Axonal spike-count regimes link spinal cord stimulation periodicity to artificial sensory detection and discrimination in rodents

Electrical stimulation is widely used to evoke artificial sensation, yet how temporal stimulation patterns are transformed into neural activity and perception remains poorly understood. Recent behavioral studies have shown that increasing the aperiodicity of spinal cord stimulation pulse trains alters both detection thresholds and discrimination performance despite constant pulse counts, suggesting a role for temporal structure beyond rate alone. However, the neural mechanisms underlying these effects remain unresolved. Approach: We coupled a finite-element spinal cord model to 420 conductance-based axons spanning 35 locations, six diameters, and two nodal phases. Nine index-axon spike-count regimes defined amplitudes applied to the entire population to produce heterogeneous recruitment and spike output. Population responses were used to investigate how stimulation periodicity shapes neural responses and resulting behavior. These regimes ranged from single spikes and small volleys to sustained spike trains, and they demonstrated stable operating ranges arising from interactions between stimulation timing and axonal membrane dynamics. Main results: No single regime sufficiently explained the experimentally observed detection threshold trends; instead, mixtures of regimes accurately reproduced frequency-dependent mean threshold behavior. Extending this framework to periodicity discrimination, rate and timing features derived from regime mixtures contained sufficient information to recover behavioral psychometric curves using constrained observer models. The per-frequency hybrid (rate + timing) model, with frequency-specific mixtures and block-specific decision parameters, had mean RMSE 0.0881, outperforming the individual rate (RMSE 0.1078) and timing (RMSE 0.0958) models in matching animal behavior. Model optimization (rate vs. timing) showed that discrimination from periodic stimuli incorporated greater rate information, while discrimination from aperiodic stimuli was more timing information-dependent. Significance: These results establish a mechanistic link between stimulation temporal structure, axonal spike generation, and perceptual behavior. This framework suggests that spinal cord stimulation does not operate within a single fixed neural regime but instead engages a spectrum of spike-count regimes whose mixtures can shape perception. These findings have important implications for the design of biomimetic stimulation strategies, highlighting temporal patterning as a key dimension for controlling sensory outcomes.