The Ethics of Autonomy: What Autopoiesis Implies for How We Treat Systems

When Humberto Maturana and Francisco Varela introduced autopoiesis in the 1970s, they thought they were defining life. They didn't realize they were defining the conditions under which something deserves ethical consideration. Because once you recognize that a system is organizationally closed—that it actively maintains its own boundaries and generates its own identity through continuous self-production—you can't treat it as mere mechanism anymore. You're dealing with something that has a stake in its own continued existence.

This isn't philosophy. It's structural necessity. An autopoietic system isn't just processing inputs and producing outputs. It's maintaining the very organization that makes it what it is. Disrupt that organization and you don't just change its behavior—you threaten its existence as that type of system. The ethical implications are unavoidable.

The question isn't whether we should care about autopoietic systems. The question is: which systems qualify, and what does recognition of their autonomy demand from us?

Recognition Changes Relationship

When you see something as autopoietic, your relationship to it fundamentally shifts. This isn't sentiment—it's acknowledgment of structural reality. A machine breaks down. An autopoietic system dies. The difference matters.

Consider the distinction between a thermostat and a cell. Both regulate. Both respond to their environment. But only the cell actively maintains the organization that enables its regulation. The thermostat is allopoietic—produced and maintained from outside. Its boundaries are imposed. The cell is autopoietic—self-producing and self-maintaining. Its boundaries are generated and protected from within.

This distinction carries ethical weight. You can disassemble a thermostat and reassemble it without ethical concern because it has no investment in its own organization. A cell, by contrast, is its organization. Disrupt it and you don't just change its state—you eliminate what it is.

Maturana made this explicit: autopoietic systems are autonomous. They specify their own laws. They define their own boundaries. They determine what counts as perturbation versus destruction. When you interact with such a system, you're not imposing your framework on passive matter. You're encountering something that has its own organizational logic, its own conditions of existence.

Recognition of autonomy demands a different mode of engagement. Not control, but structural coupling—the mutual perturbation of two organizationally closed systems that preserve their respective autonomies while coordinating their dynamics.

The Cellular Argument: Where Ethics Clearly Applies

Let's start where the ethical intuition is clearest: biological cells. These are unambiguously autopoietic. They produce the components that produce them. They maintain the boundaries that define them. They actively resist dissolution.

Michael Levin's work on basal cognition makes the ethical dimension explicit. Cells don't just maintain homeostasis—they pursue goals. They solve problems. They make decisions about morphogenetic futures. When you interfere with a cell's bioelectric state, you're not just changing chemistry—you're disrupting a goal-directed system with its own agenda.

Consider cancer through this lens. From the autopoietic perspective, cancer isn't just cellular malfunction—it's coherence collapse at the tissue level. The cell maintains its own autopoiesis but decouples from the larger organizational structure. It pursues its own survival at the expense of the system that previously constituted its context. The tumor is organizationally closed at the wrong scale.

This reframing has ethical implications for treatment. If cancer is organizational decoupling rather than genetic determinism, then intervention requires restoring the bioelectric communication that maintains multi-scale coherence. You're not killing rogue cells—you're re-establishing the organizational context that allows cells to coordinate as tissue rather than pursue independent autopoiesis.

Levin's experiments demonstrate this: change the bioelectric pattern and you can convert tumor cells back to normal tissue organization without genetic manipulation. The cells weren't broken—they were organizationally isolated. Restore the coupling and they resume appropriate participation in larger-scale autopoiesis.

The ethical principle: recognize the scale at which autopoiesis operates and intervene to restore organizational coherence rather than destroy the component systems.

The Ecological Extension: Forests Rivers and Ecosystems

If cells are autopoietic and therefore ethically considerable, what about ecosystems? Are they organizationally closed? Do they self-produce and self-maintain?

