Non-equilibrium physics of macroscopic brain dynamics
Loading...
Document Type
Document Version
Author
Director
Sebastián Gallés, Núria
Deco, Gustavo
Deco, Gustavo
Tutor
Deco, Gustavo
Other authors
Publication Date
Pages
184 p.
Embargo date
License
Citation
Geli, S. Non-equilibrium physics of macroscopic brain dynamics. Universitat Pompeu Fabra; 2026. handle: https://hdl.handle.net/10803/697297
This citation was generated automatically.
Citation
Doctoral program
Universitat Pompeu Fabra. Doctorat en Biomedicina
Abstract
Cada idea, sensación o experiencia consciente involucra un proceso físico irreversible en el cual nuestro cerebro consume energía, disipa calor y genera entropía. Este fenómeno es una consecuencia necesaria del procesamiento de información: emerge de la dinámica fuera del equilibrio de miles de millones de neuronas y células gliales. A pesar de décadas de investigación, aún no existe un formalismo que explique cómo estas dinámicas fuera del equilibrio termodinámico se relacionan con la cognición y la consciencia. En esta tesis desarrollamos medidas cuantitativas inspiradas en la física del no-equilibrio para estudiar la dinámica cerebral a gran escala, y las ponemos a prueba en datos de resonancia magnética funcional (fMRI) obtenidos de adultos sanos. Primero, estudiamos cómo las dinámicas fuera del equilibrio emergen de distintos tipos de interacción entre regiones cerebrales. Para ello, demostramos que la producción de entropía está explicada principalmente por dinámicas de pares, mientras que las dinámicas de orden superior (entre tres o más regiones) contribuyen en menor medida. Luego, introducimos un método para localizar las regiones que contribuyen más a la irreversibilidad, y exploramos mecanismos que explican esta ruptura de la simetría temporal. Finalmente, utilizamos un modelo dinámico para mostrar cómo los ``efectos de memoria'' en las señales de neuroimagen explican el balance entre procesamiento de información local y distribuido. En su conjunto, este trabajo aporta nuevas medidas para comprender los costos termodinámicos del procesamiento de información en el cerebro.
Every idea, sensation, or conscious experience involves an irreversible physical process in which our brain consumes energy, dissipates heat and produces entropy. This phenomenon is a necessary consequence of information processing: it emerges from the non-equilibrium dynamics of billions of neurons and glial cells. Despite decades of research, we still lack a framework that explains how these non-equilibrium dynamics relate to cognition and consciousness. In this thesis we develop quantitative measures inspired by non-equilibrium physics to study large-scale brain dynamics, and we test them using functional magnetic resonance imaging (fMRI) data from healthy adults. First, we study how non-equilibrium dynamics emerge from different types of interactions between brain regions. We demonstrate that entropy production is mainly explained by pairwise dynamics, while higher-order dynamics (between three or more regions) provide smaller contributions. Second, we introduce a method to localize brain regions that contribute most to irreversible dynamics, and we explore mechanisms underlying the breaking of time-reversal symmetry. Finally, we utilize a dynamical model to show how "memory effects" in neuroimaging time series explain the balance between local and distributed information processing. Overall, this work provides new measures to understand the thermodynamic costs of information processing in the brain.
Every idea, sensation, or conscious experience involves an irreversible physical process in which our brain consumes energy, dissipates heat and produces entropy. This phenomenon is a necessary consequence of information processing: it emerges from the non-equilibrium dynamics of billions of neurons and glial cells. Despite decades of research, we still lack a framework that explains how these non-equilibrium dynamics relate to cognition and consciousness. In this thesis we develop quantitative measures inspired by non-equilibrium physics to study large-scale brain dynamics, and we test them using functional magnetic resonance imaging (fMRI) data from healthy adults. First, we study how non-equilibrium dynamics emerge from different types of interactions between brain regions. We demonstrate that entropy production is mainly explained by pairwise dynamics, while higher-order dynamics (between three or more regions) provide smaller contributions. Second, we introduce a method to localize brain regions that contribute most to irreversible dynamics, and we explore mechanisms underlying the breaking of time-reversal symmetry. Finally, we utilize a dynamical model to show how "memory effects" in neuroimaging time series explain the balance between local and distributed information processing. Overall, this work provides new measures to understand the thermodynamic costs of information processing in the brain.
Keywords
Somputational Neuroscience, Whole-brain Dynamics, Non-equilibrium Thermodynamics, Cost of Neural Processing, FMRI, Neurociencia Computacional, Dinámica Cerebral Macroscópica, Física del No-Equilibrio, Coste del Procesamiento Neural, Resonancia Magnética Funcional
Subjects
616.8 - Neurology. Neuropathology. Nervous system
Publisher
Universitat Pompeu Fabra
