GradStim · Modeling the perturbational gradients of the human brain

Modeling the perturbational gradients of the human brain

Context and Overall Objectives Understanding consciousness, what it is, how it arises, and how it fades, is one of the great scientific and philosophical challenges of our time. While we typically think of the brain as being organized into specialized regions that perform different functions, recent neuroscience has revealed that brain activity is better understood as flowing along continuous gradients: large-scale patterns that smoothly transition from sensory processing areas to regions involved in abstract thought and self-awareness. These “neural gradients” provide a powerful framework for investigating how different states of consciousness such as sleep, anesthesia, or the effects of certain drugs, alter the way the brain functions globally. This project aimed to study how these gradients change as we move from wakefulness into different unconscious states. More specifically, the project investigated whether the first principal brain gradient, a marker of large-scale cortical organization, contracts systematically as people fall asleep, become sedated, or undergo anesthesia. Such a contraction would indicate that the brain becomes less globally integrated, potentially reflecting both reduced awareness and lower responsiveness to external stimuli. In addition to studying empirical brain data derived form advanced neuroimaging techniques, the project also developed computational brain models that simulate how these gradients might emerge from the brain’s underlying structure and chemistry. These models were used not only to replicate the patterns seen in real human brain scans but also to test whether “virtual stimulation” could push the brain from one state (e.g. deep sleep) toward another (e.g. light sleep or wakefulness). Approach and Main Achievements The project combined three main lines of investigation. First, using functional MRI data from healthy volunteers and patients, we analyzed how the brain’s large-scale organization changes across different states of consciousness, including light and deep sleep, sedation with propofol, and full anesthesia. The data revealed a progressive contraction of the first functional gradient, with each step down in consciousness corresponding to a more localized and fragmented brain network. The ordering of states: light sleep, moderate sleep, sedation, deep sleep, and anesthesia; consistently followed the degree of gradient contraction, revealing a neural “scale” of consciousness. Second, to understand how these gradients arise, we simulated brain activity using mathematical models based on the Hopf bifurcation framework, grounded in principles from physics and neuroscience. These models incorporated anatomical data (structural connectivity) and neurochemical information (such as neurotransmitter receptor densities) to simulate brain dynamics under varying conditions. After optimization, the models closely matched empirical data and successfully reproduced realistic brain gradients. Testing various biological priors further showed that features like resting-state network maps improved simulation accuracy. Third, we explored whether simulated stimulation could shift brain states toward greater consciousness. We applied periodic perturbations to models optimized for low-consciousness states and assessed whether the resulting gradient signatures became more similar to those of more conscious states. In many cases, stimulation induced a meaningful shift toward the desired target, suggesting that brain-wide gradients are not only passive markers of consciousness but may also be useful tools for exploring how consciousness could be modulated through virtual interventions. This opens up new avenues for research into computational neuromodulation and potential applications in disorders of consciousness. Expected Impact and Significance This project bridges a crucial gap between neuroimaging, computational modeling, and clinical neuroscience. By identifying the contraction of the brain’s principal functional gradient as a reliable marker of decreasing levels of consciousness, it provides a robust, quantifiable method for monitoring brain states. This has direct implications for fields such as anesthesia monitoring, where assessing the depth of unconsciousness is critical, and sleep research, where transitions between different sleep stages can now be studied through a gradient-based lens. It also holds promise for improving the diagnosis and assessment of patients with disorders of consciousness, such as those in minimally conscious or unresponsive states, where objective biomarkers are urgently needed. In parallel, the successful reproduction of these gradients using computational brain models demonstrates that large-scale neural dynamics can be predicted, and potentially manipulated, using simulations informed by anatomical and neurochemical data. This not only validates the utility of whole-brain modeling in understanding consciousness but also lays the groundwork for future in silico neuroscience, where hypotheses can be explored through virtual experiments before being tested in clinical or laboratory settings. The broader impact of this work extends beyond neuroscience: by contributing to a mechanistic understanding of conscious awareness, the findings may inform the development of next-generation artificial intelligence systems modeled after the brain’s dynamic organization. All datasets were anonymized and processed in compliance with ethical standards. The simulation code and analysis tools are being prepared for open-access release, ensuring transparency, reproducibility, and broader adoption by the neuroscience research community.

Текст от CORDIS, на английски · Данни: CORDIS, © Европейски съюз