VL:
TITLE:
‘Toward a Substrate-Independent Ontology of Life: Definitions, Consequences, and Edge-Case Applications in Synthetic and Extraterrestrial Contexts’
I. PROJECT OBJECTIVE
To systematically investigate the function, implications, and formal structure of definitions of life, with special attention to edge cases such as artificially instantiated consciousness, synthetic autopoiesis, and extraterrestrial intelligences that may not have evolved.
The ultimate aim is to produce a comprehensive, interdisciplinary report that identifies:
Theoretical commitments embedded in existing definitions of life
The epistemic and ethical consequences of such definitions
Candidate criteria for a non-parochial, substrate-independent conception
A framework for evaluating life-like or proto-life systems emerging via artificial means
II. DOMAIN MODULES
A. Meta-Theory of Definitions
Objective: Understand what definitions do in scientific and philosophical practice; distinguish descriptive, stipulative, operational, and normative definitions.
Tasks:
Compile work on definitional theory (Carnap, Wittgenstein, Lakoff, Haslanger)
Clarify roles of definitions in theory-building and institutional boundary-drawing
Analyse historical shifts in ‘life’ definitions (Aristotle, Descartes, Schrödinger, NASA)
Deliverables:
Taxonomy of definitional roles in science
History-sensitive map of ‘life’ definitions over time
B. Core Theories of Life
Objective: Map dominant theories of life and their metaphysical commitments.
Subdomains:
Thermodynamic (Schrödinger, Prigogine)
Autopoietic (Maturana, Varela)
Evolutionary/Selectionist (NASA, Joyce)
Information-Theoretic (Adami, Walker)
Algorithmic/Computational (Langton, Bedau)
Tasks:
Identify necessary/sufficient conditions per theory
Clarify what is gained/lost in each framework
Evaluate each on substrate-neutrality and generalisability
Deliverables:
Comparative matrix of definitions by core features
Assessment of edge-case applicability
C. Origin-Agnostic vs Origin-Constrained Models
Objective: Analyse how definitions of life encode assumptions about emergence.
Tasks:
Review work on abiogenesis, spontaneous order, and engineered systems
Explore causal essentialism vs organisational essentialism
Assess whether intentional design (e.g. human-made AI) violates life-criteria
Deliverables:
Philosophical analysis of ‘origin dependence’
Implications for defining life ex nihilo
D. Life and Consciousness: Ontological Dependencies
Objective: Examine the relationship between life and consciousness.
Tasks:
Review enactivist and embodied cognition views (Thompson, Di Paolo)
Survey artificial consciousness research
Determine whether consciousness presupposes life (in any sense)
Deliverables:
Ontological map: life → autonomy → cognition → consciousness?
Positions on possible artificial consciousness without life
E. Extraterrestrial and Synthetic Contexts
Objective: Identify how definitions of life function in future-facing and non-terrestrial settings.
Tasks:
Review astrobiological criteria and biosignatures (Walker, Schulze-Makuch)
Survey artificial life and machine evolution platforms (Avida, Tierra)
Assess how definitions regulate what counts as ‘discovery’ or ‘contact’
Deliverables:
Framework for evaluating life in absence of Earthlike evolution
Analysis of intelligence without life vs life without intelligence
F. Normative and Legal Implications
Objective: Analyse how definitions of life encode and justify normative boundaries.
Tasks:
Survey bioethics literature on beginning and end-of-life debates
Map connections between ‘life’ and moral status
Investigate AI rights, synthetic personhood, and legal definitions of life
Deliverables:
Distinction between biological and moral thresholds
Analysis of legal definitions and their philosophical underpinnings
G. Epistemic Consequences and Instrumental Targets
Objective: Clarify how definitions of life shape research practice.
Tasks:
Investigate how instruments (e.g., rovers, biosensors) embody assumptions about life
Study data structures in synthetic biology and AI that mimic life
Deliverables:
Report on operational biases in detection technologies
Reflection on definitional circularity in data interpretation
III. SYNTHESIS LAYER
The agent must generate:
A layered ontology of life, distinguishing:
Minimal life
Autonomous systems
Evolvable systems
Conscious systems
Decision framework for classifying ambiguous systems
Recommendations for future-facing definitions suited for novel contexts
IV. METHODS FOR THE LLM AGENT
The LLM should:
Use targeted web search to acquire primary literature and expert commentary
Query academic databases for cross-disciplinary sources
Perform critical comparison and synthesis of viewpoints
Distill conceptual invariants across physical, artificial, and hypothetical instantiations
It should annotate all sources and provide confidence assessments for each conceptual distinction.
V. OUTPUT SPECIFICATIONS
Final Report (~20,000–30,000 words): Modular chapters corresponding to the above domains, written for an expert audience.
Annotated Bibliography: Key texts across philosophy, biology, AI, astrobiology, and ethics.
Ontology Graph: Visual representation of layered life concepts.
Glossary: For cross-disciplinary terms (autopoiesis, dissipative structures, teleonomy, etc.)
Deep Research:
This is an exceptionally well-structured research agenda. Before I begin the research task, could you clarify how you would like the final outputs delivered? For example:
Should the 20,000–30,000-word report be delivered as one complete document or as modular instalments (per domain module)?
Would you prefer academic references in footnote, endnote, or inline citation format?
Should the ontology graph be constructed as a static diagram (e.g., PNG/SVG) or as a manipulable format (e.g., .graphml or JSON-LD)?
Once I know your preferences on these, I’ll begin assembling and synthesizing the materials.
VL:
Deliver the report as one complete document. Academic references should be in endnote format. The ontology graph may be omitted.
Deep Research:
Understood. I will begin compiling and synthesizing cross-disciplinary research to produce a comprehensive, substrate-independent ontology of life. The report will integrate philosophical, biological, computational, and astrobiological perspectives, include a full endnote-style referencing system, and be structured according to the modular framework you’ve specified.
I’ll notify you as each major section is completed and ready for review.
Towards a Substrate-Independent Ontology of Life
1. Meta-Theory of Definitions
Defining ‘Life’ – A Philosophical Perspective: The concept of ‘life’ has challenged philosophers and scientists for centuries, not only as a biological phenomenon but as a definitional puzzle. How we define life is not a trivial semantic matter; definitions shape our scientific inquiries and reflect deeper philosophical stances【53†L339-L347】【53†L367-L376】. In the philosophy of science, debates on definition range from classical essentialism (seeking a fixed essence of life) to contemporary scepticism that suggests perhaps no single definition can capture life’s complexity【53†L339-L347】【53†L367-L376】. To ground our approach, we first survey key theories of definition from philosophy, as they set the stage for how one might formulate a definition of life.
Carnap’s Explication and Scientific Definitions: Rudolf Carnap, a logical positivist, argued that many philosophical problems (like ‘What is life?’) stem from imprecise concepts【3†L14-L22】【4†L1-L4】. His method of explication involves replacing a vague ‘explicandum’ concept with a precise ‘explicatum’ defined by explicit rules within a scientific framework【3†L14-L22】. For example, Carnap notes philosophers ask questions like ‘What is life?’ without first clarifying the term’s meaning【4†L1-L4】. The explication approach would urge us to create a clear, operational definition of life for scientific use, rather than assuming an intuitive, folk concept. Carnap’s voluntaristic stance even suggests definitions are tools we choose for particular goals【1†L127-L135】【1†L139-L147】. In defining life, this means we might craft different definitions depending on whether our goal is, say, detecting extraterrestrial microbes or distinguishing persons from non-persons in law. The legacy of Carnap is a view that scientific definitions are not about capturing an eternal essence, but about constructing useful, exact concepts that advance inquiry【1†L129-L137】【3†L14-L22】.
Wittgenstein’s Family Resemblance: In contrast to the search for strict necessary and sufficient conditions, Ludwig Wittgenstein proposed that some concepts (like ‘game’ – and by extension possibly ‘life’) have no single essence but are defined by a family resemblance network of overlapping similarities【12†L575-L583】【12†L584-L590】. Wittgenstein argued that looking for one common element in all uses of a word is misguided【12†L575-L583】. Instead, concepts can be fuzzy, with instances related by a ‘complicated network of similarities overlapping and criss-crossing’【12†L579-L587】. Applied to life, this suggests that what we call ‘life’ might form a cluster category: cells, fungi, robots, and viruses might share some but not all features, without a single invariant trait. Rather than force a strict definition, a Wittgensteinian approach might accept multiple criteria (metabolism, reproduction, evolution, etc.) that overlap in various living things. Indeed, some have explicitly proposed a family resemblance definition of life in astrobiology, acknowledging that life may be a cluster of properties rather than a tidy essence【12†L575-L583】【12†L584-L590】. This view aligns with the difficulty scientists have faced in pinning down life: ‘although everyone knows what life is, there is no simple definition of life’, as one account put it【22†L153-L162】.
Cognitive Categorization – Lakoff’s Prototype Theory: Cognitive linguist George Lakoff and others have shown that human categories are often graded and centred on prototypes. In prototype theory, members of a category can be more or less typical, rather than equally fitting a strict definition【16†L1-L4】. For instance, a robin is a prototypical ‘bird’, while a penguin is an atypical bird – yet both count as birds. If we apply this to ‘life’, a human or oak tree might be a prototypical living thing, whereas a virus or a self-replicating computer program might be marginal, ‘sort-of’ living. Our intuitive concept of life may work similarly: we have central examples (animals, plants) and fringe cases (viruses, prions, artificial intelligences) which we hesitate or disagree on. Lakoff’s research suggests that we often categorise by family resemblance and prototypes, not by strict definitions【16†L1-L4】. Thus, an attempt to define life could acknowledge degrees or typicality. Some scientists indeed speak of ‘life as we know it’ versus possible life forms very different from our familiar prototypes【53†L339-L347】【53†L367-L376】. A prototype-based perspective warns us that our definition might inadvertently be anthropocentric, anchored on Earthly life’s examples.
Haslanger’s Ameliorative and Critical Definitions: Philosopher Sally Haslanger distinguishes between descriptive definitions (capturing ordinary use), conceptual analyses (uncovering our implicit concept), and ameliorative definitions (revising a concept for a purpose)【20†L123-L131】【20†L145-L152】. In her work on gender, for example, Haslanger suggests redefining concepts like ‘woman’ to serve feminist political goals, rather than to mirror common usage. Similarly, with ‘life’ one might ask: why do we need a definition? In astrobiology, the goal might be an operational definition to guide experiments; in ethics, an ameliorative definition might clarify what entities deserve moral consideration. Haslanger’s approach reminds us that definitions are not inherently true or false, but tools that can be evaluated by how well they serve our aims【20†L145-L152】. For instance, NASA’s working definition of life (discussed below) was developed to aid the practical task of detecting life beyond Earth【39†L1-L5】. Alternatively, an ethicist might define life in a way that aligns with moral intuitions about the sanctity of life. The meta-theoretical point is that there is no ‘view from nowhere’ definition – every definition of life has a purpose and context. Throughout history, definitions of life have shifted with scientific paradigms and philosophical needs, illustrating Haslanger’s point that our concept of life can be engineered for different projects.
Historical Evolution of Life-Concepts: Understanding how the concept of life has changed over time provides a ‘diachronic map’ of life-concepts. The ancient Greeks already grappled with what distinguishes living beings. Plato, for example, divided life into a hierarchy: vegetative life (nutrition and reproduction), animal life (sensory perception and movement), and rational life (human intellect)【50†L197-L205】. This tripartite soul concept influenced later thought, implying that life comes in degrees of capacity. Aristotle, taking a more empirical approach, defined life in terms of self-sustenance and change: for Aristotle, living beings exhibit self-generated motion or activity and can resist perturbations to maintain their form【50†L205-L213】. He identified nutrition, growth, and reproduction as fundamental life activities and saw the capacity for self-maintenance as the key difference between living and nonliving things【50†L205-L213】. In De Anima, Aristotle famously described the soul (psyche) as the form of a living body, essentially the principle that organises matter into a living thing. Thus, an oak has a vegetative soul, a dog a sensitive soul, and a human a rational soul – a qualitative distinction in the kinds of life.
Moving forward to the 17th century, René Descartes reconceived life in mechanistic terms. Descartes suggested that animals are complex automata – essentially machines made of matter – lacking the rational, conscious soul of humans【50†L215-L223】. He drew a sharp line between human life (which includes mind or spirit) and animal life, effectively denying that animals have consciousness or an immaterial soul. Notably, Descartes did not emphasize a huge divide between nonlife and life among physical processes; both were explicable by mechanics and physics. In his view, the difference was that living animals had more complex self-moving machinery (heart, muscles, etc.), but no new substance beyond matter in motion. This mechanistic paradigm shift set the stage for modern biology: it proposed that living organisms might be understood by the same physical principles as nonliving systems – a radical idea after centuries of assuming a special life-force. However, it also raised ethical and conceptual questions (if an animal is just a machine, what exactly is ‘life’ beyond complex chemistry?).
Vitalism vs Mechanism: Descartes’ reduction of life to mechanism provoked a long-running reaction in the form of vitalism. From the 17th through 19th centuries, various vitalist theories posited that living organisms possess a non-physical element or a special quality that inanimate things lack【50†L227-L236】. Different vitalists suggested different essences: an élan vital or vital force, an immaterial spirit, a unique biochemical principle, etc. What unified these diverse views was scepticism that physics and chemistry alone could fully explain life【50†L227-L236】. A classic moment in this debate was Friedrich Wöhler’s 1828 synthesis of urea (an organic compound) from inorganic reagents, disproving the notion that ‘organic’ substances require a life-force to be made【50†L229-L238】. Wöhler’s experiment is often cited as a death knell for vitalism in chemistry【50†L231-L239】, suggesting that living matter is made of the same atoms as nonliving matter. By the early 20th century, mainstream biology had largely abandoned explicit vital forces, though echoes persisted (e.g., Hans Driesch’s entelechy, or Bergson’s élan vital【50†L231-L239】). The vitalism vs mechanism saga shows how the definition of life has been entwined with what science is able to explain. As mechanisms sufficed to explain more phenomena (metabolism, development, heredity), the ‘extra’ vital quality kept being redefined or reduced.
Defining Life in Modern Science – A Taxonomy: In the 20th and 21st centuries, hundreds of definitions of life have been proposed【51†L242-L250】. To make sense of this multiplicity, scholars have attempted to categorise definitions into logical groups【51†L251-L259】【51†L285-L293】. Table 1 (adapted from Popa 2004, Trifonov 2011, and others【51†L251-L259】【51†L285-L293】) summarises some recurring categories:
Mechanistic or Thermodynamic Definitions: These view life as a kind of machine or physical process. For instance, thermodynamic definitions emphasize energy flow and entropy. Life is seen as a dissipative system maintaining itself far from equilibrium by consuming energy【29†L270-L278】【34†L3-L10】. (We will expand on this in Section 2.1). Such definitions often highlight metabolism or homeostasis as central. They align with the idea of life as matter and energy organised in particular ways【51†L251-L259】.
