Resilience and adaptability in self-evidencing systems
Mahault Albarracin, Dalton A R Sakthivadivel · 2025
Read Resilience and adaptability in self-evidencing systems by Mahault Albarracin, Dalton A R Sakthivadivel (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-06897.txt.
Introduction
Mahault Albarracin and Dalton Sakthivadivel offer a theory paper from twenty twenty-five on what it means for a living or artificial system to be resilient. They work within the free energy principle, a framework that treats staying alive as keeping a workable model of oneself and the world. Earlier work used that framework to model resilience from the outside. The authors argue instead that resilience sits at the center of the framework itself, because keeping an identity requires constant adjustment.
On their view, a system holds together by reshaping its self-model as conditions change, and that reshaping takes energy. They sketch a general framework for thinking this way and give simple models to show it in action. The paper will interest readers who study adaptive systems, self-organization, and theories of life.
Catalog record
| CREATOR | Mahault Albarracin, Dalton A R Sakthivadivel |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 31 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-06897
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Resilience and adaptability in self-evidencing systems Mahault Albarracin, Dalton A R Sakthivadivel Abstract In this paper we will articulate a view of resilience under the free energy principle and vice versa . The free energy principle is about existence and identity, and resilience is the condition under which things exist at all. In previous work this has been investigated as modelling resilience using the free energy principle. We will extend that work by making the case that self-organisation under the free energy principle is about resilience, in the sense that identity is a constant process of self-reconfiguration, implying the existence of a self-model and the energy to reconfigure that self-model—and hence, the resilience of a maintained identity under changes. A general framework for thinking about resilience in this context will be sketched out and some models will be provided using that framework. 1 Introduction This paper is mainly concerned with the notions of identity and resilience which are at stake when modelling adaptive systems. Self-organisation is implicitly defined in terms of identity—that is, the self. However, it is also apparent that self-organisation is never static, and the sense of identity implicit here must somehow be flexible or else be capable of accommodating changes in the environment. In dissipative adaptation, for example, a system self-organises by restructuring itself to absorb and dissipate energy from an external source in the environment [ 9 , 21 , 24 , 40 ] . This echoes enactive and autopoietic theories: living systems produce and maintain themselves [ 28 ] , and for an organism to maintain a model of itself and its environment, it must produce what is required to preserve its generative model [ 29 ] . Maturana–Varela’s concept of autopoiesis describes how a system continuously regenerates its components but also keeps its identity intact—again encountering the tension we highlighted. This paper aims to investigate what an appropriate description of this flexibility might be and how can it can be modelled, and in particular, what resilience has to do with a changing identity in dynamic environments. We will develop a first idea of a formalism to describe resilience and some simple models for how resilience features in adaptive systems. Adaptation is often achieved by, or can be modelled as, an inference process [ 15 , 37 , 39 , 47 , 51 ] . An interesting example to study is the human brain. Modern theoretical neuroscience asserts that the brain maintains an internal self-model which is continuously refined as part of its predictive engagement with the world. In predictive processing, an agent must model the causes of its sensory inputs to respond to perturbations from the environment [ 12 ] . Since the agent’s own body, traits, and actions are implicated in those causes, the brain encodes a model of ‘self’ to better predict incoming sensations [ 20 ] . In particular, Hohwy–Michael suggest that people “perceive and maintain their selfhood by an internal model of hierarchical endogenous (hidden) causes”, wherein higher-level causes like goals and traits generate predictions that shape perception and action. It follows that identity is embodied in the brain’s generative model and is continuously updated by maximising model evidence and minimising prediction error, a process one might call self-evidencing . As discussed, maintaining an identity requires not just change but also a degree of stability of core features, offering a conundrum. To self-evidence, an agent learns and constantly adjusts a generative model of the world and of itself, using sensory evidence to correct discrepancies. The brain expends informational and energetic resources to keep its self-model aligned with reality—revising beliefs or deploying actions when mismatches (prediction errors) arise. This continual updating is the computational essence of self-reconfiguration. The self-model is never static, because new experiences demand incorporation. One way to resolve this tension is to view an agent as balancing accuracy (updating to fit data) with precision or confidence in prior beliefs, as is done in active inference [ 14 , 33 ] . Some self-model predictions— e.g. deeply held traits or values—are high-precision priors that resist change, providing continuity of identity. This balance is evident in psychology: people seek to preserve a coherent self-concept in the face of discrepant information (the motive for self-consistency or self-verification) even as they update other aspects of themselves over time [ 34 ] . For example, individuals continuously gather self-relevant information to build an inner ‘self-theory’, updating beliefs about themselves with feedback from social and personal experiences. Over time, as a more stable self-concept is consolidated, people begin to give greater weight to maintaining existing self-beliefs, “at the expense of continuous updating”. Altogether this implies that the “self” is not a static essence but a working model, continually regenerated by the very organism it defines. To characterise the malleability of certain aspects of identity and the rigidity of certain other ones, we will frame it in terms of resilience . An identity (the organism’s self-model) is sustained by ongoing self-production and error-correction—the same processes that characterise resilience. Resilience generally refers to the capacity of a system to withstand shocks or disturbances without losing its core functions or identity [ 18 , 33 , 42 ] . The term resilience is polysemous: it encompasses at least three complementary aspects [ 2 , 33 ] . It denotes (i) inertia—resistance to change when disturbed, (ii) elasticity—the ability to rebound or restore a prior state after a deviation, and (iii) plasticity—the capacity to flexibly reorganize or find new stable states when conditions change. All three aspects describe processes rather than static traits. Resilience is not simple a fixed quantity one has a certain amount of; it is something one does, dynamically, over time. The link between identity processes and resilience becomes apparent when we consider what is maintaining or recovering in each case. In the face of life’s perturbations—be it trauma, loss, illness, or rapid change—it is often one’s sense of self and purpose that is challenged. Awareness and cognitive flexibility distinguish the more resilient phenotypes, allowing agents to anticipate and integrate surprises without being overwhelmed. A resilient individual manages to hold onto an integrated self-concept (core values, sense of meaning, self-esteem) or to reconstruct a new one, such that they continue to function and grow. This points to multiple scales of the self, akin to Russian dolls with varying degrees of malleability—or rates of change. Figure 1: The Matryoshka doll exhibits a nested sense of identity. As time goes on, defects and mutations in the outer dolls change their phenotypes. The innermost doll, however, remains intact and constant. Identity has been linked in different ways to resilience. the concepts have porosity in the literature and a degree of metaphorical artistic license. Empirical studies show clear connections between certain qualities of identity and resilience outcomes. For instance, people who have more complex identities— i.e. who see themselves in terms of multiple distinct roles or attributes—tend to be more resilient to stress and negative events [ 4 , 49 ] . If one facet of the self is threatened, a person with many facets can draw strength and continue to find self-worth in other aspects, preventing a total collapse of self-regard. This flexibility in self-concept, a form of