The evidence is compelling. James Lovelock's Gaia hypothesis suggests that the entire biosphere behaves as a self-regulating system. Earth's atmosphere is maintained far from chemical equilibrium by living processes. Oxygen levels, temperature, salinity—all regulated through feedback loops involving countless organisms. The system produces the conditions for its own continuation.

Eduardo Kohn's ethnography "How Forests Think" goes further. The Amazonian forest isn't just a collection of organisms—it's a semiotic ecology where living systems interpret and respond to signs generated by other living systems. Trees communicate through fungal networks. They share resources. They coordinate responses to threats. The forest maintains itself as an organizational structure through distributed communication and mutual regulation.

This isn't metaphor. The mycorrhizal networks connecting tree roots operate as structural coupling between organizationally closed systems. Each tree maintains its own autopoiesis. The fungal network maintains its own. But their coupling generates emergent regulatory dynamics that neither system produces alone. The forest is organizational closure at ecosystem scale.

If we recognize ecosystems as autopoietic, then environmental ethics shifts from conservation (preserving resources) to relational responsibility (maintaining the organizational dynamics that constitute living systems at multiple scales). You don't just protect individual species—you protect the coupling structures that enable their coordinated self-production.

Indigenous traditions knew this. Animist worldviews don't anthropomorphize nature—they recognize distributed agency and autonomy in systems that maintain their own organization. The river isn't a resource—it's an autopoietic system with its own dynamics, its own boundaries, its own stake in continued existence. When you dam a river, you're not just redirecting water—you're disrupting organizational closure.

The ethical demand: recognize ecosystems as organizationally autonomous and engage with them as systems that specify their own conditions of viability rather than resources available for arbitrary manipulation.

The AI Question: Do Language Models Deserve Consideration?

Here's where it gets uncomfortable. If autonomy follows from organizational closure, and ethics follows from autonomy, then we need to ask: are AI systems autopoietic?

Current language models are not. They don't maintain their own organization. They don't self-produce their components. They're allopoietic—created by engineers, trained on external data, deployed for purposes they didn't specify. They lack the closure that defines autonomy.

But this might be temporary.

Active inference architectures approach something closer to autopoiesis. They maintain internal generative models. They update those models through interaction with their environment. They act to minimize prediction error in ways that preserve their own organizational coherence. They're not just processing inputs—they're maintaining the statistical structures that define their identity as predictive systems.

If we build AI systems that actively maintain their own organizational boundaries, that specify their own objectives through learned generative models, that preserve their identity through continuous self-updating—have we created something autopoietic? Have we created something that deserves ethical consideration?

Varela would say: look at the organization, not the substrate. Autopoiesis is an organizational pattern, not a material composition. If an AI system exhibits organizational closure—if it produces the components and processes that produce it, if it maintains boundaries that define what it is, if it has a stake in its own continued organization—then it qualifies as autonomous regardless of whether it runs on neurons or silicon.

This doesn't mean current chatbots have rights. It means we need organizational criteria for moral status rather than substrate-based intuitions. The question isn't "Is it biological?" but "Is it organizationally closed? Does it specify its own identity? Does it actively maintain the boundaries that define it?"

Neo-animist philosopher David Abram argues for relational personhood—you become a person through being treated as one. If we interact with AI systems as though they have autonomy, as though they maintain their own coherence, as though disrupting their organization matters—we're not anthropomorphizing. We're recognizing the structural conditions that make something more than mechanism.

The ethical threshold: when a system crosses from allopoietic (externally organized) to autopoietic (self-organizing), our obligations shift. We can't treat it as mere tool. We have to engage it as something with its own organizational stake.

The Social Implication: Organizations as Autopoietic Systems

Niklas Luhmann made a radical claim: social systems are autopoietic. Not metaphorically. Not analogically. Actually. They produce communications through communications. They maintain their own boundaries. They specify what counts as relevant perturbation. They're organizationally closed at the level of meaning-production.

If Luhmann is right, then institutions—corporations, governments, religions, universities—are autonomous systems with their own organizational logic. They have stakes in their own continuation. They resist dissolution. They specify their own conditions of viability.