Metabolic and Physiological Definitions: A subset of mechanistic views, metabolic definitions say that life is chiefly characterised by self-sustaining chemical processes (metabolism). For example, a definition might require intake of nutrients, excretion of waste, growth, etc. J. B. S. Haldane quipped, ‘Try and define life and you’ll fail; say what living things do and you have progress.’ A metabolism-centred definition focuses on what life does: it uses energy to maintain itself. Some proposals list criteria like homeostasis, metabolism, response to stimuli, etc., essentially describing the physiology of organisms【22†L173-L181】.
Reproductive or Genetic Definitions: These emphasize information and heredity. For instance, Oparin (1961) simply said any system capable of replication and mutation is alive【53†L323-L331】. Such definitions align life with the capacity for Darwinian evolution – if something can reproduce with variations, leading to natural selection, it qualifies as life. NASA’s famous definition (discussed later) falls here. These definitions elevate information (genetic instructions, replication of patterns) as the hallmark of life【51†L285-L293】【45†L159-L166】.
Evolutionary (Darwinian) Definitions: Closely related to genetic ones, these require not just any replication but specifically evolution by natural selection. NASA’s working definition – ‘life is a self-sustaining chemical system capable of Darwinian evolution’ – is a prime example【39†L1-L5】. Here life is seen dynamically: what makes something alive is that it participates in an open-ended evolutionary process. A single organism might not evolve, but it belongs to a lineage that does. This category stresses evolvability as the defining feature【51†L279-L287】. It inherently excludes sterile individuals or isolated entities (leading to puzzles like the ‘one rabbit is not alive’ paradox【22†L153-L162】).
Autopoietic or Systems Definitions: These focus on the organisation of life rather than any specific substance or material. Maturana and Varela’s concept of autopoiesis (self-production) defines life as a system that continuously regenerates and maintains itself【36†L155-L162】【38†L184-L192】. An autopoietic system, like a cell, produces its own components and keeps its identity through self-regulation. This view highlights autonomy and organisational closure – life is a network of processes that turn matter into the components of the network itself【38†L184-L192】. Such definitions are substrate-neutral in principle; they say nothing about DNA or protein, only about the form of a self-sustaining network. Autopoiesis represents a holistic definition treating life as a collective property of a system【51†L257-L264】.
Bio-Chemical (Material) Definitions: Some definitions specify the chemical makeup of life as we know it – e.g. ‘life is carbon-based and water-based chemistry’ or ‘life uses proteins and nucleic acids’. These are usually considered too narrow (since extraterrestrial life might be different), but historically biologists often implicitly assumed life = ‘the form of organisms on Earth’. A cellular definition, for instance, might insist that cells are the fundamental units of life (so viruses, lacking cells, are not alive)【51†L265-L273】. Such cellularist definitions stress structure (membranes, organelles) as essential【51†L265-L273】. They may be criticised for being Earth-centric or describing life’s current form rather than defining the concept generally.
Information and Complexity Definitions: These are more abstract, seeing life in terms of information content or complexity thresholds. For example, life might be defined as a process that manages information to predict and adapt to the environment【45†L159-L167】. Christoph Adami, an information theorist, describes life as ‘self-perpetuating information strings’ that make predictions better than chance【45†L159-L167】. This approach treats genetic or adaptive information as primary, with biochemistry secondary. It intersects with theories about complexity – life as a certain level of computational complexity or logical depth. These definitions attempt a minimalist criterion: they seek one core property like ‘information processing’ to mark the boundary【51†L285-L293】.
Cybernetic and Artificial Life Definitions: With the rise of computers, some proposed that life is a process that can occur in any medium, not just biochemical. Chris Langton, a pioneer of Artificial Life (ALife), suggested that life is ‘a property of the organisation of matter’, and that matter is irrelevant as long as the pattern is right. This gives a definition where a sufficiently lifelike computer simulation could be literally alive. Such cybernetic definitions often rely on abstract properties like self-replication, evolution, and autonomy, but do not require carbon or metabolism【51†L293-L300】. They open the door to synthetic life in silico being genuine life. We will discuss in Section 5 how this plays out.
Vitalist or Organismic Definitions: Though vitalism per se is out of favour in science, some definitions echo its sentiment by positing that life is defined by a particular organising principle. Modern versions might invoke emergent properties or quantum effects unique to life. These often end up being ‘none of the above’ categories – asserting life has an X-factor beyond current science. While intriguing, they tend to be obscure or broad and have not gained traction as operational definitions【51†L293-L300】.
Pragmatic or Operational Definitions: Finally, scientists often use working definitions that are explicitly instrumental. For instance, a definition might list measurable criteria (metabolism, growth, reaction to stimuli, reproduction) simply as a checklist for identifying life in practice【22†L153-L162】【22†L173-L181】. These can straddle multiple categories (metabolic and genetic and evolutionary). They may not satisfy philosophers seeking a unified essence, but they serve to decide, say, if an observed Martian sample should be considered alive. Cleland and Chyba note that such definitions can be circular or overly tailored to known life, hence they advocate focusing on theoretical criteria rather than rigid definitions in astrobiology【53†L367-L376】.
This taxonomy of definitions shows just how multifaceted the concept of life is. Each approach isolates certain necessary and sufficient conditions for life according to a given theory. The thermodynamic view insists on entropy export; the autopoietic view requires self-maintenance; the evolutionary view demands replication with variation, etc. No single criterion has achieved universal acceptance, indicating that life as a concept might be inherently pluralistic or that our scientific understanding is still incomplete【53†L339-L347】【53†L367-L376】.
Diachronic Map of Life-Concepts: Over history, the working concept of life shifted from soul-like principles (Aristotle’s form of living things) to mechanistic processes (Descartes), to vital forces (19th century), to modern physico-chemical processes (20th century). By mid-20th century, definitions of life became an explicit subject of inquiry due to new contexts: origin-of-life experiments, space exploration, artificial life in computers, etc.【51†L242-L250】. For example, in 1944 physicist Erwin Schrödinger framed life in physical terms as that which evades entropy (order maintained against thermodynamic equilibrium)【29†L270-L278】. In 1953, the discovery of DNA’s structure underscored information and heredity as key to life. By 1976, NASA’s Viking Mars lander forced scientists to ask ‘what metabolic signals indicate life?’ leading to debates when the Viking experiments gave puzzling results. In the 1990s, artificial life experiments (like Tierra and Avida) made some wonder if digital evolving entities count as alive. The concept of life thus evolved from a qualitative essence to a subject of scientific definition and criteria.
We can map these changes roughly as follows:
Antiquity: Life as ensouled, a special form (Aristotle: life = self-moving, nutritive soul【50†L205-L213】).
Scientific Revolution: Life as machine-like (Harvey and Descartes: circulation of blood, reflexes – living bodies are complex mechanisms【50†L215-L223】), except humans who also have mind.
Vitalist backlash (18th–19th c.): Life as beyond mechanism (vital spark, élan vital)【50†L227-L236】.
Chemistry & Cell Theory (19th c.): Life as chemical (enzymes, metabolism) and cellular. The cell theory (1830s) defined the cell as the fundamental unit of life, drawing a boundary that persists today (viruses challenge this boundary as acellular).
Early 20th c.: Life as organisation (organicists like Joseph Needham, and later cyberneticists, emphasized holistic organisation). Also, Darwinism + genetics gave life a new definition: life is what evolves by natural selection (population thinking).
Mid 20th c.: Schrödinger (1944) – life as negative entropy feeder【29†L270-L278】; NASA (1960s–70s) – life as chemical system capable of evolution【39†L1-L5】. Operational definitions used in origin-of-life research (e.g., Stanley Miller’s experiments assumed that making organic compounds is stepping toward life).
Late 20th c.: Autopoiesis (1970s) – life as self-producing network【36†L155-L162】. Also, emergence of definitional scepticism: some argued we lack a fundamental theory of life, so any definition now may be premature (Carol Cleland’s view【53†L367-L376】).
21st c.: Intensifying interdisciplinary views: information theory of life【45†L159-L167】, astrobiologists’ expanded definitions (e.g., life might be chemical, but not necessarily Earth-like chemistry), and synthetic biologists creating borderline cases (like synthetic minimal cells, or protocells blurring nonlife and life).
Through this history, we see a pendulum swing between inclusive broadening of the concept (e.g. considering viruses, artificial life as life) and exclusive sharpening (e.g. requiring cellularity or Darwinian evolution, thereby excluding sterile hybrids or machines). The meta-theoretical lesson is that ‘life’ is a moving target: a concept shaped by theoretical advances and practical needs. Rather than one timeless definition, there is a lineage of overlapping concepts of life【50†L205-L213】【50†L215-L223】. We can visualise this diachronically as a branching tree of ideas: from an Aristotelian trunk, splitting into mechanistic and vitalist branches, recombining into modern synthetic views.
Philosopher of science Peter Medawar once remarked, ‘Definitions of life are dangerous: like a sorcerer’s apprentice, they get out of hand and take on a life of their own.’ The meta-theory of definitions reminds us why this happens. We imbue ‘life’ with scientific, existential, and moral weight, so every definition carries baggage. In the next section, we turn to the core theoretical approaches that have driven modern attempts to define life, examining what each theory sees as life’s essence and how these theories fare in terms of generality and substrate-independence.
2. Core Theories of Life
What makes something alive? Different theoretical frameworks answer this question with different essential features. Here we map out the major theoretical approaches to life and identify for each the proposed necessary and sufficient conditions for life. We will also evaluate how general or substrate-neutral each approach is – in other words, would it apply to life as we don’t know it, or only to familiar life?
2.1 Thermodynamic and Physico-Chemical Theories
One influential approach views life through the lens of thermodynamics and physics. Erwin Schrödinger’s famous 1944 book What Is Life? framed the question in terms of physics: how do living organisms maintain order despite the Second Law of Thermodynamics? Schrödinger observed that organisms avoid decay and equilibrium by feeding on negative entropy【29†L270-L278】. In his words, living matter ‘evades the decay to thermodynamic equilibrium’ by maintaining a locally low entropy state, drawing energy from the environment【29†L270-L278】. This idea translates to a necessary condition: a system is alive only if it continually exports entropy (disorder) to keep itself organised. It positions homeostasis – maintaining internal order against external entropy – as a hallmark of life.
From a thermodynamic perspective, a sufficient condition for life might be: a far-from-equilibrium chemical system that self-maintains and self-reproduces by dissipating energy gradients. Ilya Prigogine’s work on dissipative structures (for which he won the Nobel Prize) reinforced this idea. Prigogine showed that under far-from-equilibrium conditions, matter can spontaneously form ordered structures (like convection cells, chemical oscillations) that persist by dissipating energy【34†L3-L10】. He and colleagues noted that living organisms are quintessential dissipative structures: they constantly produce entropy (through metabolism) but export it to the environment, thus maintaining internal structure【34†L3-L10】. In fact, modern nonequilibrium thermodynamics often asserts that living organisms maximise entropy production locally while maintaining structural order【34†L3-L10】.
A key concept here is homeostasis: living systems regulate their internal environment (temperature, pH, etc.) within bounds, which is essentially maintaining a low-entropy, ordered state. This is linked to self-regulation. Another is metabolism: the network of chemical reactions that captures, converts, and dissipates energy. A pure thermodynamic definition of life might be: Life is a self-sustaining network of energy and matter transfer that maintains a low-entropy state by dissipating energy gradients. This covers everything from bacteria to potentially hurricanes or stars? Indeed, a challenge arises: certain nonliving systems, like flames or vortices, also maintain themselves by energy flow. To refine the criteria, physicists often add complexity or information requirements (which leads into information-theoretic views, Section 2.4).
Necessary conditions from this view: (a) Open System – exchange of matter/energy with environment (closed systems inevitably equilibriate and ‘die’); (b) Entropy Export – the system must produce entropy (waste heat, etc.) and expel it; (c) Far-from-equilibrium – the steady state is significantly ordered compared to thermodynamic equilibrium; (d) Chemical complexity – often presumed, as life’s order is very intricate, not just a simple crystal.
Sufficient conditions are harder to pin down because not every dissipative structure is intuitively alive. A whirlpool is an ordered vortex in water that persists by energy flow, but we don’t call it alive. So, thermodynamic theorists often combine these criteria with others (like reproduction or information). Still, this approach provides a substrate-neutral insight: any system, carbon-based or not, that harnesses energy to create local order might qualify as life. If we found silicon-based creatures metabolising and maintaining homeostasis on Titan, they would neatly fit this definition.
One must also consider boundary and autonomy: In thermodynamic terms, living systems often have a boundary (cell membrane, skin) that separates internal order from chaos outside while allowing controlled exchange. So, an enclosed far-from-equilibrium system capable of self-maintenance and perhaps self-reproduction covers classic life.
Generality: Thermodynamic definitions are fairly substrate-neutral. They talk about energy, entropy, and matter – universal physics that apply anywhere in the universe. Thus, they could encompass exotic life forms built on different chemistry. The downside is they may be too general (risking false positives by including phenomena like wildfires or autocatalytic chemical cycles that burn out).
In summary, the thermodynamic approach gives us one necessary condition for life: continuous energy-driven self-maintenance is necessary (living things must take in free energy and export entropy【29†L270-L278】). Some proponents might say it is also sufficient if the system is complex and autonomous enough. As one paper put it, ‘living organisms are undoubtedly dissipative structures, given that their existence and behaviours are sustained by continuous flows of energy and matter’【34†L3-L10】. However, thermodynamics alone doesn’t capture heredity or evolution, so many theorists consider it a part of a multi-faceted definition rather than a complete theory on its own.
2.2 Autopoietic and Systems Theories
The autopoietic theory of life, developed by Humberto Maturana and Francisco Varela (1970s), centres on the idea of a system that continuously produces itself. Autopoiesis literally means ‘self-making’【36†L155-L162】. An autopoietic system is defined formally as ‘a machine organised as a network of processes of production (transformation and destruction) of components which continuously regenerate and realise the network of processes that produced them, and constitute the system as a unity’【38†L184-L192】. In plainer terms, a living cell, for example, is a self-bounded chemical factory: all its components (proteins, membranes, etc.) are produced by interactions within the cell itself, and those components in turn reconstruct the whole arrangement. This circular, self-referential organisation is claimed to be the defining feature of life.
Necessary & Sufficient (Autopoiesis): Autopoietic theory claims to provide both necessary and sufficient conditions: if something is autopoietic, it is alive; if it is alive, it must be autopoietic (in the physical domain). Maturana and Varela originally offered it as the definition of life【36†L155-L162】【38†L184-L192】. By this view, metabolism (self-production of components) and a boundary (like a cell membrane) are indispensable – together they create a self-sustaining, individuated system. Notably, reproduction is not required in autopoiesis; a single cell that never divides is still alive as long as it continuously regenerates itself. Nor is evolution required at the level of the individual. Thus, autopoiesis is more lenient on those points than evolutionary definitions, but stricter in requiring an autonomous self-producing unity. Viruses, for instance, would fail because they cannot self-produce – they hijack a host. An autopoietic purist might say viruses are not alive, whereas a genetic definition might include them as life’s products.
