Inhibition stabilization is a widespread property of cortical networks

eLife, Journal Year: 2020, Volume and Issue: 9

Published: June 29, 2020

Many cortical network models use recurrent coupling strong enough to require inhibition for stabilization. Yet it has been experimentally unclear whether inhibition-stabilized (ISN) describe function well across areas and states. Here, we test several ISN predictions, including the counterintuitive (paradoxical) suppression of inhibitory firing in response optogenetic stimulation. We find clear evidence operation mouse visual, somatosensory, motor cortex. Simple two-population data let us quantify strength. Although some predict a non-ISN transition with increasingly sensory stimuli, effects without stimulation even during light anesthesia. Additionally, average paradoxical result only transgenic, not viral, opsin expression parvalbumin (PV)-positive neurons; theory show this is consistent operation. Taken together, these results stabilization are common features

Language: Английский

---

Context Changes Everything

The MIT Press eBooks, Journal Year: 2023, Volume and Issue: unknown

Published: March 1, 2023

From the influential author of Dynamics in Action, how concepts constraints provide a way to rethink relationships, opening intentional, meaningful causation. Grounding her work problem causation, Alicia Juarrerochallenges previously held beliefs that only forceful impacts are causes. Constraints, she claims, bring about effects as well, and they enable emergence coherence. In Context Changes Everything, Juarrero shows coherence is induced by enabling constraints, not causes, resulting then maintained constitutive constraints. Constitutive turn, become governing regulate modulate coherent entities behave. Using tools complexity science, offers rigorously scientific understanding identity, hierarchy, top-down so doing, presents new thinking natural world. argues personal which has been thought be conferred through internal traits (essential natures), grounded dynamic interdependencies keep structures whole. This challenges our ideas well notion stability means inflexible rigidity. On contrary, stable brittle cannot persist. Complexity says Juarrero, can shape we meet world, what emerges from interactions finds coherence, humans identities robust resilient. framework significant implications for sociology, economics, political theory, business, knowledge management, psychology, religion, theology. It points more expansive synthetic philosophy who living nonliving things alike.

Language: Английский

---

Excitation–Inhibition Balance, Neural Criticality, and Activities in Neuronal Circuits

The Neuroscientist, Journal Year: 2024, Volume and Issue: unknown

Published: Jan. 31, 2024

Neural activities in local circuits exhibit complex and multilevel dynamic features. Individual neurons spike irregularly, which is believed to originate from receiving balanced amounts of excitatory inhibitory inputs, known as the excitation–inhibition balance. The spatial-temporal cascades clustered neuronal spikes occur variable sizes durations, manifested neural avalanches with scale-free These may be explained by criticality hypothesis, posits that systems operate around transition between distinct states. Here, we summarize experimental evidence for underlying theory balance criticality. Furthermore, review recent studies excitatory–inhibitory networks synaptic kinetics a simple solution reconcile these two apparently theories single circuit model. This provides more unified understanding circuits, spontaneous stimulus-response dynamics.

Language: Английский

---

Approximating Nonlinear Functions With Latent Boundaries in Low-Rank Excitatory-Inhibitory Spiking Networks

Neural Computation, Journal Year: 2024, Volume and Issue: 36(5), P. 803 - 857

Published: April 23, 2024

Abstract Deep feedforward and recurrent neural networks have become successful functional models of the brain, but they neglect obvious biological details such as spikes Dale’s law. Here we argue that these are crucial in order to understand how real circuits operate. Towards this aim, put forth a new framework for spike-based computation low-rank excitatory-inhibitory spiking networks. By considering populations with rank-1 connectivity, cast each neuron’s threshold boundary low-dimensional input-output space. We then show combined thresholds population inhibitory neurons form stable space, those excitatory an unstable boundary. Combining two boundaries results rank-2 (EI) network inhibition-stabilized dynamics at intersection boundaries. The resulting can be understood difference convex functions is thereby capable approximating arbitrary non-linear mappings. demonstrate several properties networks, including noise suppression amplification, irregular activity synaptic balance, well relate rate limit becomes soft. Finally, while our work focuses on small (5-50 neurons), discuss potential avenues scaling up much larger Overall, proposes perspective may serve starting point mechanistic understanding computation.

Language: Английский

---

Synapse-type-specific competitive Hebbian learning forms functional recurrent networks

Proceedings of the National Academy of Sciences, Journal Year: 2024, Volume and Issue: 121(25)

Published: June 13, 2024

Cortical networks exhibit complex stimulus–response patterns that are based on specific recurrent interactions between neurons. For example, the balance excitatory and inhibitory currents has been identified as a central component of cortical computations. However, it remains unclear how required synaptic connectivity can emerge in developing circuits where synapses neurons simultaneously plastic. Using theory modeling, we propose wide range response properties arise from single plasticity paradigm acts at all connections—Hebbian learning is stabilized by synapse-type-specific competition for limited supply resources. In plastic circuits, this enables formation decorrelation inhibition-balanced receptive fields. Networks develop an assembly structure with stronger connections similarly tuned normalization orientation-specific center-surround suppression, reflecting stimulus statistics during training. These results demonstrate self-organize into functional suggest essential role competitive development circuits.

