Six-fold Integrated Criteria for Non-equilibrium Living Manifolds

The essential demarcation between living and non-living systems remains one of the central challenges in natural science. Traditional definitions rely on isolated lists of features such as metabolism, reproduction, and responsiveness, which fail to cover borderline cases and provide no guidance for artificial synthesis or extraterrestrial detection. This paper proposes the LINC theory (Living-system Integrated Nonequilibrium Criteria), arguing that the criterion for identifying a living system lies not in the presence or absence of any single feature, but in the simultaneous satisfaction of six constraints. These six constraints are: critical emergence of system complexity, dynamical positioning at the edge of chaos, non-equilibrium dissipative construction, topological boundary closure, finite-error information iteration, and cross-hierarchical fractal coupling. The first three constitute dynamic properties, while the latter three constitute structural properties; the two categories are mutually prerequisite and form an inseparable constraint bundle. A material organization satisfying all six constraints is defined as a non-equilibrium living manifold; the absence of any single constraint reduces the system to an ordinary dissipative structure or random turbulence. This theory provides operational negative criteria for the design of synthetic biology and the detection of extraterrestrial life. Keywords: definition of life; integrated information; edge of chaos; non-equilibrium thermodynamics; constraint satisfaction; synthetic biology 1. Introduction Attempts to define life have permeated the history of natural science. Early vitalism attributed life to mysterious vital forces, while reductionism disassembled life into mechanical combinations of molecular machines. Modern definitions typically enumerate several key features, such as metabolism, self-replication, evolutionary adaptation, and stimulus response. However, such feature-list-based definitions face three systematic difficulties. First, the problem of feature overlap. Fire consumes fuel and releases energy, satisfying the description of energy transformation, yet no one regards it as alive. Computer viruses can self-replicate and propagate, yet they are not classified as living organisms. This indicates that the satisfaction of a single feature, or the stacking of a few features, is insufficient to constitute adequate criteria. Second, the problem of boundary fuzziness. Viruses occupy a gray zone between life and non-life, lacking autonomous metabolism yet possessing information transmission and evolutionary capabilities. Prions are merely misfolded proteins, yet they can induce conformational transitions in homologous proteins. A permissive definition introduces excessive ambiguity, whereas a stringent definition excludes all marginal cases. Third, the lack of operationality. Extraterrestrial life detection requires explicit signal criteria; synthetic biology requires design endpoints and success standards; artificial life research requires criteria for determining whether a system has crossed the threshold from non-life to life. The absence of a formalized decision framework leaves these endeavors without theoretical anchors. In recent years, several independent theories have approached the formal definition of life from different angles. Kauffman and Roli (2024) proposed that life is a nonequilibrium self-replicating chemical system achieving “spatial closure, constraint closure, and catalytic closure.” Pross (2004–2023) established the framework of dynamic kinetic stability, arguing that life is a far-from-equilibrium dynamic state maintained through continuous energy supply. Langton (1990) pointed out that the optimal computational condition for living systems lies at the phase transition boundary between order and chaos. Navarro-Quiroz et al. (2026) reviewed cross-scale fractals and modular scaling laws in biological systems, proposing a recursive physicoinformational grammar from molecular networks to ecosystems. Eigen’s (1971) error catastrophe theory sets a theoretical upper limit on replication fidelity. The aforementioned works each capture one facet of the organizational principles of life, yet no framework has integrated these elements into a single computable conjunctive criterion with quantitative thresholds. This paper aims to fill this gap: systematizing and integrating existing elements, connecting the six constraints via logical conjunction to form the first computable negative criterion bundle. 2. System Definition and Basic Conventions To establish a unified language for subsequent discussion, the following basic concepts are defined. A system is defined as a directed network composed of material nodes and interaction edges, where nodes represent molecules or molecular assemblies, and edges represent chemical reactions or regulatory relationships. The system state is described by a vector of concentrations or conformational parameters of all nodes; the state space is the set of all possible states. Integrated information (Φ) serves as a metric for the density of internal causal interactions within the system, calculated based on mutual information between subsets of system states. The value domain of Φ is the non-negative real numbers; higher values indicate that the overall behavior of the system is increasingly difficult to decompose into a linear combination of subsystem behaviors. It should be particularly noted that the calculation of Φ requires explicit specification of system granularity and temporal scale; different levels correspond to different Φ values. The entropy S of the system Is divided into internal entropy and boundary-exchange entropy: internal entropy measures the degree of disorder and distribution uniformity of molecules within the system; boundary-exchange entropy measures the entropy flow resulting from material and energy exchange between the system and its environment. The chemical potential gradient μ is defined as the difference in chemical potential between the inside and outside of the system boundary. This gradient may be an ion concentration difference, charge difference, or any difference in generalized chemical potential. 3. Elaboration of the Six Constraints 3.1 First Constraint: Critical Emergence of System Complexity 3.2 This constraint requires that the integrated information Φ of the system be substantially greater than a critical threshold Φ_c. When the system has few nodes or low feedback-loop density, the system behavior can be approximated by the independent superposition of individual node behaviors. In this case, the macroscopic state of the system is a simple statistical result of microscopic states, with no novel causal structure above the molecular level. However, when the number of nodes, connection density, and nonlinear feedback intensity jointly cross a certain critical point, the system enters an emergent state. In this state, the macroscopic state of the system acquires downward causal power—that is, the macroscopic whole exerts constraints and guidance on local molecular behavior. For example, the global transcriptional state in a gene expression network can restrict the activity window of individual genes; the overall progression of the cell cycle can determine the synthesis timing of specific enzyme molecules. The existence of downward causal force renders system behavior inexplicable or unpredictable by linear superposition of microscopic chemical reactions. Understanding the macroscopic behavior of the system requires the introduction of a global state function that takes the state of the entire system, rather than local variables, as its independent variable. The necessity of this constraint lies in its exclusion of simple linear reaction chains and low-complexity oscillatory systems. A system failing to satisfy this constraint, regardless of its other conditions, lacks true holistic properties and can only be classified as an ordinary chemical system. 3.3 Second Constraint: Dynamical Positioning at the Edge of Chaos 3.4 Nonlinear dynamical systems can be classified into three ideal phases according to the nature of their phase-space trajectories. The first is periodic or quasi-periodic motion, where system orbits converge to a finite set of attractors and are insensitive to initial conditions, typified by pendulums or single-molecule oscillatory reactions. Such systems are highly ordered but lack the capacity to adapt to novel environments. The second is fully chaotic motion, where system orbits traverse dense regions of phase space with exponential sensitivity to initial conditions, typified by turbulence or threebody gravitational systems. Such systems possess rich potential for variation yet cannot preserve any information over the long term. The third is complex edge motion, where system orbits repeatedly traverse a finite region without overlapping, being neither fully predictable nor fully random. Living systems are neither periodically rigid nor chaotically collapsed; they are precisely locked onto the fractal boundary between order and disorder—the edge of chaos. The essential feature of this boundary is that the system simultaneously possesses two seemingly contradictory yet complementary attributes: long-range spatiotemporal correlation, meaning that a state change at one location within the system can influence distant or future system behavior without rapid decay due to local perturbations, thereby providing the dynamical foundation for the stable transmission of genetic and epigenetic information; and local high sensitivity, meaning that small perturbations in certain specific directions can be exponentially amplified by the system’s nonlinear mechanisms into macroscopic state transitions, thereby providing the dynam