Partial entropy decomposition reveals higher-order information structures in human brain activity

Proceedings of the National Academy of Sciences, Journal Year: 2023, Volume and Issue: 120(30)

Published: July 19, 2023

The standard approach to modeling the human brain as a complex system is with network, where basic unit of interaction pairwise link between two regions. While powerful, this limited by inability assess higher-order interactions involving three or more elements directly. In work, we explore method for capturing dependencies in multivariate data: partial entropy decomposition (PED). Our decomposes joint whole into set nonnegative atoms that describe redundant, unique, and synergistic compose system's structure. PED gives insight mathematics functional connectivity its limitation. When applied resting-state fMRI data, find robust evidence synergies are largely invisible analyses. can also be localized time, allowing frame-by-frame analysis how distributions redundancies change over course recording. We different ensembles regions transiently from being redundancy-dominated synergy-dominated temporal pattern structured time. These results provide strong there exists large space unexplored structures data have been missed focus on bivariate network models. This structure dynamic time likely will illuminate interesting links behavior. Beyond brain-specific application, provides very general understanding variety systems.

Language: Английский

---

Multivariate information theory uncovers synergistic subsystems of the human cerebral cortex

Communications Biology, Journal Year: 2023, Volume and Issue: 6(1)

Published: April 24, 2023

One of the most well-established tools for modeling brain is functional connectivity network, which constructed from pairs interacting regions. While powerful, network model limited by restriction that only pairwise dependencies are considered and potentially higher-order structures missed. Here, we explore how multivariate information theory reveals in human brain. We begin with a mathematical analysis O-information, showing analytically numerically it related to previously established theoretic measures complexity. then apply O-information data, synergistic subsystems widespread Highly typically sit between canonical networks, may serve an integrative role. use simulated annealing find maximally subsystems, finding such systems comprise ≈10 regions, recruited multiple systems. Though ubiquitous, highly invisible when considering connectivity, suggesting form kind shadow structure has been unrecognized network-based analyses. assert interactions represent under-explored space that, accessible theory, offer novel scientific insights.

Language: Английский

---

Brain network integration dynamics are associated with loss and recovery of consciousness induced by sevoflurane

Human Brain Mapping, Journal Year: 2021, Volume and Issue: 42(9), P. 2802 - 2822

Published: March 19, 2021

The dynamic interplay of integration and segregation in the brain is at core leading theoretical accounts consciousness. human dynamically alternates between a sub-state where predominates, predominantly segregated sub-state, with different roles supporting cognition behaviour. Here, we combine graph theory functional connectivity to compare resting-state MRI data from healthy volunteers before, during, after loss responsiveness induced concentrations inhalational anaesthetic, sevoflurane. We show that states characterised by high are especially vulnerable general anaesthesia, exhibiting attenuated complexity diminished small-world character. Crucially, these effects reversed upon recovery, demonstrating their association Higher doses sevoflurane (3% vol burst-suppression) also compromise temporal balance brain. Additionally, demonstrate reduced anticorrelations brain's default mode executive control networks reconfigure depending on state or segregation. Taken together, our results integrated terms both its information capacity, whose breakdown represents generalisable biomarker consciousness recovery.

Language: Английский

---

Integrated information as a common signature of dynamical and information-processing complexity

Chaos An Interdisciplinary Journal of Nonlinear Science, Journal Year: 2022, Volume and Issue: 32(1)

Published: Jan. 1, 2022

The apparent dichotomy between information-processing and dynamical approaches to complexity science forces researchers choose two diverging sets of tools explanations, creating conflict often hindering scientific progress. Nonetheless, given the shared theoretical goals both approaches, it is reasonable conjecture existence underlying common signatures that capture interesting behavior in systems. Here, we argue a pragmatic use integrated information theory (IIT), originally conceived neuroscience, can provide potential unifying framework study general multivariate By leveraging metrics put forward by decomposition framework, our results reveal effectively surprisingly heterogeneous complexity-including metastability criticality networks coupled oscillators as well distributed computation emergent stable particles cellular automata-without relying on idiosyncratic, ad hoc criteria. These show how an agnostic IIT important steps toward bridging gap informational complex

Language: Английский

---

Whole-brain modelling identifies distinct but convergent paths to unconsciousness in anaesthesia and disorders of consciousness

Communications Biology, Journal Year: 2022, Volume and Issue: 5(1)

Published: April 20, 2022

The human brain entertains rich spatiotemporal dynamics, which are drastically reconfigured when consciousness is lost due to anaesthesia or disorders of (DOC). Here, we sought identify the neurobiological mechanisms that explain how transient pharmacological intervention and chronic neuroanatomical injury can lead common reconfigurations neural activity. We developed systematically perturbed a neurobiologically realistic model whole-brain haemodynamic signals. By incorporating PET data about cortical distribution GABA receptors, our computational reveals key role spatially-specific local inhibition for reproducing functional MRI activity observed during with GABA-ergic agent propofol. Additionally, diffusion obtained from DOC patients dynamics characterise loss also emerge randomised connectivity. Our results generalise between datasets, demonstrating increased connectome perturbation represent distinct paths towards characteristic unconscious brain.