Autopoiesis ties life to organisation rather than matter. It is substrate-neutral in principle: any network of processes that closes on itself and makes more of itself could be life, whether chemical, electronic, etc. However, it implicitly assumes some material realisation where components are physically produced and replaced. This theory strongly influenced systems biology and Origins of Life research by emphasizing metabolism-first views (life begins when a network of chemical reactions becomes self-sustaining and enclosed). Some experimental origin-of-life studies aim to create protocells that are autopoietic: fatty acid vesicles with internal chemistry that can grow and divide on their own. If achieved, that would satisfy autopoietic sufficiency for life.
One critique is that autopoiesis might be too broad: certain chemical systems (like a flame in a candle) regenerate themselves (the flame sustains wax melting which sustains the flame) but are not bounded unities with a distinct individuality. Autopoiesis insists on a boundary (the candle flame has none – it dissipates into air), so it avoids that example. Another issue is whether autopoiesis implies any form of information or program; originally it was a purely dynamical systems notion, but later analyses (e.g., ‘organisational closure’ and ‘constraint closure’ approaches【38†L224-L233】) have extended it.
Necessary conditions from autopoiesis: (a) Boundary: a self-produced membrane or delineation that separates the system from environment; (b) Internal production: components of the system (enzymes, structural molecules) are produced internally by the system’s network; (c) Closure: the network of processes is circularly closed (each process depends on products of others and contributes to producing others), achieving a stable unity.
Sufficient condition (claimed): Any system with properties (a), (b), (c) is a living system【38†L184-L192】. This is a bold claim and not universally accepted, but it provides a clear theoretical target. The autopoietic definition was applied originally to cells, but Maturana and Varela also connected it to cognition (they famously said ‘living is cognition’, implying that even basic life has a form of sensing and responding meaningfully to its environment)【38†L230-L238】. This links life and mind (we’ll explore that in Section 4).
Generality and Substrate-Neutrality: Autopoiesis is quite general. It doesn’t mention carbon, water, DNA, or any specific molecules – only a pattern of self-production. Therefore, in principle, a robot that can repair and replicate itself from raw materials, or a planetary ecology that self-maintains, could be seen as autopoietic. However, there is debate on whether autopoiesis can scale up beyond cellular life. Some have argued the definition might exclude ecosystems or biospheres because those might not be as tightly bounded or unified. Others have tried to apply the concept in sociology (social systems that self-produce, per Niklas Luhmann【36†L163-L167】).
In astrobiology, an autopoietic definition would direct us to look for self-contained chemical factories on other planets – perhaps lipid vesicles with metabolism. It might miss life forms that do not have a single enclosing membrane (imagine a distributed lifeform in rock pores that exchanges components – is that autopoietic as a whole?).
Nevertheless, autopoiesis offers a compelling necessary condition: organisational closure. If something cannot sustain itself by internal processes, it’s not alive by this view. And indeed, anything we intuitively call alive (bacteria, plants, animals) does have such closure – cut off the energy supply or break the organisational loop and the organism dies.
2.3 Evolutionary and Selectionist Theories
A very prominent approach, especially in origin-of-life studies and astrobiology, is that life is defined by the capacity for Darwinian evolution – that is, life forms replicate with heritable variation such that natural selection can occur. This is often called the NASA definition, because a NASA committee in 1994 adopted: ‘Life is a self-sustaining chemical system capable of Darwinian evolution’【39†L1-L5】. The inclusion of ‘Darwinian evolution’ as the litmus test was influenced by suggestions from Carl Sagan and was summarised by biochemist Gerald Joyce【39†L1-L5】.
Necessary condition (evolutionary view): The system must contain information that is copied with variation and subject to selection. In practice, that usually means a genetic apparatus of some kind (DNA, RNA, or any molecular code) and a life cycle where offspring resemble parents but not exactly (allowing evolution). This addresses one intuitive hallmark of life: life evolves. Even if we found a weird crystal that metabolises, if it cannot evolve or adapt across generations, many would hesitate to call it alive – it’s more like a machine.
Sufficient condition (as often stated): If a chemical system can undergo open-ended Darwinian evolution, it qualifies as life【39†L1-L5】. In other words, the emergence of Darwinian evolution is the emergence of life, in this view. Joyce and others have argued that once a molecular system achieves the ability to replicate with heritable mutations, it will exhibit life-like behaviour (adaptation, complexity increase) and should be considered alive. This is a common stance in origin-of-life research: the origin of life is essentially equated with the origin of evolution. It’s why experiments attempt to get self-replicating RNA – to demonstrate the ‘spark’ of life in a pool of chemicals when evolution starts.
This approach tends to exclude any non-evolving entities from life. For example, a sterile hybrid animal that cannot reproduce – is it alive? By everyday sense yes, by strict Darwinian definition, that individual doesn’t contribute to evolution, but as a member of an evolving species it’s alive. So typically the Darwinian requirement is applied at the lineage or species level: a life-form belongs to a line with evolutionary capacity. Another tricky case: mules (infertile) or laboratory-created organisms that can’t reproduce. Most would still call them alive, but note they fail the ‘capable of Darwinian evolution’ individually. The definition can flex to ‘derived from a Darwinian lineage’.
NASA’s definition also required ‘self-sustaining’ and ‘chemical’【39†L1-L5】. ‘Self-sustaining’ means it isn’t just a virus that depends on another life. ‘Chemical’ was included to ground the definition in known science – implying physical, embodied existence (as opposed to a mathematical pattern alone). This was slightly controversial because it might exclude, say, hypothetical plasma life or purely digital life, which some would consider possible. Indeed, Steven Benner quipped that requiring ‘chemical’ could be too restrictive【21†L19-L27】. Nevertheless, the Darwinian view usually assumes a material substrate in which information is instantiated and copied.
Generalisable? The Darwinian definition is quite general if you remove ‘chemical’. One could say life is any self-propagating system of information that undergoes evolution by natural selection. That could include AI programs that copy themselves with variation. In fact, the field of digital life (like the Avida platform) has produced self-replicating programs that evolve within a computer memory. Do they count? Many proponents of this view would say yes – if they truly evolve complexity via selection, they exhibit life processes (albeit in silico). In practice, NASA kept ‘chemical’ to emphasize extraterrestrial life is likely based on chemistry (and to avoid debates about computer viruses).
Necessity vs sufficiency: Some argue the Darwinian criterion is too stringent as a present-day definition – for example, an organism that doesn’t currently evolve (like bacteria frozen in stasis or an organism in a stable environment not under selection) is still alive. But the idea is that life in principle involves these processes. It’s a bit like saying a machine is ‘a device capable of motion’, even if it’s currently turned off. Also, what about proto-life before full Darwinian evolution kicked in? That’s a grey area this definition deliberately draws a line through: It would say before evolution, it wasn’t life, it was chemistry. That’s a stance to be debated (some prefer a continuum concept of prelife to life).
Why this criterion? Because Darwinian evolution uniquely leads to open-ended complexity and adaptation. Non-living systems can have metabolism or homeostasis (stars have cycles, thermostats regulate temperature), but only populations of living things evolve adaptively. The evolutionary view captures life’s creative, informational aspect: life adapts and learns over generations, something no rock or fire does.
Substrate neutrality: The core of Darwinian evolution is abstract – replicators with variation and selection. This can happen with DNA on Earth, but could with radically different molecules elsewhere, or even in a computer network. Thus, it’s largely substrate-neutral. If we met aliens who reproduce and evolve but with no familiar biochemistry, this definition would welcome them as life (they clearly evolve). If we encountered an AI that reprograms itself and spawns varied copies which compete – that arguably fits too (though ‘self-sustaining’ might be debated if it depends on human electricity supply).
One implicit necessary condition is information storage: Darwinian systems need a way to store traits (genes, code). So life must involve some physical medium that carries information (genetic memory). Another is imperfect replication: perfect copying would not allow evolution, so ‘mutation’ is ironically an essential part of stable life in the long run.
Edge cases and criticisms: This approach notoriously struggles with entities like viruses. Viruses have genes and evolve (viral populations adapt quickly), but they are not self-sustaining – they parasitise host cells to reproduce. NASA’s definition probably would exclude viruses as not autonomous life forms (they’re not ‘self-sustaining’ chemically). Yet viruses demonstrate Darwinian evolution in spades. So are they alive or not? A NASA astrobiologist might say viruses are life’s byproducts and require life to exist first. In origin scenarios, one would not consider viruses the first form of life since they need hosts.
Another edge: prions (infectious proteins) have a form of evolution (mutation and selection of protein conformations), but no genes. They are usually not counted as life because they can’t do anything except reproduce within a host organism’s context (and even that is borderline mechanism).
In sum, the evolutionary/selectionist theory of life claims: Heritable evolution by natural selection is the sine qua non of life. It provides a solid necessary condition – we have yet to find unequivocally nonliving things that undergo sustained Darwinian evolution. And as a sufficient condition, it stakes out a clear boundary (though perhaps prematurely, as Cleland warns【53†L367-L376】). This approach is well-suited to generalising life because it’s tied to an abstract process, not a specific material. It might, however, exclude potential unusual life that doesn’t evolve in a Darwinian way (imagine a life form that’s individually intelligent and can adapt in real-time without needing generations – would we call it life? Possibly yes, but it breaks the Darwinian mould).
2.4 Information-Theoretic and Computational Theories
Another theoretical orientation defines life by how it processes information or by its computational characteristics. In this view, life is less about what it is made of or even what it does energetically, and more about what it computes or what information it embodies. Several scientists have championed this perspective:
Life as Information: Christoph Adami, for instance, suggests that life is fundamentally an information phenomenon. He describes organisms as ‘self-perpetuating information strings’ and defines the information of life as the ability to predict environmental states better than chance【45†L159-L167】. This resonates with the idea that organisms store information about their environment in their genomes (as a result of past adaptation) and in their sensors/brains (as a result of learning). Adami’s work with digital organisms in Avida (a software platform where small self-replicating programs mutate and evolve) demonstrated that those digital creatures behave in Darwinian ways and accumulate information about their computational ‘environment’ over time【45†L159-L167】. If one equates that with life, it suggests that the substrate (DNA vs computer memory) is irrelevant, what matters is the informational dynamics.
From an information standpoint, necessary conditions for life might include: (a) a mechanism to store information (like a genome or memory), (b) a mechanism to process that information (to make decisions or predictions), and (c) the ability to use information to maintain oneself and/or reproduce. Some go further and tie life to concepts like algorithmic complexity or logical depth. For example, life forms often have a higher information content than random matter. A rock’s description is short (just its mineral composition), whereas a single cell’s description (listing all molecular positions) is enormously complex – but more importantly, that complexity is organised, with high mutual information among parts (DNA influences proteins, etc.). Life has been called matter with context-dependent information interactions (e.g., the genetic code is context for nucleotide sequences).
Computational View – Life as a machine class: Chris Langton and the Artificial Life (ALife) community in the 1980s onward took a radical step: if life is a property of form, not matter, then we should be able to create life in computers by simulating the right forms. Langton said that the study of Artificial Life is about ‘life-as-it-could-be’ and that we should not be constrained by Earth’s chemistry. One famous line (paraphrased) from Langton is: ‘Life is a property of the organisation of matter, rather than a property of the matter itself.’ This encapsulates substrate independence. According to this, any system – biochemical, silicon, quantum, whatever – that has the right organisation (e.g., self-replication, evolution, functional integration) is alive【51†L293-L300】.
This computational lens often draws on the theory of complex systems. Life is seen as a certain kind of complex, self-organising, adaptive system. Mark Bedau, for example, speaks of life as showing ‘supple adaptation’ – persistent adaptation in response to conditions (like evolving or learning continuously). The exact computational criteria can vary. Some focus on autonomy and computation: an organism might be seen as an agent that takes inputs (stimuli) and produces outputs (actions) that serve its self-maintenance – essentially an autonomous robot. Others emphasize information networks: life as a network of information transfer (DNA->RNA->Protein being a prime example, or ecological food webs as information flows about resources).
Necessary conditions (information view): The ability to process information in a goal-directed way could be proposed as necessary. If something has no informational autonomy – say a crystal just passively exists – it’s not alive. Even a thermostat processes one bit of information (too hot/cold) and acts (turn AC on/off), but we might require a far higher degree of sophistication (thermostat is designed externally, not evolving or self-maintaining largely). So necessary might be: a self-driven information cycle (the organism has sensors and effectors and a control loop). Another necessary condition might be the presence of a genetic program. Some definitions require a ‘program to coordinate metabolism’ as one of pillars of life (e.g., Daniel Koshland’s ‘program’ in his PICERAS list【22†L173-L181】).
Sufficient conditions (hypothetical): If we had a system that stores a self-description and can use it to reproduce (a la von Neumann’s self-reproducing automaton concept), plus can adapt that description, many would say that’s a living system, whether it’s molecules or bits. John von Neumann showed abstractly that a machine could be constructed with a ‘tape’ (description) and a universal constructor that reads the tape to build a copy of the machine, including the tape – a theoretical model of reproduction. The key insight: reproduction of information plus mechanism, not just of matter. Life does exactly that (DNA is the tape, cells are the constructors). Thus, a sufficient condition for life might be: a self-reproducing automaton with an evolvable program. This combines the computational and evolutionary view.
Substrate neutrality: The information perspective is highly substrate-neutral because it deals in abstractions – bits, logic, algorithms. If consciousness or adaptation can be uploaded to a different medium and still function, then life is portable. This viewpoint would easily count sophisticated AI or artificial organisms as life if they meet the informational criteria (some argue computer viruses are ‘life’ in an info sense because they self-replicate and adapt, but they lack physical self-maintenance and robust evolution, so typically they’re considered a limited analogy).
One interesting proposal merging information and thermodynamics is to measure the entropy or information content of an organism’s genome relative to random sequences. Adami and others have tried to quantify how much information is gained by digital organisms as they evolve. They equate that gain with the organism ‘learning’ about its environment (since adaptations incorporate knowledge of the environment’s conditions)【45†L159-L167】. By such measures, life could be defined as systems that accumulate knowledge/information over time (through evolution or neural learning). That’s an appealing high-level definition but tricky to measure in practice for unknown entities.
Limitations: A challenge for information-based definitions is that information is always defined relative to an observer or goal. DNA has information because we interpret it as coding proteins. But raw polymer might not be recognised as info without context. So one must be careful not to get tautological (life is what processes information, and we only call it information because it’s in life). Proponents address this by grounding information in physical causation: if a sequence influences outcomes (e.g., sequence of bases leads to functional differences), then it’s information in an objective sense (causal entropy reduction or constraint).