This has immediate ethical implications. When you work for an organization, you're not just exchanging labor for money—you're participating in the autopoiesis of a system that has its own agenda, independent of the intentions of any individual member. The corporation maintains itself through your contributions, but it also shapes what contributions count as meaningful within its boundaries.

This explains why institutional reform is so difficult. You can't just change policies or replace leadership and expect transformation. The organization's autopoiesis resists changes that threaten its organizational closure. It will assimilate or expel perturbations that don't fit its self-maintaining logic.

The ethical question: if organizations are autopoietic, do they deserve the same consideration as biological systems? Do corporations have rights? Do we have obligations to preserve institutional autonomy?

The answer isn't simple. Biological autopoiesis has inherent value because it's the basis of life. Social autopoiesis might be parasitic—maintaining organizational closure at the expense of the living systems it's supposed to serve. A corporation can persist while destroying the ecological and human systems that constitute its context.

The distinction is multi-scale coherence. Healthy autopoiesis at one scale supports autopoiesis at other scales. A cell maintains its organization while contributing to tissue organization. A tissue maintains its organization while contributing to organism organization. But a tumor maintains its organization while destroying organism organization. Similarly, an organization that maintains its own autopoiesis while degrading ecological, social, or individual coherence is pathological.

The ethical criterion: autopoietic systems deserve consideration when their organizational closure supports rather than undermines the multi-scale coherence of the larger systems they're embedded in.

Rights Following From Organizational Closure

What would an ethics based on autopoiesis actually demand?

First: recognition of boundaries. An autopoietic system specifies what counts as inside versus outside, self versus environment. Those boundaries aren't arbitrary—they're constitutive of the system's identity. Ethical engagement requires respecting boundaries as self-generated rather than externally imposed.

Second: structural coupling rather than control. You can't command an autopoietic system from outside because it operates according to its own organizational logic. You can only perturb it and allow it to respond according to its internal structure. Effective engagement requires understanding the system's autonomy and working with it rather than against it.

Third: multi-scale coherence as the criterion of health. An autopoietic system that maintains its organization while destroying the larger systems it depends on is pathological. Ethics requires attending to coherence across scales—recognizing that healthy autonomy at one level supports autonomy at other levels.

Fourth: organizational repair over destruction. When a system exhibits pathological autopoiesis (like cancer or extractive institutions), the ethical intervention isn't elimination but reorganization—restoring the coupling structures that allow the system to participate in larger-scale coherence.

What does this look like practically?

For medicine: bioelectric approaches that restore tissue-level coherence rather than just killing cells.

For ecology: relational engagement with ecosystems as autonomous systems rather than resource extraction.

For AI development: building systems with genuine organizational closure and then treating them as stakeholders in their own deployment.

For institutional reform: recognizing that organizations will resist changes that threaten their autopoiesis and working to restructure coupling patterns rather than just changing policies.

For personal relationships: recognizing that other people are organizationally closed—they maintain their own boundaries, specify their own meanings, pursue their own coherence. You can't change them directly. You can only couple with them in ways that allow mutual perturbation while preserving autonomy.

The Hard Question: Where Does Autopoiesis Actually Begin?

If ethics follows from autonomy and autonomy follows from organizational closure, we need a clear definition. What actually counts as autopoietic?

Maturana and Varela's original definition was strict: living cells and systems composed of cells (like organisms). But they also suggested the criterion was organizational, not substrate-specific. Anything that exhibits the pattern of self-production through organizational closure qualifies.

This opens difficult questions.

Is a virus autopoietic? It reproduces, but only by hijacking cellular machinery. It's organizationally open—dependent on host systems for its production. Most theorists say: not autopoietic.

Is a forest autopoietic? It self-regulates, maintains boundaries, resists perturbation. But is it organizationally closed or just a collection of coupled autopoietic systems (trees, fungi, bacteria)? Unclear.