Language: Английский

---

Critical Avalanches in Excitation-Inhibition Balanced Networks Reconcile Response Reliability with Sensitivity for Optimal Neural Representation

Physical Review Letters, Journal Year: 2025, Volume and Issue: 134(2)

Published: Jan. 15, 2025

Neural criticality has emerged as a unified framework that reconciles diverse multiscale neuronal dynamics such the irregular firing of individual neurons, sparse synchrony in populations, and emergence scale-free avalanches. However, functional role remains ambiguous. Here, we investigate neural representations response to external signals excitation-inhibition balanced networks. We reveal that, contrast with case for traditional critical branching model, state network simultaneously achieves maximal sensitivity, reliability, optimal representation due presence reliable avalanches induced by signals. further demonstrate heterogeneity inhibitory connections is mechanism underlying representation. Our study addresses longstanding challenge concerning significance criticality, namely intricate coexistence reliability sensitivity.

Language: Английский

---

Computational functions of precisely balanced neuronal microcircuits in an olfactory memory network

Cell Reports, Journal Year: 2025, Volume and Issue: 44(3), P. 115330 - 115330

Published: Feb. 20, 2025

Language: Английский

---

Theory of Gating in Recurrent Neural Networks

Physical Review X, Journal Year: 2022, Volume and Issue: 12(1)

Published: Jan. 18, 2022

Recurrent neural networks (RNNs) are powerful dynamical models, widely used in machine learning (ML) and neuroscience. Prior theoretical work has focused on RNNs with additive interactions. However gating i.e., multiplicative interactions ubiquitous real neurons also the central feature of best-performing ML. Here, we show that offers flexible control two salient features collective dynamics: (i) timescales (ii) dimensionality. The gate controlling leads to a novel marginally stable state, where network functions as integrator. Unlike previous approaches, permits this important function without parameter fine-tuning or special symmetries. Gates provide flexible, context-dependent mechanism reset memory trace, thus complementing function. modulating dimensionality can induce novel, discontinuous chaotic transition, inputs push system strong activity, contrast typically stabilizing effect inputs. At unlike RNNs, proliferation critical points (topological complexity) is decoupled from appearance dynamics (dynamical complexity). rich summarized phase diagrams, providing map for principled initialization choices ML practitioners.

Language: Английский

---

Early selection of task-relevant features through population gating

Nature Communications, Journal Year: 2023, Volume and Issue: 14(1)

Published: Oct. 27, 2023

Abstract Brains can gracefully weed out irrelevant stimuli to guide behavior. This feat is believed rely on a progressive selection of task-relevant across the cortical hierarchy, but specific across-area interactions enabling stimulus are still unclear. Here, we propose that population gating, occurring within primary auditory cortex (A1) controlled by top-down inputs from prelimbic region medial prefrontal (mPFC), support selection. Examining single-unit activity recorded while rats performed an context-dependent task, found A1 encoded relevant and along common dimension its neural space. Yet, encoding was enhanced extra dimension. In turn, mPFC only ongoing context. To identify candidate mechanisms for A1, reverse-engineered low-rank RNNs trained similar task. Our analyses predicted two context-modulated populations gated their preferred in opposite contexts, which confirmed further A1. Finally, show two-region RNN how gating could be PFC, flexible communication despite fixed inter-areal connectivity.

Language: Английский

---

Top–down modulation in canonical cortical circuits with short-term plasticity

Proceedings of the National Academy of Sciences, Journal Year: 2024, Volume and Issue: 121(16)

Published: April 9, 2024

Cortical dynamics and computations are strongly influenced by diverse GABAergic interneurons, including those expressing parvalbumin (PV), somatostatin (SST), vasoactive intestinal peptide (VIP). Together with excitatory (E) neurons, they form a canonical microcircuit exhibit counterintuitive nonlinear phenomena. One instance of such phenomena is response reversal, whereby SST neurons show opposite responses to top–down modulation via VIP depending on the presence bottom–up sensory input, indicating that network may function in different regimes under stimulation conditions. Combining analytical computational approaches, we demonstrate model networks multiple interneuron subtypes experimentally identified short-term plasticity mechanisms can implement reversal. Surprisingly, despite not directly affecting activity, PV-to-E depression has decisive impact We how reversal relates inhibition stabilization paradoxical effect several demonstrating coincides change indispensability for stabilization. In summary, our work suggests role generating makes testable predictions.

Language: Английский

---