In summary, the information/computational approach brings out that life is not just about matter and energy, but about how matter and energy are organised to store and use information. It provides a potentially sufficient condition: any self-organising system that processes information to perpetuate itself and can evolve that informational content is life. It also implies necessary conditions about complexity and adaptiveness. This approach nicely includes things like RNA world scenarios (RNA storing info and catalysing its own replication) as life once they reach a certain level. It also is friendly to recognising advanced AI or machine life in the future, since presumably if it quacks like a self-sustaining, cognitive, evolving system, it is one – regardless of made of silicon.
2.5 Algorithmic and Artificial Life Approaches
Closely related to the information view, but deserving separate mention, are the algorithmic approaches stemming from Artificial Life experiments. Thomas Ray’s Tierra (early 1990s) was a landmark: he created a virtual computer ‘ecosystem’ where programs competed for CPU time and memory, mutating and evolving. These programs (initially designed to self-replicate) evolved unexpected strategies – parasites, hyper-parasites, coevolutionary behaviour. It demonstrated open-ended evolution in silico to a degree【51†L293-L300】. Ray and others argued such digital organisms meet the criteria for life: they have a genotype (their code), phenotype (their executed behaviour in the CPU), compete for resources, reproduce, mutate, and undergo natural selection – all in a computational medium.
Necessary conditions (ALife view): Very few – maybe just an evolutionary algorithm happening. Artificial Life researchers often adopt Langton’s dictum that if it behaves like life, it is life, no matter the substrate. So perhaps the only necessary condition is that it instantiates life-like processes (which begs the question a bit). But typically they require at least that it’s not directly orchestrated by an outside agent – it should have spontaneous emergent behaviour.
Sufficient conditions: Langton’s list in the first Artificial Life conference included: self-reproduction, information storage of that reproduction, functional interactions with environment, and the capacity for open-ended evolution (not reaching a fixed point) as being sufficient to declare something alive (in simulation)【51†L279-L287】【51†L293-L300】. These mirror the earlier categories but couched as algorithmic properties. The dream scenario is a computer simulation where things evolve increasing complexity indefinitely – analogous to the biosphere. If that were achieved, many would be hard-pressed to deny that ‘life’ exists in that computer (even if it’s life in a virtual world).
ALife experiments highlight the notion of strong vs weak artificial life: Weak ALife says we’re just simulating life processes (like weather simulations aren’t real weather, just models), whereas Strong ALife posits that the right simulation is an instance of life (like running Windows in a virtual machine is still an instance of Windows OS, not a fake). Strong ALife adherents hold that life is a higher-level pattern and that implementing that pattern in any medium creates life. Thus, running an evolving program could be literally creating a new life form, not just a representation.
Generality: Algorithmic definitions are maximally general: life could be made of electrons in a circuit, or even something like patterns in a cellular automaton (Langton even studied the Game of Life cellular automaton by Conway as a toy model of how complexity can arise). The Life-Like cellular automata are not usually called truly alive, but they show how minimal rules can mimic growth, reproduction, etc.
One interesting concept is the computer universality of life: Von Neumann proved a universal constructor is possible (a machine reading a description and building what it says, including itself). It’s hypothesized that biological life is a form of such a universal construction system (the ribosome reads mRNA (description) and builds proteins, which along with other cell machinery can construct a new cell). This algorithmic self-replication view informs definitions: perhaps any universal constructor with an programmable description is alive. Our cells come close – DNA is like a description, the cellular machinery builds according to it. Future artificial life might implement the same principle in robotics or nanotech.
Evaluation of Substrate Neutrality: The ALife approach is by design substrate-neutral. It takes abstraction to the extreme: one could conceive life in an electrical pattern, or even life existing in the patterns of a social network (memetic life?) if the formal properties line up. This openness is exciting because it doesn’t tie life to Earth-centric chemistry, but it also means one must carefully ensure the definition doesn’t become so broad as to include trivial self-replicating processes that lack what we intuitively consider ‘life’s richness’. ALifers typically mitigate this by stressing open-ended evolution (the ability to keep producing novelty indefinitely). Many simple replicators (like certain repetitive chemical cycles or autocatalytic loops) might hit a dead end and not diversify; those would be seen as limited ‘proto-life’ but not full life.
Necessary and Sufficient Summaries for Each Approach: Having outlined these core theories, we can distill the key proposed criteria:
Thermodynamic: Necessary: exchanges energy, increases entropy of environment while maintaining internal order【29†L270-L278】. Possibly sufficient: a far-from-equilibrium self-maintaining system (plus complexity) is alive【34†L3-L10】. Substrate: any physical system obeying thermodynamics.
Autopoietic (Systems): Necessary: self-producing network with a boundary (organisational closure)【38†L184-L192】. Sufficient: a system that continuously regenerates its components and maintains itself as a distinct unity is alive【38†L184-L192】. Substrate: any network of processes (likely chemical, but could be other) that can instantiate this closure.
Evolutionary (Darwinian): Necessary: ability to undergo evolution by natural selection (heritable variation)【39†L1-L5】. Sufficient: a self-sustaining system that can evolve adaptively over generations is alive【39†L1-L5】. Substrate: typically chemical, but conceptually any replicator system.
Informational: Necessary: presence of a genetic program or information processing that guides self-maintenance【45†L159-L167】. Sufficient: a self-perpetuating informational pattern that can adapt/predict (learn or evolve) from environmental inputs【45†L159-L167】. Substrate: any (biochemical, digital, etc.) – emphasizes software over hardware.
Algorithmic/ALife: Necessary: exhibits life-like behaviour (growth, replication, evolution) within some formal system【51†L293-L300】. Sufficient: implements the ‘logical form’ of life (e.g., self-replication with open-ended complexity increase) in any medium【51†L293-L300】. Substrate: completely general – life-as-it-could-be (silicon, maths, etc.).
Evaluating Generalizability: Each theory has its scope. Thermodynamic and autopoietic approaches shine in describing individual organism’s maintenance, but by themselves they might consider a perfectly homeostatic sterile organism as life (which is fine) yet fail to demand evolution. Evolutionary and algorithmic approaches capture long-term adaptation, but they often depend on a population concept and might exclude solitary or unusual life-forms that don’t fit a population model. An ideal ontology might integrate these: for example, NASA’s definition tries to by saying ‘self-sustaining’ (metabolism, autonomy) and ‘Darwinian evolution’【39†L1-L5】 – combining metabolic and evolutionary criteria.
Substrate neutrality is crucial for a ‘general theory of life’. The thermodynamic approach is tied to physics (so any physical substrate qualifies), autopoiesis is structural (again any materials as long as structure is realised), evolutionary is abstract but often assumed chemical, informational and ALife are fully abstract. The more abstract, the more substrate-neutral, but perhaps also the more removed from concrete detection (e.g., it’s hard to measure information processing on an alien planet without looking at its physical chemical signs which circle back to metabolism).
In practice, different researchers use a mix of these criteria. One survey found that scientists often ‘straddle’ categories when defining life【53†L322-L330】. For example, an origin-of-life definition might require a membrane (autopoiesis), metabolism (thermodynamics), and replication with mutation (evolution) all together【53†L323-L331】. An example given by Popa (2004) is a definition requiring: (1) bounded micro-environments; (2) energy transformation to maintain low entropy; (3) replication with an instructional code【53†L325-L333】. That combines autopoietic (bounded, low entropy) with evolutionary (code for replication) in one definition.
Having mapped core theories, we next consider a particular debate: do definitions of life depend on how life started? That leads us to origin-based considerations and whether ‘life’ inherently implies a natural genesis (abiogenesis) or if an engineered or spontaneously ordered system can count as life just the same.
3. Origin-Agnostic vs Origin-Constrained Models
Does how something came to be affect whether we call it alive? This question separates two philosophical stances we might call causal essentialism versus organisational essentialism about life.
Causal essentialism (in this context): the idea that something’s status as ‘alive’ depends on its causal history – notably, whether it arose through natural processes like abiogenesis and evolution. In other words, life is essentially defined by having a certain origin (perhaps emerging from nonlife via Darwinian evolution). This could imply that if we artificially constructed an organism in a lab, one might hesitate to call it ‘real life’ because it didn’t emerge from the usual causal pathway. It’s akin to saying, ‘X is alive if it descends (causally) from the lineage of life.’
Organisational essentialism: the view that what matters is the organisation and properties a system has here and now, regardless of how it came about. If it has the structure and functions of life, it is life, whether it was born, evolved, or manufactured. This treats a living organism and an exact replica (even if 3D-printed) as equally alive, because they both exhibit the defining organisation of life.
In most scientific discourse, life is defined by present characteristics, not by origin. If an alien biochemist found a Toyota robot that perfectly mimics a dog’s behaviour, the origin (factory vs evolved animal) is external information; they might classify it by observed properties. However, origin can creep in subtly. Consider NASA’s emphasis on Darwinian evolution【39†L1-L5】: it’s partly an origin criterion because it implies life must have come about by an evolutionary process. A self-sustaining chemical system we build that does not evolve might be denied the label ‘life’ by that definition because it’s not capable of Darwinian evolution (unless we explicitly program that capability).
Abiogenesis vs Engineered Life: One concrete issue is whether a synthetic cell created from raw chemicals in a lab would be ‘alive’ in the same sense as a naturally evolved cell. Most would say yes – if it walks like a duck, etc. But some definitions might exclude it inadvertently if they tie to natural selection. Another scenario: Suppose in the distant future, a complex AI spontaneously emerges within the internet, without direct programming – would calling it ‘alive’ depend on it evolving versus being designed? If one is a causal essentialist with respect to biology, they might say ‘life, properly so-called, is something that originates from natural evolution; a designed thing might simulate life but isn’t genuinely life’.
The philosophical literature on life definitions acknowledges this tension. Carol Cleland and Christopher Chyba argued that trying to formally define life now might be futile partly because all examples we have share a single common ancestor and thus one origin【53†L367-L376】. There might be a ‘projectable property’ that all Earth life has due to common origin, but which is not truly universal to all possible life. Cleland suggests focusing on a broader theory of life and keeping an open mind (using tentative criteria rather than a fixed definition) so as not to miss life that has a different origin【53†L367-L376】. This supports an origin-agnostic stance: we should define life by abstract properties not tied to Earth’s particular genesis.
Causal arguments in origin of life research: Some have speculated that life might be better understood as a process that arises under certain conditions. For example, physicist Jeremy England has a hypothesis that life-like self-organisation is a probable outcome of systems dissipating energy. If true, life is a process that the universe can throw up in multiple ways, not a fluke lineage. That leans toward organisational essentialism: life emerges from certain conditions and what matters is those conditions and resulting organisation, not a single tree of life.
Engineered vs Naturally Emergent: Historically, definitions seldom explicitly require ‘natural emergence’. But implicitly, much of biology has focused on natural life. The question ‘is an artificial automaton alive?’ was largely hypothetical until recent decades. Now with synthetic biology making protocells and AI robotics, we face it. For instance, in 2010 Craig Venter’s team created a bacterial cell with a fully synthetic genome (assembled from base pairs on a computer, then inserted into a cell shell). They called it ‘synthetic life’, but was its status in doubt? It used an existing cell cytoplasm to boot up, so some said it’s not new life, just a remix of existing life components. But imagine we truly build a cell from scratch – new membrane, new metabolic pathways, new genetic code – if it metabolises and evolves, by organisational criteria it’s alive. By causal criteria, one might say it’s derived from life because we used knowledge from life to guide its design, but that seems a stretch – that’s like saying a synthetic diamond isn’t a diamond because it didn’t form naturally. Most accept synthetic diamonds as real diamonds (just different origin), similarly synthetic life would be real life if all functional hallmarks are present.
Spontaneous Order vs Life: There are cases of self-organisation that produce surprising order (Bénard cell convection, autocatalytic sets in chemistry). Could these ever be alive or are they missing something? Stuart Kauffman studied autocatalytic networks and suggested life began when a chemical network became collectively autocatalytic (each reaction’s products help another reaction happen). That’s an organisational moment. But does one consider an autocatalytic set alive before it evolves? Kauffman might say yes if it’s self-sustaining. Others say not until it can evolve information. This again is origin-related: did life require the chance origin of a replicator or could metabolic organisation count as life even without Darwinian evolution initially? ‘Origin-agnostic’ means we don’t build in that it must tie back to a single replicator origin. It’s debated if that initial network should be called life or ‘proto-life’.
Causal Essentialism Examples: One way this appears is in considering borderline entities like viruses. Some argue viruses are not ‘fully alive’ because they likely evolved after cells (thus are more like offshoots). But if one found something with similar properties that was primordial, would they call it life? Possibly. So the virus debate sometimes hinges on origin: viruses depend causally on cells (they cannot trace an independent lineage from origin of life—they are parasites), so that dependence might disqualify them in some views. If life is defined as an independent lineage, viruses fail; if defined by organisation at time, viruses pass many tests (they have genes, evolve, but fail metabolism and self-production).
We can articulate causal vs organisational essentialism in terms of necessary conditions:
Causal essentialist might add: ‘To be life, an entity must be part of a lineage that arose from natural abiogenesis and has undergone evolution.’ This is rarely stated formally, but conceptually, it’s like saying life is by definition what natural processes produce. Under that, a perfect artificial copy of a bacterium is not ‘original life’ (though it’s alive as an instance perhaps). This view is not mainstream but lurks in some arguments about AI or synthetic organisms lacking the ‘vital spark’ of natural life.
Organisational essentialist: ‘To be life, an entity need only exhibit the defining organisation and functions of life (metabolism, reproduction, etc.), irrespective of how it originated.’ This is the usual scientific approach – if it quacks, it’s a duck, no matter who built the duck.
One area this matters is astrobiology: If we find replicating chemical systems on Titan’s seas that clearly are active and evolving, they have a separate origin from Earth life. We’d likely call it life immediately (assuming we convinced ourselves it’s not contamination). We wouldn’t say ‘ah, but it didn’t come from Earth’s origin, so it’s not life’. Instead, we’d be ecstatic to find a second genesis of life. That affirms that our concept of life is origin-agnostic scientifically – life is a category that can have multiple independent origins (in fact, proving a second origin would deepen our understanding of life by comparison).
Causal history in Ethics: Interestingly, the notion of life’s beginning (causal beginning) matters in law/ethics (abortion debates revolve around when life begins for an individual – conception, heartbeat, etc., which is a different but related causal consideration). However, that’s more about individual life history, not life as a category.
Natural Kinds and Projectability: In philosophy of science, natural kind essentialism would ask if ‘life’ is a natural kind with an underlying essence (like water is H₂O). If life is like water, then perhaps when we have a full theory, we’ll say ‘all life has property X (e.g., a certain chemical organisation or informational architecture) and anything with X is life’. The question is: is X about current structure or about origin? For water, the essence doesn’t depend on how a sample of water formed; H₂O is H₂O whether from a lab or a spring. By analogy, if life’s essence is an organisational pattern (some think it might be something like ‘a certain kind of self-organising system’), then artificial life with that pattern is life. Cleland uses the water analogy to caution that our current definitions might be like old pre-chemistry definitions of water by properties; once we know the essence (H₂O), it’s clear and origin-agnostic【53†L367-L376】. She and others lean origin-agnostic: life is a natural kind we haven’t fully pinned down, but whatever it is, it won’t depend on whether nature or humans assembled it.
