How effective delays shape oscillatory dynamics in neuronal networks

Synaptic, dendritic and single-cell kinetics generate significant time delays that shape the dynamics of large networks of spiking neurons. Previous work has shown that such effective delays can be taken into account with a rate model through the addition of an explicit, fixed delay [Roxin et al. PRL 238103 (2005)]. Here we extend this work to account for arbitrary symmetric patterns of synaptic connectivity and generic nonlinear transfer functions. Specifically, we conduct a weakly nonlinear analysis of the dynamical states arising via primary instabilities of the asynchronous state. In this way we determine analytically how the nature and stability of these states depend on the choice of transfer function and connectivity. We arrive at two general observations of physiological relevance that could not be explained in previous works. These are: 1 - Fast oscillations are always supercritical for realistic transfer functions. 2 - Traveling waves are preferred over standing waves given plausible patterns of local connectivity. We finally demonstrate that these results show a good agreement with those obtained performing numerical simulations of a network of Hodgkin-Huxley neurons.

💡 Research Summary

The paper investigates how effective time delays—arising from synaptic transmission, dendritic propagation, and intrinsic cellular kinetics—shape the collective dynamics of large spiking neuronal networks. Building on the rate‑model framework introduced by Roxin et al. (2005), the authors generalize the approach to incorporate arbitrary symmetric connectivity matrices and generic nonlinear transfer functions that more faithfully represent the input–output relationship of real neurons.

First, a linear stability analysis of the asynchronous (homogeneous) state is performed. The eigenvalue spectrum of the linearized system reveals two distinct primary instabilities: a Hopf bifurcation, which generates oscillatory activity, and a real‑eigenvalue bifurcation, which gives rise to spatial patterns (standing or traveling waves). The critical parameters governing these bifurcations are the effective delay τ_eff, the shape of the transfer function F(V), and the Fourier transform of the connectivity kernel.

Next, the authors conduct a weakly nonlinear analysis near the bifurcation points. By expanding the transfer function around the fixed point and applying multi‑scale techniques, they derive amplitude equations of the form (\dot{A}= \mu A - \beta A^{3} + \dots). The sign of the cubic coefficient β is shown to be dictated by the third derivative of F at the operating point. For realistic, saturating transfer functions (as measured in cortical pyramidal cells), β is positive, guaranteeing that the Hopf bifurcation is supercritical. Consequently, fast oscillations (in the gamma range) emerge gradually and settle onto a stable limit cycle, a result that contrasts with earlier studies that reported both super‑ and subcritical scenarios.

The spatial analysis focuses on distance‑dependent, symmetric connectivity kernels (e.g., Gaussian or Mexican‑hat profiles). By evaluating the dispersion relation λ(k) for wavevectors k, the authors demonstrate that the growth rate of traveling‑wave modes exceeds that of standing‑wave modes when the effective delay aligns with the propagation time of the wave. This asymmetry arises because the delay introduces a phase shift that preferentially stabilizes modes with a non‑zero group velocity. The theoretical prediction is that, under biologically plausible local connectivity, traveling waves are the dominant pattern, while standing waves are generally unstable.

To validate the analytical predictions, large‑scale simulations of Hodgkin‑Huxley neurons (≈10,000 cells) are carried out. Synaptic delays are varied between 1 and 5 ms, and the connectivity is tuned to match the assumed kernels. The simulations reproduce the supercritical emergence of gamma‑band oscillations (≈80–120 Hz) and exhibit traveling waves with speeds of 0.2–0.5 m/s, matching the analytical growth rates and wave numbers derived from the dispersion relation. Standing‑wave configurations decay, confirming the theoretical instability.

In summary, the study provides two physiologically relevant insights that were not captured by earlier models: (1) realistic neuronal transfer functions ensure that fast oscillations are always supercritical, leading to smooth onset and stable rhythmic activity; (2) plausible local connectivity patterns favor traveling waves over standing waves, offering a mechanistic explanation for the prevalence of propagating activity observed in cortical recordings. These findings deepen our understanding of how delays and network architecture jointly sculpt neuronal dynamics and suggest design principles for neuromorphic systems that aim to emulate brain‑like rhythmic behavior.