Patterned perturbation of inhibition can reveal the dynamical structure of neural processing

Sadra Sadeh, Claudia ClopatheLife 2020 · DOI

TL;DR

This paper introduces a novel method using "patterned perturbations" of neural activity to uncover the specific dynamics of inhibitory-stabilized networks (ISNs) in the cortex. By precisely targeting inhibitory neurons based on their functional properties, the authors demonstrate a unique "specific paradoxical effect" that reveals how these networks maintain stability and enable rapid transitions between selective neural states, crucial for complex brain functions.

What Problem Does This Paper Address?

Understanding how the brain processes information relies heavily on studying neuronal activity. A key challenge is that simply observing neurons or perturbing them non-specifically (e.g., activating or deactivating large groups randomly) often masks the intricate functional organization of neural circuits. Specifically, the brain's cortex is thought to operate in an "inhibitory-stabilized network" (ISN) regime, where strong excitatory activity is balanced by potent inhibition. However, recent findings suggest that this inhibition might not be uniform but rather "feature-specific," meaning it selectively targets and stabilizes specific groups of neurons that process similar information. The problem is that traditional, non-specific perturbation methods are insufficient to detect and characterize these feature-specific ISNs, leaving their dynamical properties and functional implications largely unexplored.

Methods

The authors employ computational modeling to investigate their hypothesis. They simulate large-scale networks of both rate-based and spiking neurons. The core methodological innovation lies in the patterned perturbation technique. Instead of randomly activating or deactivating inhibitory neurons, they selectively target them based on their functional properties, such as preferred orientation or receptive field (RF) similarity. This allows them to probe the network's response to perturbations that are specifically aligned with its internal functional structure. They introduce a quantitative measure, the "specific paradoxical effect," characterized by a negative slope in the relationship between inhibitory neuron input perturbation and their response change, as a signature of feature-specific ISNs. They also explore different perturbation strategies, including varying the stimuli used to define response similarity and the amplitude of perturbations.

Datasets

Not discussed in the paper. The study relies on simulated data generated from their computational models.

Key Insights and Ideas

  • Patterned perturbation is essential: Non-specific perturbations, while revealing general ISN properties, fail to uncover the more nuanced dynamics of feature-specific ISNs. Only perturbations aligned with the network's functional architecture can reveal these specific dynamics.
  • "Specific paradoxical effect" as a signature: A key finding is the identification of a "specific paradoxical effect" – a negative correlation between the magnitude of patterned inhibitory perturbation and the resulting change in inhibitory neuron activity. This effect is distinct from the general paradoxical effect seen in non-specific ISNs.
  • Functional specificity matters: The degree of specificity in both excitatory and inhibitory connections plays a critical role in network stability and dynamics. Feature-specific ISNs are crucial for balancing potentially unstable, feature-specific excitatory subnetworks.
  • Spontaneous activity reflects structure: Feature-specific ISNs exhibit spontaneous transitions between highly selective neural states, mirroring patterns observed in experimental studies of the visual cortex. This suggests that the network's internal structure dictates its spontaneous activity patterns.
  • Response similarity is a practical guide: While RF similarity can guide patterned perturbations, using response similarity to naturalistic stimuli proves to be a more effective and potentially more feasible experimental approach for identifying specific ISNs.

Key Results

  • Simulations of rate-based and spiking networks with feature-specific excitatory-inhibitory connectivity demonstrated the "specific paradoxical effect" when inhibitory neurons were perturbed in a patterned manner aligned with their functional properties.
  • Non-specific perturbations, even in networks with specific connectivity, only revealed the general paradoxical effect, failing to capture the specific dynamics.
  • Networks with feature-specific ISNs exhibited rapid spontaneous transitions between selective neural activity patterns, a phenomenon observed experimentally.
  • Perturbations based on RF similarity and, more effectively, response similarity to naturalistic stimuli, successfully revealed specific ISN properties.
  • The study provides a quantitative metric (the slope of the response change vs. input perturbation) for identifying feature-specific ISNs.

Limitations and Future Work

  • Experimental validation: While the study provides strong theoretical and simulation-based evidence, direct experimental validation in biological systems is the crucial next step. The authors suggest specific experimental protocols.
  • Complexity of biological networks: The models, while incorporating increasing realism (e.g., spiking neurons, more realistic RFs), are still simplifications of the vastly complex biological cortex. Future work could explore the role of different interneuron subtypes and more intricate connectivity patterns.
  • Stimulus choice for response similarity: The effectiveness of using response similarity as a perturbation guide depends on the choice of stimuli. Naturalistic stimuli were found to be more effective than simplified gratings, suggesting that the ecological relevance of stimuli is important.
  • Mechanism of specific ISN emergence: The paper touches upon potential mechanisms for the emergence of feature-specific ISNs (e.g., inhibitory plasticity) and suggests future studies to explore these.
  • Clinical implications: The authors hint at potential applications in understanding neurological disorders, such as specific epileptic modes of activity, which could be a future research direction.

Summary updated

Abstract

Perturbation of neuronal activity is key to understanding the brain's functional properties, however, intervention studies typically perturb neurons in a nonspecific manner. Recent optogenetics techniques have enabled patterned perturbations, in which specific patterns of activity can be invoked in identified target neurons to reveal more specific cortical function. Here, we argue that patterned perturbation of neurons is in fact necessary to reveal the specific dynamics of inhibitory stabilization, emerging in cortical networks with strong excitatory and inhibitory functional subnetworks, as recently reported in mouse visual cortex. We propose a specific perturbative signature of these networks and investigate how this can be measured under different experimental conditions. Functionally, rapid spontaneous transitions between selective ensembles of neurons emerge in such networks, consistent with experimental results. Our study outlines the dynamical and functional properties of feature-specific inhibitory-stabilized networks, and suggests experimental protocols that can be used to detect them in the intact cortex.
BibTeX
@inproceedings{sadeh2020patterned,
  author = {Sadra Sadeh and Claudia Clopath},
  title = {{Patterned perturbation of inhibition can reveal the dynamical structure of neural processing}},
  booktitle = {eLife},
  year = {2020},
  doi = {10.7554/elife.52757},
}
Figure