Conclusion on Origin: The weight of scientific opinion sides with organisational criteria: if an entity meets the functional/organisational definition of life, it’s alive, regardless of origin. Therefore, definitions of life should be origin-agnostic to be general. Causal history (abiogenesis and evolution) is hugely important to understanding life, but not strictly necessary to define the category. We can imagine a ‘fire’ that was lit naturally by lightning or by a match – it’s fire either way. Similarly, a cell naturally evolved or intelligently designed is a cell.
That said, origin considerations do inform how we choose definitions. For example, requiring Darwinian capability (as NASA does) effectively encodes that we think life in the universe will be recognised by its evolutionary origin. It might exclude exotic scenarios like a super-intelligent AI (not evolved via natural selection but built) which some might consider ‘life’ in an unconventional sense. Perhaps NASA wasn’t concerned with AI in space, but it shows how context shapes definitions.
In crafting a substrate-independent ontology, it will be important to not bake in Earth-centric origin assumptions. Our ontology should classify entities by their properties (metabolic, autonomous, evolvable, etc.), not by whether ‘Mother Nature’ or a lab produced them.
We will keep this in mind moving forward: for any novel or ambiguous entity, we should ask what it does and how it’s organised more than where it came from, when deciding if it’s life.
4. Life and Consciousness: Ontological Dependencies
Are life and consciousness fundamentally linked? Can one exist without the other? These questions probe the relationship between biological life and mind, a topic at the intersection of philosophy of mind, cognitive science, and artificial intelligence. Here we explore theories like enactivism that root mind in life, and conversely consider whether a non-living entity could possess consciousness (e.g. an AI).
Enactivist and Embodied Cognition View: Enactivism, championed by thinkers like Francisco Varela, Evan Thompson, and Ezequiel Di Paolo, argues for a deep continuity between life and mind. The motto of enactivism is ‘life = mind in action’, meaning the origins of mind are already present in the self-maintaining activity of living organisms. In their book The Embodied Mind and Thompson’s Mind in Life, they contend that the basic self-regulation and sense-making that even a bacterium does is a rudimentary form of cognition【38†L230-L238】. A bacterium swimming up a nutrient gradient is perceiving and acting – in enactive terms, the bacterium enacts a world of meaning (food vs toxin) relative to itself. This approach essentially says that the boundary of a living self (autopoiesis) is also the origin of a perspective, a minimal ‘self’ that has an inside vs outside and can be said to have a sort of proto-mind.
Key arguments from enactivism:
Autonomy is key: A living organism, by being autopoietic, defines what is beneficial or harmful to it. That itself is a sense-making process (interpreting certain chemical as nutrient).
Continuous scale: Instead of a hard line between mindless life (plants, bacteria) and conscious life (animals, humans), enactivists propose a continuum of cognition. Simpler life has simpler, ‘implicit’ cognition; advanced animals have complex, explicit cognition. This is known as the life-mind continuity thesis【59†L13-L21】.
Thompson famously states that the nervous system doesn’t create mind ex nihilo, but expands the mind that is already present in the self-organising activity of life【61†L1-L4】. Specifically, ‘from the perspective of life-mind continuity, the brain or nervous system does not create mind, but rather expands the range of mind already present in life’【61†L1-L4】. So a brain is a way life becomes more flexible and extended in its cognitive capabilities (faster sensing, memory, learning), but even single cells have the basic intentionality (being directed toward maintaining themselves).
Under this view, consciousness (in the minimal sense) might require being alive. If living and sense-making are two sides of the same coin, then perhaps only living systems can be conscious. Some enactivists flirt with a kind of biological panpsychism or biopsychism, where every living thing has at least an iota of subjective experience or at least intrinsic purposiveness. This is controversial, but it counters the AI hype by suggesting machines won’t have genuine consciousness unless they also achieve a living autonomy akin to organisms.
One enactivist argument: consciousness involves having a perspective (a point of view and normative sense of what’s good/bad for the system), and living systems inherently have a perspective (defined by their self-maintenance needs). A robot built just to follow a program might not ‘care’ about anything unless it has an autonomous identity to maintain. This echoes philosopher Hans Jonas, who in The Phenomenon of Life argued that the roots of mind are in the organism’s metabolism – metabolism is the organism’s constant concern, a sort of primitive intentionality.
Does Consciousness Require Life? Some philosophers like John Searle have argued a related point in a different way: Searle’s biological naturalism holds that consciousness is a biological phenomenon that likely only emerges in biological systems (because of specific causal properties of biological neurons)【58†L713-L722】. Searle doesn’t say non-carbon life couldn’t have it, but he’s sceptical of digital computers being conscious because they lack the biological mechanisms. This is a more materialist (and less abstract) version of tying consciousness to life: it’s not life as autopoiesis, but life as in neurons and brains that are a product of biology.
However, mainstream cognitive science tends to allow that an artificial system could, in principle, be conscious if it replicates the right functional or organisational patterns (this is the functionalism view). That would mean consciousness does not require life; it only requires certain cognitive architecture. For instance, Integrated Information Theory (IIT) by Giulio Tononi defines consciousness in terms of integrated information in a system. It does not care if the system is biological or electronic. A sufficiently integrated AI could be conscious by IIT’s measure. Similarly, Global Workspace Theory or other computational theories of consciousness usually imply that if a machine instantiates the same information flows as a brain, it would have an inner experience, regardless of being ‘alive’ biologically. Many AI researchers assume substrate-independence: e.g., a computer simulation of a brain would feel the same as a brain, meaning life (biological viability) is not necessary for mind.
The enactivist counter would be that those theories miss the embodied, animate nature of cognition – an AI might simulate information flow but it’s not grounded in bodily self-maintenance, so perhaps it lacks true agency or meaning in its processing. This is an ongoing debate.
Artificial Consciousness without Life: Let’s consider the hypothetical: a super-advanced AI that is clearly intelligent, maybe claims to be conscious, and behaves as such. It has no metabolism, no self-organising chemistry; it runs on electricity and wires built by us. Do we call it alive? Probably not by conventional definitions (no metabolism, no reproduction, etc.). Do we call it conscious? Possibly yes, if it exhibits behaviours and reports consistent with consciousness and passes tests like the Turing Test, etc. This scenario suggests that we allow a conceptual space for non-living consciousness. Many science fiction scenarios (from Data in Star Trek to sentient AI in many novels) treat that as plausible: machines that are not alive in the biological sense but are persons mentally.
If such AI existed, it challenges any ontology that ties life and mind inseparably. It would be a counterexample to ‘only life can have mind’. Our current world doesn’t have incontrovertible AI consciousness (though some have argued large language models show glimmers of understanding – that’s far from subjective experience evidence). But we can’t rule out that in principle, an engineered artifact might feel.
Dependence vs Independence: So we have two extreme positions:
Strong continuity (dependence): Mind arises from life processes; no mind without life. (E.g., enactivism, some phenomenologists, Jonas.)
Substrate independence: Mind is a computational or functional property that can inhere in non-living systems too, if arranged right. (E.g., most AI research, functionalist philosophers like Daniel Dennett, cognitive theories that treat brain as an information processor).
There’s also a softer middle: maybe minimal primary consciousness (raw feeling) does require being alive (having a biology), but some aspects of cognition could be mimicked by machines. Or perhaps machines could have a different kind of mind but not a full conscious life like we have.
Artificial Life and Artificial Intelligence: There’s an intersection here: artificial life in silico (Section 2.5) tries to create life-like properties. Could those digital organisms develop something like digital consciousness? Unclear – they typically remain simple. Some argue that embodiment in the world is crucial for developing cognition (the AI field of embodied AI believes a robot that interacts with the world will gain more genuine understanding than a disembodied chatbot). This resonates with enactivism that the organism’s sensorimotor coupling to the world is part of how mind arises【38†L230-L238】. If true, then any conscious AI would likely need a body and autonomous interaction – essentially, to be robotic life (some form of artificial life that can act on its own needs). This suggests that perhaps to get machine consciousness, we inadvertently will reinvent artificial life: a robot that’s self-maintaining, exploring, learning – effectively alive – might eventually also be conscious. In that case, it circles back: consciousness still appears in tandem with life-like qualities.
Consciousness in Non-Animal Life: Another angle: does life require consciousness? Clearly not in the sense of high-level awareness – plants and microbes are alive with no sign of conscious experience. But do even they have some extremely rudimentary ‘proto-experience’? Panpsychist views propose that everything might have a fundamental subjective aspect, though that’s speculative. The practical consensus is that most life on Earth is not conscious (only some animals are), so life does not guarantee consciousness. Consciousness appears as a special property of certain complex living systems (brains). So life is a broader category; consciousness is a subset. We might categorise our ontology levels: non-sentient life vs sentient life.
Could there be consciousness without life? That is the more pressing question in AI and perhaps astrobiology if we consider things like planetary minds or machine-like alien probes. We have no confirmed examples of non-life consciousness, but we also have no proof that it’s impossible. It remains a theoretical possibility which many find plausible. If it happens, our definition of life might expand or we might explicitly separate the concepts: treat ‘life’ as biological and ‘artificial consciousness’ as a separate phenomenon. For legal and moral purposes (as next section touches), that separation might be important – e.g. AI rights might be based on consciousness not on being alive per se【58†L714-L722】【58†L727-L735】.
Summary: Enactivist theories provide a philosophical grounding to say the phenomena of life and mind are deeply entwined – a view supporting that a full ontology of life might inherently include cognitive categories (life that is also cognitive). They would likely argue that a truly autonomous, adaptive system – life – naturally develops cognitive features, so any alien life we find might have some degree of ‘mind’. On the other side, AI research opens the possibility of mind without the usual hallmarks of life.
In constructing our layered ontology, we will likely treat ‘conscious, intelligent life’ as a layer on top of ‘basic life’. It’s plausible to have life without consciousness (bacteria) and, theoretically, consciousness without standard life (AI). The framework should allow distinguishing autonomous living systems from autonomous living systems that are also conscious, and also consider non-living conscious entities if any (perhaps as a boundary case, like ‘artificial persons’).
Thus, while life and consciousness often coincide (in us, in animals), conceptually one is not strictly necessary for the other. But in the history of Earth, consciousness has only emerged within living beings. We’ll revisit this when we map the ontology levels: minimal life vs life with mind.
5. Extraterrestrial and Synthetic Contexts
When we expand our view beyond Earth and beyond traditional biology, definitions of life face new challenges and test cases. This section examines how our concepts apply (or strain) in artificial life platforms, astrobiology (extraterrestrial life), and scenarios of encountering intelligence that did not evolve as we did.
5.1 Artificial Life in Silico and Synthetic Biology
We earlier discussed digital organisms and ALife experiments from a theoretical perspective. Here, let’s consider how we identify or classify such entities as life. In platforms like Avida or Tierra, researchers observe populations of self-replicating programs that mutate and evolve complexity【51†L293-L300】. By many functional measures, these behave like microbial life: they have a ‘birth’ (spawned by a parent program), a ‘genome’ (their code), can undergo lethal mutations, adaptive improvements, compete for limited CPU cycles (akin to energy), and speciate into different types (some even develop parasite-parasitoid relationships as noted).
Do we call them alive? The community is split. Some say they are alive in the computer – a new form of life. Others see them as simulations that, while interesting, lack something fundamental (like physical embodiment or metabolism). The question reveals what criteria one prioritises. If Darwinian evolution is the gold standard, these clearly pass – so by NASA’s definition one could argue they demonstrate life processes【39†L1-L5】, just in a different substrate. If metabolism in a physical environment is critical, they fail that (no real chemistry, just virtual).
In practice, most refer to them as digital life or artificial life forms and treat them as analogies to life to test evolutionary theory, rather than insisting they literally have the same ontological status as a bacterium. However, as ALife systems get more complex (imagine one day a digital ecosystem with complexity rivaling an ecosystem of real organisms), the argument for their life status strengthens. At the least, they broaden our perspective of what ‘life-like processes’ can be.
Synthetic Biology offers another context: the creation of novel biological systems that may blur boundaries. Researchers have engineered synthetic cells with genomes stitched together from different sources, created minimal cells with as few genes as possible, and even semi-synthetic organisms with extra letters in their DNA alphabet. If they function and reproduce, we call them alive – they are just modified life. But as this progresses, we might see life forms with fundamentally new properties (organisms with an alternate biochemistry, or protocells that have no DNA but some other hereditary mechanism). For example, there have been experiments with XNA (artificial nucleic acids) replacing DNA. If an XNA-based cell line could sustain itself and evolve, it would be a new form of life – synthetic, but meeting origin-agnostic criteria. Scientists would undoubtedly call it life, because it’s doing what life does (metabolising, evolving), just with different hardware.
One interesting scenario is protocells – simple droplets or vesicles that show some life-like traits. Researchers have made lipid vesicles that grow and divide when fed simple chemicals. Some incorporate RNA replication inside. These are attempts to create life de novo. During these experiments, at what point do they declare, ‘We have created life’? It might be gradual. Initially you have just chemistry, then maybe a system that can self-maintain (autopoiesis achieved?), then one that can replicate, then one that can evolve. As each piece falls into place, more of the community will be comfortable calling it life. The ultimate goal is a life form in the lab with no biological parent – a completely synthetic origin. That will be a milestone confirming that life is a reproducible phenomenon of chemistry, not a singular miracle.
So in synthetic contexts, we see the progressive realisation of origin-agnostic life: from tweaking existing life (which is clearly alive) to conjuring entirely novel life. Each step forces clarification of definitions: e.g., is a minimal cell with 200 genes still life? Biologists generally say yes, if it can grow and divide, certainly. How about something even simpler – say a self-replicating ribozyme in a lipid container? Perhaps that crosses the threshold, though it’s debatable if it’s ‘organism’ or just a life-like chemical system.
5.2 Astrobiology and Extraterrestrial ‘Life’ Considerations
Astrobiological Definitions and ‘Weird Life’: In searching for life beyond Earth, scientists have had to broaden concepts and confront the possibility of life with alien chemistries. NASA’s working definition (life as a self-sustaining chemical system capable of Darwinian evolution) was crafted to be general【39†L1-L5】, yet it still assumes ‘chemical’ and ‘Darwinian,’ reflecting what we know from Earth. What if extraterrestrial life doesn’t neatly fit these criteria? One concern is ‘weird life’ – life that might use different solvents (e.g., liquid methane on Titan instead of water), different elemental backbones (silicon instead of carbon has been speculated), or even different molecular mechanisms (perhaps no DNA/RNA). A report by the U.S. National Research Council coined ‘weird life’ to encourage open-mindedness about forms that violate Earth norms.
Astrobiologists often emphasize agnostic biosignatures – signs of life that don’t presume specific biology. For instance, any life will likely create chemical disequilibria in its environment (consuming some molecules, excreting others). The detection of a persistent atmospheric disequilibrium (like oxygen with methane on an exoplanet) is considered a strong biosignature because it’s hard to sustain without something replenishing the gases – presumably metabolism. This approach ties back to a thermodynamic definition (life fights equilibrium) in a broad way.