Language: Английский

---

Decomposing past and future: Integrated information decomposition based on shared probability mass exclusions

PLoS ONE, Journal Year: 2023, Volume and Issue: 18(3), P. e0282950 - e0282950

Published: March 23, 2023

A core feature of complex systems is that the interactions between elements in present causally constrain their own futures, and futures other as system evolves through time. To fully model all these (between elements, well ensembles elements), it possible to decompose total information flowing from past future into a set non-overlapping temporal describe different modes by which can be stored, transferred, or modified. achieve this, I propose novel information-theoretic measure dependency ( τsx ) based on logic local probability mass exclusions. This integrated decomposition reveal emergent higher-order within dynamics system, refining existing measures. demonstrate utility this framework, apply spontaneous spiking activity recorded dissociated neural cultures rat cerebral cortex show how processing are distributed over system. Furthermore, being localizable analysis, provide insight computational structure single moments. explore time-resolved neuronal avalanches find types atoms have distinct profiles course an avalanche, with majority non-trivial happening before first half cascade completed. These analyses allow us move beyond historical focus measures such transfer integration, panoply relationships (and groups elements) systems.

Language: Английский

---

Reduced emergent character of neural dynamics in patients with a disrupted connectome

NeuroImage, Journal Year: 2023, Volume and Issue: 269, P. 119926 - 119926

Published: Feb. 3, 2023

High-level brain functions are widely believed to emerge from the orchestrated activity of multiple neural systems. However, lacking a formal definition and practical quantification emergence for experimental data, neuroscientists have been unable empirically test this long-standing conjecture. Here we investigate fundamental question by leveraging recently proposed framework known as "Integrated Information Decomposition," which establishes principled information-theoretic approach operationalise quantify in dynamical systems - including human brain. By analysing functional MRI our results show that emergent hierarchical character dynamics is significantly diminished chronically unresponsive patients suffering severe injury. At level, demonstrate capacity positively correlated with extent organisation activity. Furthermore, combining computational approaches network control theory whole-brain biophysical modelling, reduced severely brain-injured can be mechanistically explained disruptions patients' structural connectome. Overall, suggest chronic unresponsiveness resulting injury may related impairment infrastructures required support emergence.

Language: Английский

---

Local orchestration of distributed functional patterns supporting loss and restoration of consciousness in the primate brain

Nature Communications, Journal Year: 2024, Volume and Issue: 15(1)

Published: March 11, 2024

Abstract A central challenge of neuroscience is to elucidate how brain function supports consciousness. Here, we combine the specificity focal deep stimulation with fMRI coverage entire cortex, in awake and anaesthetised non-human primates. During propofol, sevoflurane, or ketamine anaesthesia, subsequent restoration responsiveness by electrical thalamus, investigate loss consciousness impacts distributed patterns structure-function organisation across scales. We report that activity under anaesthesia increasingly constrained structure scales, coinciding anaesthetic-induced collapse multiple dimensions hierarchical cortical organisation. These signatures are observed different anaesthetics, they reversed recovery behavioural markers arousal. No such effects were upon stimulating ventral lateral demonstrating specificity. Overall, identify consistent orchestrated specific thalamic nuclei.

Language: Английский

---

Contributions of network structure, chemoarchitecture and diagnostic categories to transitions between cognitive topographies

Nature Biomedical Engineering, Journal Year: 2024, Volume and Issue: 8(9), P. 1142 - 1161

Published: Aug. 5, 2024

Abstract The mechanisms linking the brain’s network structure to cognitively relevant activation patterns remain largely unknown. Here, by leveraging principles of control, we show how architecture human connectome shapes transitions between 123 experimentally defined cognitive maps (cognitive topographies) from NeuroSynth meta-analytic database. Specifically, systematically integrated large-scale multimodal neuroimaging data functional magnetic resonance imaging, diffusion tractography, cortical morphometry and positron emission tomography simulate anatomically guided states can be reshaped neurotransmitter engagement or changes in thickness. Our model incorporates neurotransmitter-receptor density (18 receptors transporters) thickness pertaining a wide range mental health, neurodegenerative, psychiatric neurodevelopmental diagnostic categories (17,000 patients 22,000 controls). results provide comprehensive look-up table charting brain organization chemoarchitecture interact manifest different topographies, establish principled foundation for systematic identification ways promote selective topographies.

Language: Английский

---

Mechanisms Underlying Disorders of Consciousness: Bridging Gaps to Move Toward an Integrated Translational Science

Neurocritical Care, Journal Year: 2021, Volume and Issue: 35(S1), P. 37 - 54

Published: July 1, 2021

In order to successfully detect, classify, prognosticate, and develop targeted therapies for patients with disorders of consciousness (DOC), it is crucial improve our mechanistic understanding how severe brain injuries result in these disorders.

Language: Английский

---