TLDR: This research introduces a dual computational framework to model human ‘gravity-awareness’ in altered gravity environments. It uses an EEG Fourier MLP to predict brain activity changes and an IGP-Physio model for physiological responses (HRV, EDA, motor behavior). Both models were trained on literature-derived data. Additionally, a large language model (Claude 3.5 Sonnet) simulated subjective human experiences, which aligned well with the quantitative findings. The framework provides insights into how the brain and body adapt to microgravity, partial gravity, and hypergravity, offering valuable tools for personalized astronaut training and monitoring human performance in space.
Our planet’s gravity has profoundly shaped human development, guiding our brains to integrate sensory information into an internal model of gravity. This internal model allows us to predict and interpret gravitational forces, forming what researchers call ‘Gravity-Awareness’. A new study delves into how this awareness is affected when humans encounter altered gravitational environments, ranging from the weightlessness of space to the intense forces of spacecraft launch and re-entry.
The research, titled Gravity-Awareness: Deep Learning Models and LLM Simulation of Human Awareness in Altered Gravity, presents a novel dual computational framework to quantitatively model these adaptations. Authored by Bakytzhan Alibekov, Alina Gutoreva, and Elisa Raffaela Ferre, the study combines deep learning models with large language model (LLM) simulations to offer a comprehensive understanding of human responses to varying gravity.
Modeling the Brain’s Response
The first component of this framework is a lightweight Multi-Layer Perceptron (MLP) model, called EEG Fourier MLP. This model predicts how different brainwave frequencies (EEG bands) change depending on the gravitational load. These brainwave changes are crucial indicators of the brain’s cortical state. For instance, the model tracks alpha-band activity, associated with the Default Mode Network (DMN) and reflective, calm states, and beta-band activity, linked to the Prefrontal Cortex (PFC) and alertness or cognitive workload.
The EEG Fourier MLP was trained on a large synthetic dataset, carefully constructed from an extensive review of parabolic flight literature. This allowed the researchers to simulate neural activity across a continuous spectrum of gravitational forces, from microgravity (0g) to hypergravity (1.8g), ensuring the model’s predictions are grounded in established neurophysiological findings.
Tracking the Body’s Physiological State
Complementing the brain activity model, the second component, named IGP-Physio (Independent Gaussian Processes for Physiological State Prediction), focuses on the body’s broader physiological state. This model uses Gaussian Processes to predict changes in various physiological parameters, including Heart Rate Variability (HRV), Electrodermal Activity (EDA), and motor behavior like trunk and wrist activity.
Similar to the EEG model, the IGP-Physio model was trained using data derived from parabolic flight studies, using published findings as ‘anchor points’ to build robust, continuous functions. This approach allows for flexible, noise-aware modeling of how different body systems respond to gravity changes.
Simulating Subjective Experience with AI
Beyond quantitative modeling, the study introduced a unique qualitative approach: using a large language model, Claude 3.5 Sonnet, to simulate subjective human experience under different gravitational loads. The LLM was prompted with physiological parameters and asked to generate introspective narratives of alertness and self-awareness across conditions like microgravity (0g), lunar gravity (0.17g), Martian gravity (0.38g), Earth gravity (1g), and hypergravity during take-off (4g) and re-entry (6g).
Remarkably, Claude’s narratives closely aligned with the quantitative findings from both the EEG and physiological models. For example, in microgravity, the LLM described disorientation and a ‘floating, untethered’ feeling, consistent with the EEG model’s prediction of suppressed alpha activity and the physiological model’s findings of high vagal tone and reduced trunk muscle engagement. In hypergravity, Claude reported immense physiological strain and narrowed self-awareness, mirroring the models’ predictions of high cortical stress and extreme sympathetic arousal.
Key Insights into Gravity-Awareness
The combined framework revealed several consistent patterns. In microgravity, the brain shows suppressed alpha and mu rhythms, indicating altered resting-state cognition and a shift in sensorimotor strategy due to postural unloading. In hypergravity, there’s an enhancement of beta and gamma power, reflecting heightened arousal and cognitive stress.
Physiologically, both microgravity and hypergravity are perceived as significant stressors, leading to increased sympathetic arousal (shown by a V-shaped response in skin conductance, lowest at 1g and rising sharply at extremes). Heart rate variability also increases in microgravity, consistent with initial cardiovascular deconditioning.
Also Read:
- Optimizing Sleep Analysis for Wearable Devices
- Predicting High-Performing AI for Wearables Without Extensive Training
Implications for Space Exploration
This dual computational framework, integrated with LLM simulations, offers a powerful new approach to understanding and predicting human performance in altered gravity environments. The findings have direct implications for astronaut training and aerospace safety. The lightweight EEG model could be deployed on wearable devices for real-time monitoring of neural patterns during high-G training or VR simulations, detecting early signs of cognitive instability.
By building an individual ‘gravity-adaptation fingerprint’ for each crewmember, training can be personalized to target weak points, accelerate adaptation, and enhance resilience. This could help predict individual susceptibility to gravitational stress, including risks of G-induced loss of consciousness (G-LOC) or disorientation, ultimately leading to safer and more effective space missions.