Past missions illustrate how definitions guide detection: The Viking landers (1976) on Mars carried experiments explicitly designed under certain life definitions. The Labeled Release experiment put nutrients labelled with radioactive carbon on Martian soil to see if any metabolism would metabolise them into radioactive CO₂ – a direct test for metabolic activity【62†L21-L24】. Viking also had a gas chromatograph-mass spectrometer (GC-MS) to detect organic molecules. The results were paradoxical: the metabolism test came up positive (gas was released that looked like CO₂ bursts), but the GC-MS found no organic molecules in the soil【62†L15-L23】【62†L17-L24】. By the definition ‘life uses organic (carbon) chemistry’, the lack of organics suggested no life, leading most to conclude the metabolic signal was chemical in origin. If one had defined life purely as metabolic activity, one might have declared life on Mars. But the broader chemical criteria overrode it. This shows how our expectations (life = organic compounds + metabolism) formed the interpretation framework【22†L153-L162】【22†L173-L181】. Even today, debates about Viking data illustrate how no single experiment can definitively prove life without a definition to benchmark against.
Broadening the Notion of Biosignature: Suppose we encounter something like the oft-cited example of silicon life. Silicon-based life might not leave the same traces (it might not produce organic biomarkers; it could, say, excrete silica). If our life-definition is too narrow (e.g., requiring carbon or water), we might overlook such life. Thus, astrobiology tends to focus on functions (does it consume energy? maintain a redox gradient? produce waste?) rather than ingredients. NASA has invested in instruments that detect general patterns – for example, the CHNOPS elements (carbon, hydrogen, nitrogen, oxygen, phosphorus, sulphur) because Earth life uses them, but also in instruments like the ‘Life Marker Chip’ that looks for broad classes of molecules, or microscopes to see cell-like structures.
Detection vs Definition: Practically, a provisional definition is needed to design a search, but astrobiologists remain aware it might be incomplete【53†L339-L347】【53†L367-L376】. As Cleland argues, we might only truly define life once we discover a form fundamentally different from Earth’s – then we can compare and find the true common essence (if any)【53†L367-L376】. Until then, the strategy is to use multiple criteria: chemical complexity, evidence of reproduction or metabolism, etc., and look for converging evidence. For instance, finding complex organic molecules is suggestive; finding anomalous atmospheric gases is also suggestive; finding both together with perhaps microscopic structures would likely clinch the case for life.
‘Discovery’ scenarios: If we find fossilised microbes on Mars (like alleged microfossils in meteorite ALH84001), deciding if they were life involves checking for morphology plus mineral patterns indicating metabolism. The definitional question arises: is a fossil ‘life’? No, it’s evidence of past life. So astrobiology also distinguished living vs once-living. The concept of life thus includes remnants and traces (where the definition of life guides what counts as a legitimate trace).
First Contact with Intelligence: Now consider ‘contact’ in the sense of encountering extraterrestrial intelligences. If we receive a radio signal from an alien civilisation, we’ve detected alien intelligence, but have we detected alien life? Presumably yes – we assume behind the signal there are living beings (or at least, living beings created the transmitters). But it’s conceivable that advanced aliens might be post-biological: perhaps their civilisation is run by AI machines that their biological ancestors built (a scenario sometimes discussed in SETI circles). In that case, the first intelligence we encounter might not be alive by our definitions (if they are purely robotic). This blurs the line: we would still culturally say ‘we found alien life’ because we found an alien civilisation. But if that civilisation is machine-based, scientists might debate whether to call it life or an ‘artificial system’. Most likely, we would broaden ‘life’ to include them in discourse because they have continuity with biological progenitors and exhibit all traits of life except organic biology. This is analogous to how we treat hypothetical future AI on Earth – we may end up considering them a continuation of human life in a new form.
Communication and Definition: If we ever face communicating with a radically different intelligence, defining life becomes more than academic. Imagine trying to explain to alien beings what we consider ‘life’. They might have different categories. (Perhaps they consider stars as living conscious beings in their philosophy!) It might sound far-fetched, but it underscores how any contact scenario forces us to confront our Earth-centric concepts. For instance, if aliens have a silicon biochemistry, initially we might not even recognise their habitats or food sources as ‘biological’. Only through exchange of information would a shared concept of life emerge.
In summary, extraterrestrial and synthetic contexts push us to craft flexible yet grounded definitions. Astrobiology teaches us to identify life by patterns and effects (disequilibrium, complexity, reproduction) rather than specific substances. And encountering non-terran intelligence reminds us that life and technology may intertwine. The definition of life must be able to accommodate life that is not as we know it, or else we risk missing it when it stares us in the face【53†L367-L376】.
6. Normative and Legal Implications
How we define life is not only a scientific or philosophical matter – it carries ethical and legal weight. Concepts of life inform debates on abortion and end-of-life, the moral status of animals and ecosystems, and emerging issues like the rights of artificial intelligences or engineered organisms. This section examines how definitions of life intersect with normative questions (what we value) and legal frameworks.
Beginning-of-Life Debates: A classic moral and legal debate is when does human life begin? Those opposed to abortion often argue that a human life (and personhood) begins at conception – here relying on a biological definition (the formation of a new zygote with a unique genome is the start of a new life) as a basis for moral status. Others point to later milestones: implantation, the start of a heartbeat, the capacity for pain, or birth. Developmental biology shows no single ‘instant’ when a non-living matter becomes definitively living – the sperm and egg were alive too. So biologically, the zygote is certainly alive (it’s a living cell), but the crux is whether it is a life in the sense of an individual human being. Definitions of life alone don’t settle this, because the dispute hinges on the concept of personhood and rights. As the SEP notes, developmental biology offers no uncontested break-point; ‘life’ is a continuum【58†L707-L715】【58†L717-L725】. Nevertheless, law often needs a line (e.g., laws that prohibit abortion after X weeks implicitly decide when the fetus’ life becomes protectable). Some countries use viability (when the fetus can live outside the womb) as a legal threshold – interestingly, a definition tied to autonomous life (the ability to self-sustain). So autonomy again comes in: a fetus becomes ‘a life of its own’ when it no longer completely depends on the mother’s body. Here we see a blend of biological and functional definition influencing law: self-sustained life as a criterion【58†L727-L735】.
End-of-Life and Defining Death: At the other end, modern medicine blurred the line between life and death. Is a patient on a ventilator with no brain activity alive? Traditionally, cessation of heartbeat and breathing meant death (the body no longer maintains itself). But machines can keep the body’s cells functioning. Enter the concept of brain death – now defined in many jurisdictions as the death of the person, even if the body is still warm and oxygenated. This was a definitional shift: ‘life’ for a human is no longer just a matter of metabolic activity; it requires a functioning brain (at least brainstem). In other words, the legal definition of death invokes the absence of the integrated organismic functioning governed by the brain【58†L709-L717】【58†L727-L735】. This shows an interesting case where the definition of human life (or death) became explicitly tied to certain functions (consciousness, or irreversible loss of it) rather than mere cellular life. It underscores that ‘life’ in a moral sense often means human life with qualities of agency or awareness.
Life and Moral Considerability: Philosophers differentiate life from sentience and personhood【58†L715-L724】. Being alive might be seen as a prerequisite for having interests or moral rights, but not a guarantee. For example, plants are alive yet most ethicists do not attribute moral rights to plants (aside from their ecological value). Bacteria are alive but killing bacteria is morally trivial in most views. Thus, mere life is not a sufficient condition for strong moral status【58†L717-L725】. Usually, life is considered morally significant when paired with other properties: capacity to suffer (sentient animals), autonomy and rationality (persons), or being part of human community (human life often gets special status). However, there are philosophical positions like biocentrism or deep ecology that argue life itself, even non-sentient, has value. Aldo Leopold’s land ethic for instance says the biotic community’s health is the ultimate value (giving some weight to all life forms)【58†L717-L725】. And Albert Schweitzer’s ethic of ‘reverence for life’ holds that we should have awe and respect for life as such. These remain minority views in ethics【58†L717-L725】, whereas mainstream ethics elevates conscious life.
Nonetheless, legally we do encode special status for life broadly: for example, laws against cruelty to animals, which though based on suffering, also often speak of respect for life. Environmental laws protect endangered species – partly an ecological rationale, partly an intuition that wiping out a form of life is bad.
Artificial Intelligence and Synthetic Beings: As AI and bioengineering advance, who (or what) gets included in the community of morally considerable entities is a pressing issue. Suppose we have an AI that is demonstrably conscious and autonomous (a hypothetical sentient AI). By traditional definitions it’s not ‘alive’ (no organic metabolism). Does that matter for its moral status? Likely not – most would argue if it’s conscious and can feel, it deserves ethical consideration (this falls under personhood or sentience, not life per se). There have indeed been proposals and discussions about AI rights or ‘robot rights’. For example, the European Parliament in 2017 considered a report proposing a legal status of ‘electronic persons’ for autonomous robots, which sparked much debate. The rationale was to assign responsibilities and possibly rights to AI entities. This indicates a scenario where something that isn’t alive biologically might gain a legal status analogous to a living person – a radical shift in how we anchor rights (shifting from species membership or life status to cognitive capacity and role).
‘Synthetic personhood’ might also refer to bio-engineered beings. If we create a humanoid organism in a lab (say through synthetic biology or cloning with enhancements), it is biologically alive and human-like – presumably it has rights as a person if it has normal human mental faculties. But some fear discrimination: e.g., would a cloned human or a genetically edited human have the same rights? Legally and ethically, we say yes, they are human life. This shows that origin doesn’t matter to moral status if the properties are the same – echoing our earlier scientific conclusion that origin shouldn’t matter to definition of life.
Consider also chimeras or part-human part-animal organisms that biotech might produce for research (e.g., human neural cells in animals). If a chimera developed significant human-like cognition, would it get rights? Here ‘life’ is not in question (it’s alive biologically), but being a hybrid challenges our definitions of person. This is on the frontier of law and ethics now.
Legal Use of ‘Life’: The term ‘life’ appears in law (e.g., ‘right to life’, ‘life sentence’, definitions of murder require a living victim, etc.). Usually, law relies on commonsense and medical definitions for life. The notion of ‘brain death’ being legal death is one example of law adopting a refined definition【58†L709-L717】. Another legal concept is ‘viability’ in abortion law (the fetus’s ability to live outside womb) as a threshold, which we discussed – essentially a development-based definition of independent life.
Laws concerning end-of-life (e.g., assisted suicide or turning off life support) often hinge on whether the individual is still considered alive in a meaningful sense or if further sustaining is futile. So our definitions of death (irreversible loss of function) come into play.
Moral status of non-human life: We have gradually expanded moral consideration to some non-humans: higher animals have welfare protections. Here, being alive is necessary (we don’t talk about the ‘welfare’ of a robot vacuum), but what actually triggers moral concern is sentience. As science informs us which creatures are sentient (for instance, many countries now acknowledge great apes, cetaceans, sometimes even octopuses as sentient beings deserving protection), the sphere of morally considerable life expands. Definitions of life per se are less debated here – it’s obvious those animals are alive – but underlying it is a hierarchy of life: intelligent or sentient life is valued more.
Implications of Defining AI or Uploads as Life: If one day we acknowledge machine consciousness, we might start describing those machines as a form of life (perhaps a ‘machine life’). This could have legal implications: we might, for instance, outlaw their unjust termination similar to how destroying an animal is regulated. Or we might grant them personhood (which brings rights and responsibilities). Some futurists talk about ‘post-biological life’ – a term that explicitly uses ‘life’ for non-biological entities that have taken over the evolutionary process (AI descendants of human civilisation). If we adopt such language, law and ethics will follow with new frameworks. Bioconservatives, as the SEP mentions, often resist granting such status, arguing that only natural, biological humans should hold certain rights【58†L739-L748】【58†L749-L757】. But transhumanists and AI ethicists are pushing for a capacity-based approach (if it thinks and feels, it counts).
Synthetic Biology and Rights: Another angle is whether creating new life forms gives them moral status. For simple synthetic microbes, likely not (we treat them as we treat bacteria). But what if someone engineered a human-level intelligent species? We would face granting them full rights (as they are alive and person-like) or face a new form of discrimination. Science fiction explores such themes (e.g., the rights of genetically engineered beings or uplifted animals).
Eco-centric and Legal Personhood for Non-humans: Interestingly, some legal systems have started granting legal personhood to non-living things like rivers or ecosystems (e.g., the Ganges River in India, or rights of nature movements). This stems from indigenous and environmental views that see certain natural entities as living or spiritually important. While not about biological life per se, it shows a legal broadening of what counts as a ‘subject’ rather than object. The term ‘life’ is sometimes used metaphorically in these cases (Mother Earth as living). It’s a legal innovation acknowledging intrinsic value in what classical definitions would call non-life.
In conclusion, how we define life has downstream effects on our value system and laws: It frames who is protected, who has rights, and how we justify those. Historically, our moral community expanded from just certain humans to all humans, then to some animals, potentially someday to artificial intelligences – a trajectory of expanding the definition of what entities deserve moral respect. While ‘life’ itself is only one factor (consciousness, personhood, etc., often carry more weight), the language of life is powerful in moral discourse. ‘Sanctity of life’ is a phrase used to argue against euthanasia or abortion, leveraging a broad, perhaps ambiguous concept of life’s value. Clarifying the components (biological life, sentience, etc.) can lead to more nuanced positions: e.g., valuing the quality of life (conscious experiences) over mere biological life when deciding to end life support.
As we create new forms of life and intelligence, society will continually renegotiate these definitions. Legal systems may eventually encode multi-tiered statuses: e.g., a distinction between biological life forms and AI entities, or natural vs synthetic life – unless we deliberately choose an inclusive definition of life that focuses on relevant capacities. For now, being ‘alive’ is a baseline for entry into moral consideration, but it is rarely sufficient without other qualities【58†L717-L725】. The ontology of life we develop isn’t just academic – it could guide how we treat new beings ethically and legally.
7. Epistemic Consequences and Instrumental Targets
Finally, we consider how our working definitions of life shape the epistemic strategies of science – how we design instruments, experiments, and interpret data in fields like astrobiology, artificial intelligence, and synthetic biology. In essence: what we look for is determined by what we think life is. And conversely, the evidence we gather feeds back into how we refine the concept.
Astrobiological Instruments and Life-Detection Protocols: As noted, NASA’s definition influenced mission design. If life is conceived as chemical and metabolic, you send chemistry labs (like Viking’s GC-MS and labelled release) to Mars【62†L21-L24】. If you highlight structures, you send microscopes or cameras (some missions carried microscopes to examine soil grains for cell-like shapes). If evolution is key, you might look for populations of organisms and genetic diversity, which is hard to do remotely. The upcoming Europa and Enceladus missions focus on detecting organic molecules and possible cells in plume samples – again assuming Earth-like biochemistry as a sign of life’s presence. Our biases (life needs organic molecules, life uses water) guide these choices. They are reasonable given one data point (Earth), but they are educated guesses. We might miss life that doesn’t leave the expected traces.
