Fluctuation-dissipation of the Kuramoto model on fruit-fly connectomes

Géza Ódor, István Papp, Gustavo Deco · 2025

Read Fluctuation-dissipation of the Kuramoto model on fruit-fly connectomes by Géza Ódor, István Papp, Gustavo Deco (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-20708.txt.

Introduction

Géza Ódor, István Papp, and Gustavo Deco studied sync in two thousand twenty-five in a model placed on the full brain wiring of a fruit fly. The Kuramoto model describes how coupled oscillators fall into step. The fly connectome holds about one hundred thousand neuron nodes, with weights that point more one way than the other.

Working near the sync transition, they measured how responses to small nudges drift from a balance rule that holds at equilibrium. That drift grew with the one-way bias of the wiring, with more swings over time. Ageing exponents matched a theory of broken time symmetry. A random graph of like size gave a baseline for contrast.

Catalog record

CREATORGéza Ódor, István Papp, Gustavo Deco
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE55 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-20708

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Fluctuation-dissipation of the Kuramoto model on fruit-fly connectomes Géza Ódor, István Papp, Gustavo Deco Abstract We investigate the distance from equilibrium using the Kuramoto model via the degree of fluctuation-dissipation violation as the consequence of different levels of edge weight anisotropies. This is achieved by solving the synchronization equations on the raw, homeostatic weighted and a random inhibitory edge variant of a real full fly (FF) connectome, containing ≃ 10 5 \simeq 10^{5} neuron cell nodes. We investigate these systems close to their synchronization transition critical points. While the topological(graph) dimension is high: d ≃ 6 d\simeq 6 the spectral dimensions of the variants, relevant in describing the synchronization behavior, are lower than the upper critical dimension: d s ≃ 2 < d c = 5 d_{s}\simeq 2<d_{c}=5 , suggesting relevant fluctuation effects and non mean-field scaling behavior. By measuring the auto-correlations and the auto-response functions for small perturbations we calculate the fluctuation-dissipation ratios (FDR) for the different variants of different anisotropy levels of the FF connectome. Numerical evidence is presented that the FDRs follow the level of anisotropy of these non-equilibrium systems in agreement with the expectations, with increasing temporal oscillations. Numerical values for the aging exponents provide confirmation for the predictions of the generalized time-translational-invariance theory. We compare the non-reciprocal connectome results with those on a symmetric Erdős-Rényi random graph of similar size and provide estimates for the Kuramoto model aging behavior. We also present a network analysis of the FF connectome and calculate the level of hierarchy, also related to the anisotropy. Finally, we provide some partial results for the periodically forced Shinomoto-Kuramoto model describing fluctuations in the non-resting state. We provide numerical evidence that fluctuations are maximal in the resting state of the critical brain. I Introduction Thanks to the explosion of informatics and precise neurodata analysis large structural connectomes are available in different repositories (see for example: [ 1 , 2 ] ) by now, but one has to be cautious with non-proofread ones [ 3 ] . Here we used the fruit-fly connectome as a complex, empirical case study of a directed graph, as close to the reality of a biological neural system as possible, rather than as the primary driver of the research. In this study, we harness this modeling framework to introduce perturbability metrics, which assess how external perturbations influence brain dynamics. Specifically, perturbability quantifies the brain’s susceptibility to external inputs, while perturbational recovery measures the system’s capacity to return from a pathological state toward a healthy dynamical regime. Crucially, the theoretical backbone of these measures is grounded in the Fluctuation-Dissipation Theorem (FDT) from statistical physics [ 4 ] . Neural systems, being inherently out of equilibrium, naturally exhibit complex fluctuations across spatial and temporal scales. The FDT characterizes the fundamental relationship between these spontaneous fluctuations and the system’s response to external perturbations. In the brain, this reflects a balance between internal noise and dissipation mechanisms. Deviations from FDT predictions highlight a departure from equilibrium, often marking an increased sensitivity or altered responsiveness to stimuli—a hallmark of pathological brain states. Non-equilibrium systems can be characterized by the violation of the fluctuation-dissipation (vFD) relation [ 5 , 6 ] ). It is related to the non-equilibrium nature of the dynamics; the prominent example is the case of systems with absorbing states [ 7 ] . Spatial anisotropy also breaks the detailed balance condition (necessary to be in equilibrium state). A prominent example is when we add an up/down symmetry-breaking term to surface growth, which changes the equilibrium Edwards Wilkinson growth to the non-equilibrium KPZ equation [ 7 ] . Similarly, when we add spatial anisotropy to the diffusion or reaction terms, like in the case of driven diffusive particle models, vFD is achieved. Modular structure, or other constraints in the dynamics, may also generate vFD. Very recently the vFD in neural systems has been studied by [ 8 , 9 ] and the relation between vFD and the asymmetry of the interactions has been demonstrated using empirical human neuroimaging data as well as whole-brain models. Using this approach they uncovered the thermodynamic hierarchy of conscious states and quantified disruptions in disorders of consciousness. They demonstrated that wakefulness shows significant violations of FDT, whereas deep sleep exhibits a major decrease in FDT violations, indicating a flattening of the brain’s hierarchical organization toward more symmetrical interactions. Evidence was found that global and resting-state network dynamics in patients with disorders of consciousness is closer to equilibrium than in healthy controls, decreasing stepwise with lower consciousness. In this work we extend this research, by taking into account the critical brain hypothesis [ 10 , 11 , 12 ] , i.e we investigate the vFD on large scale whole brain models tuned at the critical synchronization point. This is based on our previous whole brain Kuramoto model studies of different structural connectomes, such as the fruit-fly hemi-brain (HB) and large human gray mater graphs [ 13 , 14 , 15 , 16 ] . We investigate here what consequences of this hypothesis implies for the vFD results. Note, that for dynamical simulations we could have used large synthetic complex networks, with randomness as we had done before, like Erdős Rényi graphs [ 17 ] , Watts-Strogatz and Benjamini-Berger type long-link graphs [ 17 , 18 ] or hierarchical modular ones [ 19 , 20 ] , but these assume some features: dimensions, symmetries, modularity features… etc, relevant for the critical behavior. In order to extend the previous studies towards more brain like systems one should tuned them manually, before the simulations. Instead of that, now we decided to use a real FF connectome, which sets these relevant features. To be sure we present the necessary structural graph analysis in the first part of this publication and later we test by simulations what happens with the dynamical behavior if vary the level of interaction anisotropy, i.e. the reciprocity symmetry of edges. In this work we investigate the relation of anisotropy and non-equilibrium critical dynamical behavior within the framework of this connectome, using an oscillatory model for the function. We studied the network for three different anisotropy levels: the original network (o); the case with renormalized weights (w); and the case with 20\% negative weights chosen randomly to model inhibition (i), see section “ III.1.4 Asymmetry of the FF connectome variants ” . Far-from-equilibrium systems often display dynamical scaling, even if the stationary state is very far from being critical [ 21 , 22 ] . Criticality, often occurring at continuous, second-order phase transitions, is common in nature and beneficial for systems. In neural systems, criticality allows for the generation of working memory, spontaneous long-range interactions, and maximal sensitivity to external signals due to diverging correlations and fluctuations [ 23 , 24 , 14 ] . Information-processing is also optimal near the critical point [ 25 , 11 , 26 , 24 ] , leading systems to self-organize close to criticality (SOC) [ 27 , 11 ] , or slightly below it to avoid over-excitation. This can also happen in the dynamical, semi-critical so called Griffiths phase [ 28 , 29 , 30 , 19 , 31 ] . Another important, open question in the level of hierarchical anisotropy of

Read the complete text