Sensitive living systems: Balancing disorder and complexity in holobionts for communication across ecosystems

¹ Sorbonne Université, INSERM, Immunoregulation-Immunopathology-Immunotherapy (I3); UMRS959; Paris - France ; CNRS-France
² IMT Atlantique, Lab-STICC, CNRS UMR 6285, Technopôle Brest-Iroise, CS 83818, Brest - France

^* Corresponding author : This email address is being protected from spambots. You need JavaScript enabled to view it.

Published online: 29 April 2026

Abstract

How do living systems sustain communication, identity, and resilience while remaining exposed to disorder and environmental perturbations? How do poly-genomic organisms integrate evolutionary history (including non-genetic transmission), development, and ecological context to generate memory and tolerance? Addressing these questions requires a paradigm shift, from reductionist descriptions toward living systems conceived as sensitive subjects, historically embedded organizations. We consider the holobiont—the host organism together with its microbiota—and the immune system as a model of multiscale living co-organization shaped by onto-phylogenesis. We show how somatic diversification, degeneracy, and stochastic interactions enable biological networks to regulate integrity and coherence, transforming variability into adaptive organization rather than noise. Across intertwined temporalities—developmental time, critical decision windows, and long evolutionary time—biological sensors detect and interpret perturbations, supporting emergent properties such as dominant tolerance, distributed memory, and resilience. We introduce the Generic Sensor-Actuator (GenSA) framework as a scale-independent, subject-centered abstraction: living systems are modeled as networks of multiscale holons that sense, integrate, memorize, and respond to distortions through energy- or affinity-based interactions, providing a common language of communication from molecules to ecosystems. Adaptation relies on probabilistic evaluation of perceived distortions through random distortion functions/tests across scales, making “memory” and “tolerance” history-dependent dynamical networks.