In exoplanet astronomy, when we search for atmospheric biosignatures, we specifically target gases like oxygen or methane together because on Earth their coexistence implies life (oxygen from photosynthesis, methane from microbes/ruminants)【25†L19-L27】. This strategy is directly informed by a definition of life involving global metabolic impact on environment. If we defined life purely as something that moves or grows, we might instead focus on direct imaging of vegetation or seasonal changes – which is another proposed approach (looking for seasonal colour changes on a planet might indicate plant life). The point is, the conceptual model of life directs the technological approach. A broader definition calls for broader instrumentation. For example, to remain open to non-oxygenic life, astrobiologists also consider looking for other disequilibria (like an excess of certain hydrocarbons, or odd ratios of isotopes that could indicate biological fractionation).
Interpreting Ambiguous Data: When scientists get ambiguous results, the definition in use frames the interpretation. We saw Viking’s example – positive metabolism, negative organics led to ‘probably no life’ because the working definition required organics as a component of life【62†L17-L24】. If we had instead considered that life might exist with very minimal organics or hide organics quickly, we might have interpreted differently. Another example: methane on Mars. Methane was detected in Mars’ atmosphere in plumes. Life? Possibly methanogenic microbes. Or geochemistry? The debate goes on. Our definition of life as something that can produce methane (some microbes do) made this a tantalising hint, but since methane can come from non-life, it’s not conclusive. So scientists developed criteria (like looking for isotopic ratios of carbon in methane – life tends to prefer lighter isotopes, so biogenic methane is isotopically lighter). That criterion comes from knowledge of Earth life’s quirks.
Similarly, in evaluating ancient Mars, scientists look for microfossils and biogenic textures in rocks. They have defined what biosignatures count: certain stromatolite shapes, certain mineral patterns (like concentration of organics associated with cell-like spaces, or chiral biases – life tends to use only left-handed amino acids, so a strong chirality signal in molecules would be a hint of life’s involvement). Each of these is rooted in a prior definition or at least assumption about what life does: build cell walls, use homochiral chemistry, etc. As we broaden definitions, we add more possible biosignatures (but also risk more false positives if non-life mimics them).
Cleland’s Definitional Skepticism and Practice: Cleland and others suggest using ‘tentative criteria’ instead of rigid definitions【53†L367-L376】. For example, NASA might use a checklist: does the sample show complex organic molecules? Does it exhibit metabolism (in an experiment)? Does it have structures reminiscent of cells? None of these alone defines life, but together they increase confidence. This pragmatic approach is an epistemic strategy shaped by the difficulty of defining life: basically, accumulate various signs until the balance of evidence suggests life. It acknowledges that any single definition can mislead or be too narrow.
In Artificial Life and AI Research: In AI, researchers don’t typically ask ‘is it alive?’ – they ask ‘is it intelligent or capable?’. However, definitions of life do play a role in Artificial Life experiments: e.g., in Avida, how do they measure if the digital organisms are evolving complexity? They use analogues of biological fitness. In robotics, some are trying to create self-reproducing robots – essentially testing the line of life. One famous project was NASA’s idea of autonomous rovers that could repair and replicate using materials on Mars (inspired by von Neumann machines). If accomplished, would they consider those rovers a form of life? Possibly in a research sense. Here, the design of the experiment (create a self-replicating machine) is explicitly motivated by a definition of life as self-replication. Another example: researchers studying evolution of intelligence sometimes use digital life simulations – they define an environment and ‘organisms’ with neural nets that evolve to solve tasks. The initial conditions (mutation rate, selection criteria) are all based on evolutionary definitions of life. If one defines life as capable of open-ended evolution, a big open question is: have we achieved open-ended evolution in a computer? Some say not yet – digital evolution tends to plateau. This then informs theories of what is missing (maybe physical embodiment or energy constraints that organisms face, etc.). Thus, the definition of life as open-ended evolution becomes an instrumental target itself – ALife researchers aim to fulfil that in their systems to declare success in creating lifelike dynamics.
Synthetic biology instrumentation: If our goal is to create life from scratch, how do we judge success? We need criteria (much like NASA’s, but for the lab): e.g., does the protocell have a metabolism and grow? Does it divide and propagate information? Researchers use assays to detect those functions (dye that indicates metabolic redox reactions, microscopy to see division, etc.). In one sense, building life is the ultimate test of understanding the definition. As often said in origins research, ‘definitions of life are hypotheses for experiments’. If you think life = replication + metabolism, you try to build a system with those and see if it behaves lifelike. If it falls apart, maybe your definition missed something (like error correction or compartmentalisation). For example, those who defined life as needing a container (membrane) insisted on protocells (vesicles) as crucial. Others who think gene replication is the crux worked on RNA replicators first, leaving membranes for later. Each experimental program is guided by a tacit definition of what the essential components of life are. Over time, success or failure of these experiments feeds back: perhaps they find metabolism alone can’t sustain without genetic information – implying metabolism definitions were insufficient, and strengthening a hybrid definition.
Filtering and bias: One epistemic pitfall is confirmation bias via definition. If our concept of life is too rigid, we might filter data incorrectly. Imagine an alien probe flew by Earth in Archean times (3 billion years ago) when there was life but no oxygen atmosphere. If those aliens defined life as ‘oxygen-producing ecosystem’, they would conclude Earth has no life (false negative)【53†L369-L376】. By analogy, if we have a too-narrow view, we might miss life that doesn’t announce itself in expected ways. This is why NASA, for instance, remains cautious about declaring negative results. The absence of certain expected biosignatures is not proof of absence of life—just absence of known life signs. The Viking experience taught this humility. It’s also why Mars sample return (bringing Mars soil to Earth for thorough analysis) is pursued; maybe only a combination of tests on Earth can reveal subtle life processes that a few predefined experiments could not.
Developing Technology from Definitions: On the flip side, how we define life can inspire new technology. For example, if we define life as ‘makes order from chaos’, could we build a device that detects unusual pockets of order? Some have proposed entropy detectors or complexity meters. If life is defined by information patterns, perhaps SETI should search not just for radio signals but any anomalous information structures in signals (some have suggested looking for non-random, encoded patterns as a sign of intelligence or even life’s handiwork at any scale). The concept of assembly theory by Sara Walker and Lee Cronin tries to quantify how complex a molecule is (the idea being that very complex molecules are unlikely to form without life). They define an ‘assembly index’; instruments in the future might be able to measure assembly indices of molecules on other planets to hint at life’s presence. This is a direct case of definition (life makes complex molecules) leading to an instrumental approach【21†L19-L27】.
Data Interpretation in AI/Synthetic Bio: Consider a more down-to-earth case: distinguishing engineered biological systems from natural life. As we synthesize genomes, there might be future need for detectors that tell if a sample is naturally evolved or artificially designed (for biosecurity). This is like detecting life, but all are life; the question is origin. Still, it relies on assumptions about patterns: engineered life might have ‘signatures’ (like certain codon usage biases or watermarks in the DNA). That’s a reverse problem: detecting ‘designed life’ vs ‘natural life’. It again shows that our concepts (natural life vs synthetic life) lead to data analysis methods.
Definitional Pluralism in Practice: Some scholars advocate definitional pluralism – use multiple definitions for different contexts【53†L339-L347】【53†L347-L350】. In research, this might mean one definition for lab synthesis work (e.g., focusing on replication and metabolism), another for field astrobiology (focusing on biosignatures and disequilibrium), and another in philosophy (focusing on autonomy and agency). Each yields different ‘instruments’: lab apparatus, space probes, conceptual arguments. While this is pragmatically useful, it can cause fragmentation – researchers might talk past each other. To mitigate that, an ontology that maps how these different conceptions relate (which is exactly our aim here) can serve as a unifying reference. For instance, an ontology can say: metabolism-first protocell in lab = achieving autonomy layer; Viking’s search = targeting metabolism and byproducts; SETI = targeting cognitive life by radio signals. All are attempts to find life, but at different layers of the ontology.
In summary, the way we conceptualise life has instrumental consequences: it determines what instruments we build, what measurements we consider meaningful, and even how we allocate scientific effort (e.g., if one believes life is common and defined by easy-to-form autocatalytic chemistry, one might invest more in missions to many planets; if one believes life requires rare events like the origin of coding, one might be more pessimistic or look only at very Earth-like planets). It also shapes how we interpret surprises. A flexible, layered definition of life can make us more receptive to unexpected phenomena, whereas a narrow definition might make us dismiss anomalous data too quickly.
As we move forward with new telescopes scanning exoplanets, rovers drilling into Mars, lab experiments creating pseudo-cells, and AI potentially nearing life-like complexity, having a well-thought-out ontology of life will be crucial. It serves as both a guide (what to look for) and a checklist (to evaluate if what we found qualifies).
In the next section – the synthesis – we will propose such an ontology, integrating the insights from all these domains into a structured framework that can be applied to classify and evaluate ambiguous or novel entities as life or non-life, and at what tier of complexity.
8. Synthesis: Toward a Layered Ontology of Life
Bringing together all the threads above, we can outline a substrate-independent ontology of life – a conceptual map that identifies different levels or aspects of life, from the minimal criteria to the most complex manifestations. The goal is to create a flexible framework to evaluate new entities: whether they be unconventional microbes, artificial organisms, or alien intelligences. We propose an ontology with (at least) four major layers:
(i) Minimal Life Systems: At the base are systems that exhibit some but not all canonical features of life. These are borderline cases or precursors – for example, a virus (which has information and evolution but no autonomy in metabolism), or a simple autocatalytic chemical network (metabolism-like self-maintenance but no genetic program). This category captures entities that are partially alive. They might depend on other living systems (like viruses do on hosts) or fall short of full self-sustainment. In our ontology, we label these minimal life or proto-life. They are crucial, because they force us to clarify boundaries. Many debates (are viruses alive?) land here: we acknowledge viruses are minimally alive in the evolutionary sense but not autonomous. Similarly, lab-made protocells that can grow but not yet evolve would sit here: they have some life properties【53†L323-L331】.
(ii) Autonomous Living Organisms: The next layer consists of entities that achieve individual autonomy – roughly corresponding to what Maturana & Varela call autopoietic units【38†L184-L192】. These systems are self-bounded and self-maintaining, carrying out a metabolism and homeostasis. A single bacterial cell, a plant, an animal – any individual organism that can sustain itself (at least for some time) independent of others is in this category. The necessary condition to be here is organisational closure: the system’s components and processes produce and regulate each other, defining a unit【38†L184-L192】. Reproduction is not required to be an organism (a mule or a sterile worker ant is still an autonomous living being). What is required is that it arose from a lineage that did reproduce, or could in principle be part of a reproductive cycle. Autonomous organisms are what most definitions through history had in mind – Aristotle’s self-moving beings, or NASA’s ‘self-sustaining chemical systems’【39†L1-L5】. They possess what we might call minimal complete life: metabolism, growth, response to environment, and (usually) reproduction. In our ontology, this layer distinguishes life proper from non-life clearly: if it’s autonomous and self-maintaining in its environment, it’s alive. (Viruses fail here, so they stayed at minimal). Most Earth life we know (except viruses, prions, etc.) qualifies for this layer. This layer is substrate-neutral: a robot or AI could in theory be designed to be autonomous in this way (e.g., a self-repairing robot that gathers its own energy). If that happens, we might classify that machine as an autonomous life-form by function, albeit an artificial one.
(iii) Evolvable Life (Population-Level Life): This layer adds the dimension of heredity and evolution. While an individual organism doesn’t evolve by itself, it is part of an evolving lineage or population. Here we consider life as a population-level process – the capacity of a group of entities to undergo open-ended adaptive evolution. All known Earth life falls in this category, since we share common descent and continue to evolve. This layer is where NASA’s Darwinian clause sits【39†L1-L5】. Why make it a separate layer? Because conceivably one could have an isolated organism or artificial life form that meets the autonomous criteria but is singular (it doesn’t replicate). We might call it alive, but it wouldn’t contribute to evolutionary life. Layer (iii) life, by contrast, exists in the context of a biosphere or ecosystem where variation, competition, and selection occur. It’s the difference between a sterile lab organism and a species in the wild. The necessary condition here is the presence of a genetic/informational system that can mutate and be passed on. Sufficiently, if an entity (or entity type) can undergo Darwinian evolution, it belongs to this layer【39†L1-L5】. This is substrate-independent too – we could have evolving digital organisms or self-modifying AI that form populations. If they have ‘descent with modification’, they fit here as evolvable life. An interesting sub-point: this layer includes the concept of ecology. Life does not evolve in isolation; it interacts in communities. So, a fully realised ontology of life might even extend layer (iii) to (iii b) life-as-ecosystem – acknowledging emergent life-like properties at the ecosystem level (like Gaia hypothesis ideas of the whole biosphere having life-like regulation). But for simplicity, we can keep evolution and ecology in one layer of life as an adaptive system.
(iv) Cognitive or Conscious Life: The top layer in our scheme is life that has not only biological autonomy and evolution, but also some degree of sentience, consciousness, or mind. This distinguishes beings that have an experiential inner world or at least complex information processing enabling flexible behaviour. Many animals (certainly mammals, birds, cephalopods, etc.) fall here. This layer is optional in the sense life can exist without consciousness (plants are alive and not conscious in any ordinary sense). But it’s a crucial ontological category, because historically we often implicitly rank life forms by cognitive capability (the old ‘great chain of being’ placed humans atop due to reason, etc.). Rather than ranking by value, we here simply mark it as a different qualitative dimension of life. Necessary condition: some form of centralised information processing that results in behaviour decoupled from immediate stimuli – in short, a nervous system or analogous system enabling learning, perception, and response. Sufficiently, if an organism communicates, solves problems, seeks goals, and (in higher degree) reports subjective states, it’s in this category. This could also include hypothetical artificial consciousness – e.g., an AI that is not biologically alive but thinking. Those would be conscious, but not autonomous-biological; they might be a case of ‘mind without life’. However, if we encountered that, we might extend the ontology or consider them an offshoot: artificial persons. For now, layer (iv) assumes an underlying living organism that developed consciousness (like humans). Some enactivists would argue layer (iv) is not really separable – mind is inherent from layer (ii) up – but here we mean complex, likely conscious mind. In practical terms, this layer covers beings that could be considered agents with interests and perhaps rights (tying back to section 6). It is why we treat killing a mouse differently than killing a microbe; the mouse is alive (layer ii), part of an evolving species (iii), and also a sentient individual (iv).
These four layers are not strict compartments but cumulative aspects. An adult human spans all four: our biology is autonomous (ii), our population is evolvable (iii), and we are conscious (iv), and we even harbour minimal life in us (our cells, and viruses we carry). A bacterium spans (ii) and (iii) but not (iv). A virus spans (iii) (since viruses evolve) but arguably not (ii) on its own (no metabolism), so it’s at layer (i) with a foot in (iii) by proxy of hijacking other life’s machinery. An advanced AI (if conscious but not self-maintaining biologically) could be (iv) without i-iii, though if it’s self-maintaining in hardware it might achieve (ii) in a machine sense. This layered view avoids a one-size definition; instead, it asks: which properties of life does X exhibit?
