Local inhibitory plasticity tunes global brain dynamics and allows the emergence of functional brain networks

Peter J. Hellyer, Barbara Jachs, Robert Leech, Claudia ClopatharXiv 2014 · DOI · arXiv

Abstract

Rich, spontaneous brain activity has been observed across a range of different temporal and spatial scales. These dynamics are thought to be important for efficient neural functioning. Experimental evidence suggests that these neural dynamics are maintained across a variety of different cognitive states, in response to alterations of the environment and to changes in brain configuration (e.g., across individuals, development and in many neurological disorders). This suggests that the brain has evolved mechanisms to stabilize dynamics and maintain them across a range of situations. Here, we employ a local homeostatic inhibitory plasticity mechanism, balancing inhibitory and excitatory activity in a model of macroscopic brain activity based on white-matter structural connectivity. We demonstrate that the addition of homeostatic plasticity regulates network activity and allows for the emergence of rich, spontaneous dynamics across a range of brain configurations. Furthermore, the presence of homeostatic plasticity maximises the overlap between empirical and simulated patterns of functional connectivity. Therefore, this work presents a simple, local, biologically plausible inhibitory mechanism that allows stable dynamics to emerge in the brain and which facilitates the formation of functional connectivity networks.

Introduction: The Challenge of Brain Stability

Computational models of the brain often suffer from a "Goldilocks" problem. To simulate the rich, spontaneous activity seen in real brains—a state often called criticality—researchers usually have to manually fine-tune model parameters, such as global connection strength or noise levels, to a very narrow "sweet spot." If the parameters are slightly off, the model either goes silent or explodes into chaotic, "pathological" activity.

In reality, our brains are remarkably robust, maintaining stable dynamics despite development, aging, and changing environments. This research addresses the mystery of how the brain "self-tunes" to this optimal regime using only local biological rules, rather than global optimization.

Methods: Modeling the Self-Tuning Brain

The authors constructed a macroscopic computational model of the human brain consisting of 66 nodes, each representing a distinct cortical region.

Structural connectivity and distance matrices used to define the model's global architecture.

The model utilized Wilson-Cowan equations to simulate the interaction between pools of excitatory and inhibitory neurons. The nodes were connected based on real-world human white-matter structural connectivity derived from Diffusion Spectrum Imaging (DSI) of five healthy control subjects.

Schematic of the local Wilson-Cowan circuit with homeostatic inhibitory plasticity.

The key innovation was the introduction of a local homeostatic inhibitory plasticity rule. In this rule, the inhibitory connections within each node automatically adjust their strength to maintain a specific target level of excitatory activity ($\rho$). If a node is too active, inhibition increases; if it is too quiet, inhibition decreases. This allows the model to adapt to its own structural complexity without a global controller.

Results: From Local Rules to Global Criticality

The most striking finding is that a purely local rule—one that only "senses" activity within its own node—can tune the entire global network to a state of criticality.

Stability and Criticality

When the plasticity rule was active, the model naturally produced activity patterns (avalanches) that followed a power-law distribution, a hallmark of efficient information processing. The adaptive model maintained a criticality measure ($\kappa$) near 1.0 across almost the entire parameter space, whereas the non-adaptive model reached ceiling activity once connection strength exceeded a low threshold. The fit to power-law behavior was significantly better in the homeostatic mode ($t=59.01, p<0.001$).

Functional Network Emergence

The adaptive model was significantly better at replicating real-world functional connectivity (as measured by fMRI) than the non-adaptive version. Furthermore, the adaptive model showed significantly higher metastability ($t=17.81, p<0.001$), meaning brain regions could flexibly transition between being synchronized and independent—a key feature of cognitive flexibility.

Inhibition and Topology

Inhibitory weights evolved to reflect the node's position in the network. Nodes with high "clustering" or "local efficiency" required different inhibitory tuning, suggesting that inhibition is the brain's way of compensating for its own structural complexity.

Discussion and Future Work

This study presents a simple, local, and biologically plausible inhibitory mechanism that allows stable dynamics to emerge in the brain. While the model currently uses a 66-node atlas and a small subject sample, it provides a robust proof-of-concept for how the brain maintains stability.

Future work could explore higher-resolution models or the inclusion of diverse inhibitory interneuron types. Additionally, this mechanism could be highly useful in reservoir computing and machine learning, helping artificial agents maintain rich internal dynamics in changing environments.

Summary updated

Abstract

Rich, spontaneous brain activity has been observed across a range of different temporal and spatial scales. These dynamics are thought to be important t for efficient neural functioning. Experimental evidence suggests that these neural dynamics are maintained across a variety of different cognitive states, in response to alterations of the environment and to changes in brain configuration (e.g., across individuals, development and in many neurological disorders). This suggests that the brain has evolved mechanisms to stabilize dynamics and maintain them across a range of situations. Here, we employ a local homeostatic inhibitory plasticity mechanism, balancing inhibitory and excitatory activity in a model of macroscopic brain activity based on white-matter structural connectivity. We demonstrate that the addition of homeostatic plasticity regulates network activity and allows for the emergence of rich, spontaneous dynamics across a range of brain configurations. Furthermore, the presence of homeostatic plasticity maximises the overlap between empirical and simulated patterns of functional connectivity. Therefore, this work presents a simple, local, biologically plausible inhibitory mechanism that allows stable dynamics to emerge in the brain and which facilitates the formation of functional connectivity networks.
BibTeX
@misc{hellyer2014local,
  author = {Peter J. Hellyer and Barbara Jachs and Robert Leech and Claudia Clopath},
  title = {{Local inhibitory plasticity tunes global brain dynamics and allows the emergence of functional brain networks}},
  year = {2014},
  eprint = {1412.6325},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1412.6325},
}
Figure