Investigating neuronal network dynamics : scale-invariance, preferred firing rates, and plasticity via phase-shift encoding

Understanding how neuronal networks process information and adapt to external stimuli is critical for advancing our knowledge of neural plasticity and computation. Scale-invariant properties, which suggest self-organised criticality (SOC), have been observed in these networks, but the mechanisms underlying these dynamics have remained unclear. Despite theoretical models, there is still limited experimental evidence on how neuronal networks exhibit scale-invariant dynamics when exposed to varying stimulation conditions. In particular, the role of phase-shift encoding in driving network adaptability and plasticity has not been thoroughly explored. Here, we show that neuronal networks can dynamically integrate and process complex stimuli through phase-space modulation using a novel phase-shift encoding protocol. Using Finalspark’s cutting edge wetware computing platform, we conducted experiments involving baseline recordings, inhomogeneous Poisson point process (IPP) stimulations, spike-timing-dependent plasticity (STDP) protocols, and phase-shift encoding experiments, we investigated neuronal network dynamics. Preferred firing rates were identified across multiple neural circuits in an iterative design process, facilitated by Finalspark’s high-resolution, real-time stimulation and data collection capabilities. Targeted stimulation based on Hebbian learning rules with Granger causality analysis was used to evaluate pre-synaptic and post-synaptic relationships. Distinct firing patterns emerged from the baseline and Poisson stimulations via Fourier transform analysis, suggesting stable, scale-invariant temporal structures. K-S test, Shapiro-Wilk test, and Whitney-U test all showed P <0.005 under phase-shift encoding, comparing pre-stimulation behaviour to post-stimulation, indicating significant changes in network dynamics. Moreover, dimensionality reduction techniques such as UMAPs, PCAs, and PRCs revealed shifts in neural manifolds before and after training. Our findings, enabled by Finalspark’s platform, provide experimental evidence of scale-invariant plasticity in neuronal networks and suggest that networks can adapt to and retain externally introduced data. We anticipate that this research offers new insights into neural encoding mechanisms and could inform future developments in wetware computational systems.

Grantor dc:publisher

Oxford Brookes University

Author dc:creator

• Bayle, Fraser William

Contributors dc:contributor

• olde Scheper, Tjeerd
• Rast, Alexander

Statement dc:rights

• All rights reserved

Language dc:language

en

DOI dc:identifier

https://doi.org/10.24384/ersy-4c88

OAI identifier oai:identifier

tle:ab76a193-afe5-4961-b2ba-acbe2a6717dd:d6bd9758-527a-46cd-bfe2-c433766e8fca:1