Fast rate coding in heterogeneous spiking networks

Heterogeneous networks of non-Poisson integrate-and-fire-type neurons can produce fast rate-based dynamics, with feedforward signal transmission described by a history-dependent rate network that becomes asymptotically exact as network size grows.

• Why it matters: Rate coding is defined by continuous rate-network dynamics, but prior computational work relied on Poisson-like spikes or slow dynamics, leaving uncertainty about fast, spike-driven emergence under persistent heterogeneity.
• What they did: The authors analyze feedforward spiking networks and derive a history-dependent rate-network description that is asymptotically exact as network size increases, contrasting it with linear-nonlinear rate networks.
• The result: The constructive fast rate-coding framework shows that extending rate-based analyses of neural recordings is feasible without assuming slow dynamics or Poisson rate codes.

AI summary of the abstract · may contain errors · not medical advice · read the paper.