible. Moreover, ion channel densities vary considerably across neurons of the same type, so the effect of a given neuromodulator cannot be understood from a single “average” model. A key obstacle is neuronal degeneracy: distinct combinations of ion channels can generate similar spiking patterns. Standard approaches that attempt to fit a unique model from data, therefore, miss the richness of possible solutions. To address this challenge, we introduce a computational tool that reconstructs rich populations of conductance-based models (CBMs) from spike times, the most widely available form of experimental data. The method combines deep learning with Dynamic Input Conductances (DICs), a compact representation that aggregates the dynamical influence of ion channels on excitability, enabling fast and interpretable inference of diverse conductance sets from activity. By design, the reconstruction preserves degeneracy, producing heterogeneous ion channel configurations that all reproduce the observed dynamics. With this capability, neuromodulation can be investigated at the population level. Instead of treating modulation as isolated parameter changes, the tool reveals how entire distributions of conductances move through parameter space under modulatory action. This makes it possible to identify which channels are likely targeted and how robustness emerges from shifting populations rather than fixed parameters. Applied to distinct CBMs, the method shows that different modulators reorganize excitability regimes by moving whole populations in conductance space, providing a new perspective on the interplay between variability, robustness, and modulatory control.