| name | brain-digital-twins-execution-semantics-v4 |
| description | Brain digital twins execution semantics survey bridging computational neuroscience to neuromorphic systems - arXiv:2604.13574 (April 2026). Covers physically constrained executability taxonomy, execution regimes, hybrid-time correctness, and neuro-neuromorphic physical systems. Supersedes v3 with systems/runtime perspective. |
Brain Digital Twins: Execution Semantics and Neuro-Neuromorphic Systems
arXiv: 2604.13574
Date: April 15, 2026
Author: Alexandre Muzy
Categories: cs.SE, cs.CE, q-bio.NC, cs.NE
Core Thesis
Brain digital twins aim to provide faithful, individualized computational representations of brains as dynamical systems. Current approaches are fragmented across data pipelines, model classes, temporal scales, and computing platforms, preventing preservation of execution semantics across end-to-end workflows.
The paper introduces physically constrained executability as a unifying perspective — comparing approaches at the level of execution rather than model form.
Key Concepts
Physically Constrained Executability
A framework comparing brain digital twin approaches based on:
- Execution state persistence — whether state survives across executions
- Permitted events — simulation, measurement, actuation
- Temporal/causal coupling — how strongly execution is coupled to neurobiological dynamics
- Physical constraints — shared physical constraints between biological and computational dynamics
Execution Regimes Taxonomy
The paper proposes a taxonomy ranging from:
| Regime | Description | Coupling |
|---|
| Isolated offline models | Traditional simulation, no live data | None |
| Coordinated co-simulation | Multiple models running with coordination | Weak |
| Continuously executing digital twins | Sustained by online data assimilation | Strong |
| Neuro-neuromorphic physical systems | Biological and computational dynamics co-executed under shared physical constraints | Physical |
Why Accuracy Alone Is Insufficient
Accuracy metrics ignore:
- Semantic interoperability — how systems communicate meaningfully
- Hybrid-time correctness — handling continuous and discrete time together
- Evaluation protocols — standardized testing across regimes
- Scalable reproducible workflows — reproducibility at scale
- Safe closed-loop validation — safety in live brain-computer interfaces
Research Agenda
- Semantic Interoperability — standardizing how brain models communicate
- Hybrid-Time Correctness — formal methods for mixed continuous/discrete execution
- Evaluation Protocols — standardized benchmarks across execution regimes
- Scalable Reproducible Workflows — tools for reproducible brain modeling
- Safe Closed-Loop Validation — safety frameworks for live brain interfaces
Applications
- Clinical intervention prediction
- Mechanistic understanding of brain dynamics
- Personalized medicine
- Brain-computer interfaces
- Neuromorphic computing systems
Related Work
- Brain-DiT foundation models (arXiv:2604.18469)
- Alzheimer's disease progression models (arXiv:2604.18470)
- Higher-order brain interactions (arXiv:2604.17713)
- EEG foundation model adaptation (arXiv:2604.16926)
Implementation Notes
- Focus on execution semantics, not model architecture
- Consider physical constraints in system design
- Evaluate across multiple coupling regimes
- Prioritize semantic interoperability for integration
Citations
@article{muzy2026brain,
title={From Brain Models to Executable Digital Twins: Execution Semantics and Neuro-Neuromorphic Systems},
author={Muzy, Alexandre},
journal={arXiv preprint arXiv:2604.13574},
year={2026}
}
