Identifying the impact of local connectivity patterns on dynamics in excitatory-inhibitory networks

Yuxiu Shao, David Dahmen, Stefano Recanatesi, Eric Shea-Brown, Srdjan OstojicarXiv 2024 · DOI · arXiv

Introduction

Standard models of brain circuits often assume connections are random, but real biology is far more structured. While synaptic strengths depend on cell types (Excitatory vs. Inhibitory), they are often treated as statistically independent. However, recent high-resolution connectomics data from mice and humans have revealed that the brain is rich in specific "motifs" or local connectivity patterns. This paper investigates how these non-random patterns, specifically "chain motifs," fundamentally reshape network stability and dynamics.

Defining Connectivity Motifs

The researchers focus on "second-order motifs"—correlations between pairs of synapses. While previous work often highlighted reciprocal connections (where neuron A connects to B and B connects back to A), this study emphasizes the understudied impact of chain motifs. A chain motif occurs when a neuron that receives a strong input also provides a strong output to another neuron.

Illustration of network populations, the connectivity matrix J, and various local motifs including reciprocal, chain, divergent, and convergent patterns.

Methodology: Bridging Local Structure and Global Dynamics

To analyze these complex networks, the authors developed an analytical framework using:

  • Low-Rank Approximation: Approximating the $N \times N$ connectivity matrix using its dominant eigenvalues (outliers) to describe recurrent activity in fewer dimensions.
  • Eigenvalue Outlier Analysis: Deriving how specific motifs "push" eigenvalues out of the random bulk, predicting when a network will become unstable.
  • Effective Connectivity Matrix: A simplified deterministic matrix that incorporates motif strengths to predict network responses.

The Disproportionate Influence of Chain Motifs

A key insight is that not all motifs are created equal. While reciprocal, convergent, and divergent motifs have negligible effects on the dominant eigenvalues in large networks, chain motifs exert a massive influence. This influence scales with the network size ($N$), meaning even a subtle over-representation of chains in a large brain circuit can fundamentally alter its stability.

Eigenvalue distributions showing how chain-motif strength generates positive outliers that scale with network size, potentially leading to instability.

Key Results and Implications

The study demonstrates that chain motifs generate specific positive outliers in the eigenspectrum. In inhibition-dominated networks, which are typically stable, an excess of chain motifs can create strong positive feedback loops, leading to runaway activity.

Furthermore, chain motifs can act as a "control knob" for the "paradoxical response"—a phenomenon where stimulating inhibitory neurons actually decreases their activity due to recurrent feedback. The authors show that these motifs can induce or suppress such responses regardless of average connection strengths, suggesting that classical E-I balance criteria must be updated.

Conclusion and Future Directions

This work highlights that the brain's "non-randomness" is not just a structural curiosity but a primary driver of its dynamical state. While this study focuses on linear rate units, future work will need to explore how these findings translate to non-linear regimes and spiking neurons. Understanding these "dynamic motifs" may reveal how the brain tunes its own stability through synaptic plasticity.

Summary updated

Abstract

Networks of excitatory and inhibitory (EI) neurons form a canonical circuit in the brain. Seminal theoretical results on dynamics of such networks are based on the assumption that synaptic strengths depend on the type of neurons they connect, but are otherwise statistically independent. Recent synaptic physiology datasets however highlight the prominence of specific connectivity patterns that go well beyond what is expected from independent connections. While decades of influential research have demonstrated the strong role of the basic EI cell type structure, to which extent additional connectivity features influence dynamics remains to be fully determined. Here we examine the effects of pairwise connectivity motifs on the linear dynamics in EI networks using an analytical framework that approximates the connectivity in terms of low-rank structures. This low-rank approximation is based on a mathematical derivation of the dominant eigenvalues of the connectivity matrix and predicts the impact on responses to external inputs of connectivity motifs and their interactions with cell-type structure. Our results reveal that a particular pattern of connectivity, chain motifs, have a much stronger impact on dominant eigenmodes than other pairwise motifs. An overrepresentation of chain motifs induces a strong positive eigenvalue in inhibition-dominated networks and generates a potential instability that requires revisiting the classical excitation-inhibition balance criteria. Examining effects of external inputs, we show that chain motifs can on their own induce paradoxical responses where an increased input to inhibitory neurons leads to a decrease in their activity due to the recurrent feedback. These findings have direct implications for the interpretation of experiments in which responses to optogenetic perturbations are measured and used to infer the dynamical regime of cortical circuits.
BibTeX
@misc{shao2024identifying,
  author = {Yuxiu Shao and David Dahmen and Stefano Recanatesi and Eric Shea-Brown and Srdjan Ostojic},
  title = {{Identifying the impact of local connectivity patterns on dynamics in excitatory-inhibitory networks}},
  year = {2024},
  eprint = {2411.06802},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2411.06802},
}
Figure