Spike-to-Field Mechanisms of Turbulence-Like Dynamics in Spatial Spiking Neural Networks
Organizations: ZeQome, Zurich, Switzerland
Abstract
Neural avalanches and turbulence-like brain dynamics are usually measured after neural activity has been coarse-grained into continuous fields, leaving open which spike-level mechanisms can generate these macroscopic observables. We build a spatial leaky integrate-and-fire spiking neural network and explicitly transform spikes into activity-density and Hilbert phase fields. A delayed, distance-dependent, passive-dendrite regime reproducibly generates turbulence-like field structure, including high-amplitude phase-defect tracks, structure-function scaling, and power-law-favored continuous-field avalanche tails. Controls show that firing rate, vortex count, and avalanche tail fits are individually insufficient: destroying delayed propagation, dendritic filtering, or spike timing suppresses key field signatures, while random-space nulls can retain many mathematical phase defects but show weaker spatial structure and synaptic-delay transfer. Biologically motivated gates further shape the regime: PV-like perisomatic inhibition clamps phase dynamics, SST-like dendritic inhibition and projection-specific inhibitory facilitation preserve long-lived tracks, and a rate-constrained projection-STP regime expresses a roughly 200-ms episode at 12.25 Hz (full-record mean) across 12 seeds. These results use SNNs as a mechanistic testbed for spike-derived brain turbulence observables and show why phase defects and avalanche-like statistics must be interpreted jointly with spatial scaling and spike-level propagation.
Figures & tables
| Model | Rate | Vortices | Tracks | slope | |
|---|---|---|---|---|---|
| Passive + distance delay | 115.43 | 6.05 | 82.50 | 0.94 | 2.05 |
| No distance delay | 128.67 | 3.70 | 58.33 | 0.96 | 2.01 |
| Short range + delay | 113.65 | 6.10 | 92.00 | 1.15 | 1.77 |
| No passive dendrite | 164.25 | 2.29 | 37.25 | 0.98 | 2.06 |
| Uniform fast control | 129.23 | 4.28 | 64.83 | 0.84 | 1.92 |
| State | Global-rate shuffle | Spatial permutation | Time shift |
|---|---|---|---|
| High-rate baseline | |||
| Low-rate episode |
| Topology | Bits/opportunity | Cycle edges | re-entry | Raw spike excess |
|---|---|---|---|---|
| Recurrent EDR | 1.00 | 0.736 | 0.138 | |
| Feedforward acyclic | 0.00 | 0.588 | 0.107 | |
| Forward-biased same-degree | 1.00 | 0.652 | 0.118 | |
| Loop-enhanced same-degree | 1.00 | 0.757 | 0.134 |
Appendix figures & tables15 assets
Supplementary material from the paper’s appendix.
Appendix
| Representation / decoder | Features | CV accuracy | Test accuracy |
|---|---|---|---|
| Raw / linear ridge | Event temporal bins | 0.625 | 0.95 |
| Projection STP / linear ridge | State temporal bins | 0.450 | 0.45 |
| Projection STP / RBF ridge | State temporal bins | 0.425 | 0.50 |
| Balanced E/I / linear ridge | State temporal bins | 0.325 | 0.45 |
Spiking Neural Networks
Spiking neural networks process information through discrete, time-dependent neural events, aiming for efficient computation and biologically inspired learning.
See all 383 papers