Neural oligarchy: how synaptic plasticity breeds neurons with extreme influence

Florence I. Kleberg, Jochen TrieschbioRxiv 2018 · DOI

TL;DR

This paper proposes a computational model that explains how skewed distributions of synaptic strengths and neuronal firing rates arise in the brain. The model demonstrates that a combination of spike-timing dependent plasticity (STDP) and synaptic normalization naturally leads to a "neural oligarchy," where a small subset of neurons gains disproportionately large influence.

What Problem Does This Paper Address?

Cortical neural networks exhibit a peculiar characteristic: the distribution of synaptic strengths and neuronal firing rates are not uniform. Instead, they follow "right-skewed, long-tailed" distributions, meaning most neurons and synapses have moderate influence, but a few have exceptionally high influence. This phenomenon, observed in experimental recordings, raises fundamental questions about how such distributions emerge and are maintained in the brain. Understanding this is crucial for comprehending neural computation, learning, and even the origins of neurological disorders like epilepsy.

Methods

The authors developed a spiking neural network model called the "LIF-SORN" (Leaky Integrate-and-Fire Self-Organizing Recurrent Network). This model incorporates two key biological mechanisms:

  • Spike-Timing Dependent Plasticity (STDP): This learning rule modifies synaptic strength based on the precise timing of pre- and post-synaptic spikes. The model uses both excitatory STDP (eSTDP) and inhibitory STDP (iSTDP).
  • Synaptic Normalization (SN): This mechanism ensures that the overall synaptic input to a neuron remains within a certain range, preventing runaway excitation or inhibition. The model implements a multiplicative form of SN.

Additionally, the study explores the impact of small variations in neuronal spiking thresholds and spontaneous fluctuations in synaptic weights.

Datasets

Not discussed in the paper. The study uses a simulated neural network model.

Key Insights and Ideas

  • STDP and SN are sufficient: The paper demonstrates that the combination of STDP and synaptic normalization, without requiring more complex mechanisms like structural plasticity, can naturally generate the observed skewed distributions of synaptic weights and firing rates.
  • "Neural Oligarchy" emerges: The model predicts the existence of a small group of "hub" neurons that exert a disproportionately large influence on the network. This arises from the interplay of high firing rates and strong outgoing synaptic connections, which are themselves shaped by plasticity.
  • Firing rates and synaptic strength are linked: Neurons that fire more frequently tend to develop stronger connections, creating a positive feedback loop that amplifies their influence.
  • Robustness to noise: The lognormal-like distributions are surprisingly stable, even when spontaneous, random fluctuations in synaptic efficacies are introduced.

Key Results

  • The model successfully reproduces lognormal-like distributions for both excitatory and inhibitory synaptic efficacies, consistent with experimental findings.
  • Similarly, the model generates lognormal-like distributions for neuronal firing rates, driven by small variations in spiking thresholds.
  • A measure of "impact" reveals a small population of neurons with exceptionally high influence on the network dynamics.
  • The lognormal-like distributions persist even in the presence of spontaneous weight fluctuations and when neuronal activity is suppressed (TTX condition).

Limitations and Future Work

  • Simplified model: The authors acknowledge that their model is a simplification. For instance, they did not include plasticity for synapses onto inhibitory neurons, which has been observed experimentally. Future work could explore fully plastic excitatory-inhibitory networks.
  • Homeostatic mechanisms: While SN is included, the model doesn't account for homeostatic mechanisms that operate over much longer timescales.
  • Synaptic connections: The model does not address the pruning and creation of synaptic connections (multi-synapse connections). Investigating how this affects inhibitory weight distributions is suggested.
  • Inhibitory plasticity: The precise role and mechanisms of inhibitory plasticity in shaping network dynamics remain an active area of research, and the authors note that a comprehensive theoretical foundation is still lacking.
  • Experimental validation: The authors suggest that their prediction of "hub" neurons with high firing rates and strong efferent connections could be experimentally tested by combining electrophysiological recordings with detailed anatomical reconstruction of neuronal connections.

Summary updated

Abstract

Abstract Synapses between cortical neurons are subject to constant modifications through synaptic plasticity mechanisms, which are believed to underlie learning and memory formation. The strengths of excitatory and inhibitory synapses in the cortex follow a right-skewed long-tailed distribution. Similarly, the firing rates of excitatory and inhibitory neurons also follow a right-skewed long-tailed distribution. How these distributions come about and how they maintain their shape over time is currently not well understood. Here we propose a spiking neural network model that explains the origin of these distributions as a consequence of the interaction of spike-timing dependent plasticity (STDP) of excitatory and inhibitory synapses and a multiplicative form of synaptic normalisation. Specifically, we show that the combination of additive STDP and multiplicative normalisation leads to lognormal-like distributions of excitatory and inhibitory synaptic efficacies as observed experimentally. The shape of these distributions remains stable even if spontaneous fluctuations of synaptic efficacies are added. In the same network, lognormal-like distributions of the firing rates of excitatory and inhibitory neurons result from small variability in the spiking thresholds of individual neurons. Interestingly, we find that variation in firing rates is strongly coupled to variation in synaptic efficacies: neurons with the highest firing rates develop very strong connections onto other neurons. Finally, we define an impact measure for individual neurons and demonstrate the existence of a small group of neurons with an exceptionally strong impact on the network, that arise as a result of synaptic plasticity. In summary, synaptic plasticity and small variability in neuronal parameters underlie a neural oligarchy in recurrent neural networks. Author summary Our brain’s neural networks are composed of billions of neurons that exchange signals via trillions of synapses. Are these neurons created equal, or do they contribute in similar ways to the network dynamics? Or do some neurons wield much more power than others? Recent experiments have shown that some neurons are much more active than the average neuron and that some synaptic connections are much stronger than the average synaptic connection. However, it is still unclear how these properties come about in the brain. Here we present a neural network model that explains these findings as a result of the interaction of synaptic plasticity mechanisms that modify synapses’ efficacies. The model reproduces recent findings on the statistics of neuronal firing rates and synaptic efficacies and predicts a small class of neurons with exceptionally high impact on the network dynamics. Such neurons may play a key role in brain disorders such as epilepsy.
BibTeX
@misc{kleberg2018neural,
  author = {Florence I. Kleberg and Jochen Triesch},
  title = {{Neural oligarchy: how synaptic plasticity breeds neurons with extreme influence}},
  year = {2018},
  doi = {10.1101/361394},
  howpublished = {bioRxiv},
}
Figure