Autopoiesis: Self-Creating Systems and the Organization of Life

Defining Life Through Organization

Science values standardized definitions—shared criteria enable comparison, classification, and consensus. Yet life has resisted universal definition. Consider the familiar attempts: metabolism? Fire metabolizes fuel yet isn’t alive. Reproduction? A mule is sterile but unquestionably living. Growth? Crystals grow through accretion, not alive. Response to stimuli? A thermostat responds to temperature changes, not alive. Information processing? Computers process information far faster than neurons, not alive. Evolution through natural selection? Viruses evolve but their status as living remains contested—they carry nucleic acids and mutate across generations, yet possess no metabolism and require host cellular machinery to replicate.

The difficulty reveals something fundamental: life cannot be captured through listing properties, because any property list produces borderline cases that confound classification. This isn’t mere philosophical inconvenience. Astrobiology depends on knowing what biosignatures to seek. Origins of life research requires knowing what organization constitutes the threshold between chemical systems and living ones. The search for a definition persists because humans evolved to distinguish living from nonliving—to run from lions and sit on rocks—giving this categorization survival weight even when scientific boundaries remain unclear.

My approach with Francisco Varela shifted the question entirely. We asked not “what properties must life possess?” but rather “what organization defines living systems?” The distinction is critical. Properties describe components or behaviors—carbon chemistry, DNA replication, energy consumption. Organization describes relationships among components—how processes relate to produce and maintain the system as a distinct unity. A car possesses organization: components arranged to enable transportation. What makes a system living is not its materials (carbon appears in rocks), nor its energy flows (rivers flow without being alive), but a specific type of organization—one that produces itself.

Autopoietic Organization

Autopoiesis—from Greek auto (self) and poiesis (creation, production)—names this self-producing organization. An autopoietic system is a network of processes that, through interactions and transformations, continuously regenerates the components that constitute the network, and realizes the network as a distinct unity in space by constituting boundaries. The definition is precise and operational. Three criteria must be satisfied: first, the processes regenerate the components making the network through their own interactions. Second, these processes constitute the system as a unity distinct from its environment by creating boundaries. Third, this organization maintains itself despite continuous component turnover.

Consider the minimal living system: a bacterial cell. The lipid membrane forms a boundary distinguishing inside from outside. Metabolic pathways—enzymatic reactions, protein synthesis, RNA catalysis—occur within this boundary. These metabolic processes generate the very components that enable metabolism: proteins catalyze reactions producing lipids, nucleic acids encode proteins, lipids form membranes containing the metabolism. This is circular causality, organizational closure. The cell produces itself continuously. A bacterium’s lifespan before division is roughly twenty minutes, yet during this interval proteins degrade and are replaced, ATP molecules are consumed and regenerated, membrane lipids are synthesized and exchanged. The molecular components change completely, yet the cell persists as the same unity. Its identity resides not in particular molecules but in the organizational pattern—the network of production processes that regenerate one another.

Contrast this with an allopoietic system—one that produces something other than itself. A car factory produces cars, not more factory. The assembly line doesn’t regenerate its own components through its operation. Components wear and must be replaced from outside, but the factory organization doesn’t produce them. Similarly, a river flows as a stable pattern—water molecules continuously replace one another while the riverbed shape persists—but the river doesn’t produce its own banks. It’s a pattern in flowing matter, not a self-producing organization. The cell, by contrast, produces its own membrane. This self-production is autopoiesis.

The tardigrade demonstrates organizational invariance despite structural extremes. During cryptobiosis, metabolism drops to 0.01% of normal levels as the organism contracts into a tun state, losing 98% of body water. It survives decades in suspended animation. Yet upon rehydration, the autopoietic organization resumes—metabolic processes restart, components regenerate, the boundary reforms. The organization wasn’t destroyed, only slowed to near-cessation. Furthermore, tardigrades exhibit eutely: adult individuals possess a fixed cell number established during development. Growth occurs through cell enlargement, not division. Every cell in the adult’s claws and segments is a cell it possessed since larval stages. Identity persists not through replacing cells, but through maintaining organizational relationships among them. Even when structural turnover halts almost completely, the autopoietic pattern—the potential for self-production—remains the defining characteristic.