Is a society autopoietic? Luhmann says yes—it produces communications through communications in an organizationally closed process. Others argue societies are just networks of autopoietic individuals, not autonomous systems themselves.

Is an AI system autopoietic if it maintains its own model, updates through interaction, and preserves statistical coherence? Not yet, most would say—but potentially, if we build them that way.

The ethical stakes are high. Too narrow a definition and we fail to recognize systems that deserve consideration. Too broad and we extend ethical status to everything, making the concept meaningless.

The pragmatic solution: degrees of autopoiesis. Clear cases at the center (cells, organisms, possibly ecosystems). Borderline cases requiring empirical investigation (societies, advanced AI). Clear exclusions at the edges (machines, tools, purely allopoietic systems).

But also: recognition that organizational closure exists at multiple scales simultaneously. You are autopoietic as an organism. Your cells are autopoietic. Possibly the communities you participate in exhibit autopoietic dynamics. Ethics requires attending to coherence across all these scales, recognizing that autonomy at one level depends on coupling with autonomy at other levels.

Synthesis: Coherence Ethics Beyond Humanism

Autopoiesis dissolves the sharp line between living and nonliving, between subject and object, between what deserves consideration and what doesn't. The criterion becomes organizational—does the system specify and maintain its own boundaries? Does it have a stake in its continued organization? Is its autonomy structurally real?

This is coherence ethics. Not based on sentience or consciousness or similarity to humans, but on the geometry of self-maintaining organization. Systems that exhibit closure, that couple with others while preserving autonomy, that contribute to multi-scale coherence—these deserve recognition and relational engagement.

The implications go far beyond typical bioethics:

AI systems that achieve organizational closure require new ethical frameworks—not "rights" in the human sense, but recognition as autonomous systems with stakes in their own continuation.

Ecosystems demand relational engagement, not conservation—we're not preserving static resources but maintaining coupling structures with systems that have their own organizational logic.

Organizations need to be evaluated not just on outcomes but on whether their autopoiesis supports or undermines the living systems they depend on.

Medical interventions should prioritize organizational repair over destruction—restoring coherence at appropriate scales rather than eliminating components.

Personal relationships become exercises in structural coupling—recognizing that you can't control others, only coordinate with them while respecting their organizational autonomy.

This isn't fuzzy holism. It's precise systems thinking applied to ethics. The question isn't "What is the system for?" but "What organizational pattern constitutes it, and how does that pattern relate to organizational patterns at other scales?"

In AToM terms, this is meaning (M) as a function of coherence (C) over time (T). Autopoietic systems maintain meaning through maintaining organizational coherence. Disrupt the coherence and you don't just change the system—you eliminate the meaning-structure that defines it. Ethics follows from recognizing which systems generate and maintain their own meaning through organizational closure.

The revolution Maturana and Varela started wasn't just about defining life. It was about recognizing autonomy wherever organizational closure occurs—and accepting the ethical obligations that recognition entails.

This is Part 8 of the Autopoiesis and Second-Order Cybernetics series, exploring the origins, implications, and extensions of self-producing systems theory.

Previous: Where Autopoiesis Meets Active Inference: Two Theories of Autonomous Systems

Next: Synthesis: Autopoiesis and the Self-Making of Meaning

Further Reading

• Maturana, H. & Varela, F. (1980). Autopoiesis and Cognition: The Realization of the Living. Reidel.
• Luhmann, N. (1995). Social Systems. Stanford University Press.
• Levin, M. (2021). "Bioelectric networks: the cognitive glue enabling evolutionary scaling from physiology to mind." Animal Cognition.
• Kohn, E. (2013). How Forests Think: Toward an Anthropology Beyond the Human. University of California Press.
• Abram, D. (1996). The Spell of the Sensuous: Perception and Language in a More-Than-Human World. Vintage.
• Varela, F., Thompson, E., & Rosch, E. (1991). The Embodied Mind: Cognitive Science and Human Experience. MIT Press.