Autonomous Yet Coupled

Autopoietic systems are autonomous. This autonomy is specific and paradoxical: such systems are operationally closed yet materially open. Operational closure means the organization is self-referential—processes produce processes, defining the system’s own boundaries without external specification. Material openness means components are exchanged with the environment—energy and matter flow through the system. This isn’t contradiction but complementarity. The organization (pattern of relationships) is closed and self-producing. The structure (actual components) is open and exchanging.

Structural coupling names the ongoing interaction between autonomous system and environment. The environment perturbs the system. The system’s structure determines its response to perturbation—the environment triggers but doesn’t instruct. Consider bacteria in a glucose solution. Glucose presence perturbs the cell’s membrane receptors. This triggers metabolic adjustments—gene expression changes, enzyme production shifts—but the specific response depends entirely on the bacterium’s internal structure, its current metabolic state, its genetic organization. Different bacteria in the same glucose solution respond differently. Some ferment glucose, others respire it aerobically. The environment selected them (those unable to metabolize glucose didn’t survive), but it didn’t determine their response mechanisms. Response arises from internal structure, not external instruction. This is autonomy: the system determines its own states according to its own organization.

Through recurrent interactions, structural coupling produces co-ontogeny—mutual structural drift. The organism and its niche co-evolve. The organism’s responses change its environment, the changed environment triggers different responses, and both drift together through ongoing coupling. There is no predetermined fitness landscape, no external criterion of optimization. Only continuous mutual perturbation and structural modification preserving autopoiesis. Natural drift, not natural selection (a misleading phrase implying external agency selecting). The coupling maintains organization while allowing structural change—adaptations emerge from the history of coupling, not from achieving predefined functions.

Ecosystems demonstrate emergent autopoietic patterns at collective scales. Forests cycle nutrients through growth, decay, predation—each process regenerating components enabling other processes. No central manager orchestrates this. Order arises from distributed interactions following simple local rules. Self-sustaining systems close loops so effectively that each process generates the next. This is not perpetual motion violating thermodynamics—energy flows through from sunlight—but rather perpetual organization maintained through open recursion. Historical inventors sought isolated perpetual motion machines, attempting mechanisms sealed from external influence. They failed by design. Nature’s self-sustaining systems never operate in isolation. A heart beats because oxygen flows through it. A planet orbits because it moves within a gravitational field. Isolation kills circulation. Autopoietic systems sustain themselves precisely through participation with surrounding fields, exchanging matter and energy while maintaining organizational closure.

Cognition as Living

Everything said is said by an observer. This isn’t relativism but biological epistemology. Observers are human beings, languaging beings who bring forth worlds through their structure and history of coupling. We cannot step outside our cognition to describe reality objectively, because description is itself a cognitive act. Knowledge is observer-dependent—not arbitrary, but constituted through the biological organization of the knower.

Living systems are cognitive systems. Cognition is not something certain organisms possess—it is what living is. Cognition equals effective action, action enabling the continued autopoiesis of the system. A bacterium swimming up a glucose gradient performs cognition: it detects concentration differences through membrane receptors, adjusts flagellar rotation through signal transduction cascades, and maintains metabolism through nutrient acquisition. No brain required. No representations required. Cognition is sensorimotor coordination preserving organization. The bacterium doesn’t process “information” about glucose—it couples structurally with its chemical environment in ways its organization determines. Its action is cognitive because it’s effective for maintaining autopoiesis.

Nervous systems evolved to coordinate sensory-motor coupling in multicellular organisms, where different tissues must integrate activity. Brains elaborate this coordination, eventually enabling language—not information transfer but coordination of coordinations of behavior in a consensual domain. Human consciousness is a further elaboration, bringing forth distinctions and coordinating them through recursive languaging. But the fundamental cognitive process remains: bringing forth a world through effective action grounded in autopoietic organization. There is no external world the organism represents. There is only structural coupling, where organism and environment co-specify one another through recurrent interaction.

This understanding transforms how we approach living systems. We stop searching for external programs, information codes, or controlling blueprints. We recognize that organization is the locus of identity, not components. We see autonomy as structural determinism, not freedom from material constraint. We understand cognition as living itself, not a property added to certain complex organisms. And we ground our epistemology in biology: everything said is said by an observer, and the observer is constituted by the very autopoietic processes being observed. We participate in what we study. This is not circularity defeating explanation—it is the biological reality of cognition recognizing itself.