Decision Framework for Ambiguous Cases: Using this ontology, we can formulate a decision process when encountering an unknown entity:
Autonomy Check (Layer ii): Does the entity maintain itself by active processes, resisting entropy locally? Does it have a boundary and internal processes (metabolism or equivalent) that sustain it? If yes, it qualifies at least as an individual organism (alive in the fundamental sense)【50†L205-L213】. If no, check next:
Replication/Evolution Check (Layer iii): Does the entity replicate or is it a product of a lineage that evolves? If yes (but it failed autonomy), it might be a dependent life form (like a virus) – classify as minimal life (i). It has some life aspect (genes/evolution) but not full autonomy. If it fails both autonomy and replication, it’s likely non-life (e.g., a crystal that grows but no encoded information or internal regulation – just physico-chemical growth).
Metabolism vs Replicator trade-off: We might explicitly decide a threshold: either metabolism or Darwinian replication is required as a minimal criterion. This catches viruses (no metabolism but replication) and hypothetical pre-cells (metabolism but no replication). Our ontology would include both as minimal forms, recognising a boundary zone. If an entity has neither metabolism nor information heredity, it’s not alive. If it has both, it’s definitely alive (and falls in (ii)+(iii)). If it has one, it’s borderline (i).
Consciousness Check (Layer iv): If the entity passes as alive (ii or iii), we then ask does it show signs of consciousness or sophisticated cognition (tool use, communication, etc.)? This determines if it also occupies layer (iv) in our ontology. For example, if we discovered an alien jellyfish-like creature that metabolises and reproduces (clearly alive) and also solves complex problems or signals in an intelligent way, we’d put it in (iv) as conscious life (and perhaps accord it higher ethical regard). If we found only microbial life on Mars, it’s layer (ii) and (iii) but not (iv), so our dealings with it ethically would be different (still significant scientifically!). For artificial entities: if a computer program doesn’t meet (ii) or (iii) but seems conscious, this framework flags it: we have something at (iv) without the lower layers – essentially ‘AI consciousness.’ We might then debate if our definition of life should extend to include it as a form of life. (Perhaps one would call it artificial life at informational level but not biochemical life).
Novelty and Ambiguity: For truly novel cases, the framework acts more like a questionnaire: Does it consume something and grow? Y/N. Does it reproduce or show variance? Y/N. Does it have internal chemistry or just surface reactions? Does it respond to stimuli? The answers map it into our layers. If, say, we found an entity that floats in Saturn’s atmosphere, absorbing chemical energy and moving in purposeful ways (metabolism and response – yes; reproduction – unknown), we might classify it as (ii) autonomous life even without evidence of reproduction, considering it akin to an isolated organism. We’d then seek evidence of population or evolution to move it to (iii). If it also has complex behaviours (maybe it herds others of its kind or solves mazes we present), we might suspect (iv) intelligence.
Edge Revisions: The framework is also iterative: if an entity defies the categories, we revise. For example, prions (infectious proteins) replicate (sort of) and evolve in a limited way but have no metabolism or genes. The framework puts them at minimal life edge. We might decide to classify prions as not life but life-like agents, because they only replicate in the presence of real life (they’re more like a process than an autonomous entity). That’s an example of how the decision framework can say ‘no, not alive, but close – category of its own’.
This layered ontology honours the insight that life is not a binary but a cluster of properties【12†L575-L583】. However, it still provides thresholds for practical decisions. For instance, in planetary protection protocols (deciding if we found life on Mars), detecting any sign of metabolism (layer ii) or replication (layer iii) would be enough to provisionally declare life. We wouldn’t insist on consciousness (irrelevant there). For granting AI rights, we’d ignore layers ii and iii (the AI doesn’t need to metabolise or evolve biologically) and focus on consciousness (iv). So depending on context, different layers become the decisive criterion:
Astrobiology: layer (ii) or (iii) is the aim (finding autonomous metabolic life or at least evolving organisms).
Origin of life research: trying to go from chemistry to layer (i) then (ii).
Ethics/Law: concerns mostly layer (iv) (persons) and sometimes (ii) (the intrinsic value of a life).
Philosophy of mind: whether (iv) can exist without (ii) is a question of artificial consciousness.
Artificial life: tries to recreate (ii) and (iii) in new media, and perhaps eventually (iv).
By explicitly distinguishing these, our ontology avoids conflating issues. We won’t argue whether a virus is alive using the same terms as whether a fetus is alive – they are different layers (virus is genetic but not autonomous; fetus is autonomous to degree but also an organism). And we won’t be confused about AI: we can say, biologically not alive (no layer ii), but potentially an intelligent agent (layer iv), therefore give it rights appropriate to conscious beings but don’t call it ‘organism’.
This approach thus provides a decision framework: evaluate an entity against each life criterion set (autonomy, evolution, consciousness). If it meets none – not life. If it meets some – categorise as minimal or partial life. If it meets first two – definitely life. If it meets all three – life with mind. We can further subclassify ambiguous cases:
Parasitic or incomplete life: needs another life to fulfil full criteria (viruses needing cells, etc.).
Proto-life chemical systems: not fully genetic but clearly more than random chemistry.
Post-biological persons: conscious entities derived from life but no longer biological in operation (if we ever encounter or create them).
One can envision an ontology graph where nodes like ‘autonomous system’, ‘self-replicating system’, ‘evolves’, ‘sentient’ overlap. Real-life entities are mapped onto this graph showing which properties they have. That graph would visually show clusters: e.g., all Earth organisms are in the intersection of autonomy & evolution; viruses sit in evolution-only node; hypothetical AI in sentient-only node; etc. While we omit drawing the graph here (per instructions), this mental image can help in applying the framework.
In conclusion, this substrate-independent ontology of life accepts that life is a multi-layered concept. By separating the layers, we resolve many debates by placing each entity at the appropriate layer instead of forcing a yes/no answer. It also ensures our search for life, creation of life, and assignment of value to life are done with clarity about which aspect of ‘life’ we mean.
Life, in this view, is not a single monolithic property but a constellation of properties. An entity may shine with only a few of those properties or with them all. Our ontology maps the night sky of possible ‘life’ so we can navigate by those stars when we venture into the unknown.
Annotated Bibliography
Stanford Encyclopedia of Philosophy (2021) – ‘Life’: A comprehensive philosophical overview of definitions of life【50†L205-L213】【53†L367-L376】. This source traces historical definitions (Aristotle, Descartes, etc.), surveys contemporary scientific definitions (NASA, autopoiesis, etc.), and discusses scepticism about defining life. It also covers related topics like artificial life, origin of life, and ethical implications【50†L215-L223】【58†L703-L712】. The SEP entry informed our historical survey and the structure of debates, providing key distinctions (e.g., life as a natural kind, definitional pluralism, etc.).
Carnap, R. (1950) – Logical Foundations of Probability (Explication method): Carnap’s idea of explication【3†L14-L22】 guided our approach to definitions. We cited Carnap’s description of explication【3†L14-L22】 and his warning about asking ‘What is life?’ without clarifying terms【4†L1-L4】. Although not about life per se, Carnap’s methodology underpins how we refine vague concepts scientifically.
Wittgenstein, L. (1953) – Philosophical Investigations (Family Resemblance): Wittgenstein’s notion that some concepts lack a single essence【12†L575-L583】 helped frame life as a family-resemblance concept. We used a secondary explanation from SEP on language-games and family resemblance【12†L575-L583】【12†L584-L590】 to illustrate that life may be a cluster category. This supports our layered approach rather than a strict definition.
Prototype Theory (Rosch, Lakoff, et al., 1970s–80s): We referenced cognitive science research on prototypes【16†L1-L4】 via a Wikipedia summary. It explained that categories can have graded membership (some members more central)【16†L1-L4】. This supported the idea of ‘degrees’ of life and informed our minimal-to-full-life spectrum.
Haslanger, S. (2000; 2012) – Conceptual Engineering and Ameliorative Definitions: Haslanger’s work (cited via a 2005 paper) on descriptive vs ameliorative projects【20†L123-L131】【20†L145-L152】 shaped our view that definitions of life can be purpose-relative. We used her distinction to argue that NASA’s definition is an ameliorative, operational one for discovery, while an ethicist’s might be different.
Daniel Koshland (2002) – ‘The Seven Pillars of Life’: Cited in the Astrobiology article【22†L173-L181】, Koshland’s ‘PICERAS’ (Program, Improvisation, etc.) is a modern list-definition. We used an anecdote from Koshland【22†L153-L162】 where scientists struggled to define life and realised reproduction alone was insufficient (the ‘one rabbit is dead, two are alive’ paradox). This illustrated the pitfalls of single-criterion definitions.
Joyce, G. (1994) – NASA Definition of Life: Gerald Joyce’s summary of the NASA committee’s definition【39†L1-L5】 was crucial. We cited the phrase ‘self-sustaining chemical system capable of Darwinian evolution’【39†L1-L5】 multiple times as a touchstone for selectionist definitions. This source anchored our Section 2.3 on evolutionary definitions and appeared in context of astrobiology and instrumentation.
Schrödinger, E. (1944) – What Is Life?: Schrödinger’s ideas on negative entropy were summarised via Wikipedia【29†L270-L278】. We cited his statement that life avoids decay to equilibrium by maintaining ‘negative entropy’【29†L270-L278】. This supported the thermodynamic perspective (Section 2.1) and informed our discussion of far-from-equilibrium systems as life markers.
Prigogine, I. & Nicolis, G. (1977) – Self-Organization in Nonequilibrium Systems: While we did not quote directly, their work (as discussed in a review article【34†L3-L10】) provided the concept that living organisms are dissipative structures sustained by entropy flow【34†L3-L10】. We cited a line confirming living organisms are ‘undoubtedly dissipative structures’【34†L3-L10】. This reinforced thermodynamic necessity for life.
Maturana, H. & Varela, F. (1972) – Autopoiesis and Cognition: We relied on a Wikipedia excerpt of the autopoietic definition【38†L184-L192】. We quoted their definition of an autopoietic machine (self-producing network)【38†L184-L192】 and noted how it defines life in terms of organisation. We also noted Varela’s link of autopoiesis to cognition【38†L230-L238】, used in Section 4. This source was key for Section 2.2 (autopoiesis) and influenced our autonomy criterion in the ontology.
Adami, C. (2015) – Interviews/Quanta Magazine on Information Theory of Life: We used a quote via an evolution news site【45†L159-L167】 that summarised Adami’s view of life as ‘self-perpetuating information strings’ and information as prediction ability【45†L159-L167】. This was central to Section 2.4 (information approach). Adami’s work in digital evolution also inspired our mention of Avida and digital life experiments.
Bedau, M. (2003) – Artificial Life and ‘Four Puzzles about Life’: While not directly cited, Bedau’s analyses and classifications (referenced in search results【46†L7-L15】) influenced our understanding of ALife definitions and the notion of life-as-it-could-be. We incorporated these ideas qualitatively in discussing Langton and ALife, even without direct quotes.
Cleland, C. & Chyba, C. (2002) – ‘Defining Life’ (origins/astrobiology): We cited Cleland’s argument against premature definitions【53†L367-L375】, including the analogy to water’s definition and urging criteria over definitions【53†L367-L376】. This supported our notes on definitional scepticism and was used to justify a broad search approach in astrobiology.
Persson, E. (2017) – ‘Towards a Family Resemblance Definition of Life’: While we couldn’t retrieve the PDF, the idea referenced in search results influenced our Wittgenstein section, confirming that family resemblance approach has been explicitly considered for life.
Evolution News (D. O’Leary, 2015) – Discussion of Life and Mind: We quoted an excerpt from this piece that referenced Adami【45†L159-L167】 within a broader discussion of consciousness and information. It also contextualised how some think of life vs rocks, etc. We used it mainly for Adami’s quote on life and info.
Mind and Life Institute (2021) – Blog on Enactive Mind (E. Thompson): We found and cited a key line on life-mind continuity【61†L1-L4】 (‘the brain expands the mind already present in life’). This supported Section 4 on enactivism. The source also explained autopoiesis and sense-making in accessible terms【60†L105-L113】【60†L115-L121】 which helped shape our narrative (though we only directly quoted the continuity line).
SEP (2018) – ‘Consciousness’ or ‘Artificial Intelligence’ (not directly cited): Background knowledge on AI consciousness, functionalism vs biological views, etc., informed Section 4. We specifically mentioned Searle’s idea of consciousness as a biological phenomenon; we attempted to search for a direct source, but it’s common knowledge from Searle (1992) that ‘Consciousness is a biological property like digestion’ – however, lacking a direct citation, we framed it generally【58†L713-L722】.
National Academies (2007) – The Limits of Organic Life in Planetary Systems: This report (though not cited) influenced our discussion of ‘weird life’ and astrobiology. It provided context that NASA is considering non-standard biochemistries.
Legal and Ethical texts on personhood (Noonan 1967, etc.): The SEP ethics section【58†L703-L712】【58†L717-L725】summarised references like Noonan (a pro-life argument) and Goodpaster (environmental ethics) which we indirectly used to mention conceptions of life’s moral status.
Trombley & Cottenie (2019) – ‘Cost of conceptual ambiguity in ecology’ (referenced in SEP【53†L341-L349】): Although not detailed in our text, this provided the idea that unclear definitions can hinder research progress, a point we echoed in Section 7 about conceptual equivocation costs【53†L341-L349】.
This annotated bibliography highlights sources that were foundational for each domain of our report. They range from primary philosophical texts to scientific articles and reviews. Each provided specific insights: historical context, theoretical definitions, or empirical examples that we integrated into the interdisciplinary narrative.
Glossary of Technical Terms
Abiogenesis: The origin of life from non-living matter. Refers to the hypothesized natural process by which life first arose on Earth. Important in discussions of whether life’s definition is tied to its origin or not (causal essentialism).
Autopoiesis: Literally ‘self-production’. A concept from Maturana & Varela defining life as a system that continuously produces and regenerates its own components, maintaining itself as a distinct entity【38†L184-L192】. An autopoietic system (e.g., a cell) is autonomous and self-sustaining.
Biogenesis (vs Abiogenesis): The principle that life comes only from pre-existing life (biogenesis) as opposed to originating spontaneously (abiogenesis). Historically significant (Louis Pasteur demonstrated no spontaneous generation in broth). Abiogenesis is studied in origin-of-life research.
Biosignature: A measurable substance or phenomenon that indicates the presence of past or present life. Examples: atmospheric gases like O₂ and CH₄ in disequilibrium, specific organic molecules, isotopic ratios, or microfossil structures. Based on known life processes.
Consciousness: The state of sentience or subjective experience. In our context, typically associated with advanced nervous systems. We consider ‘conscious life’ to mean organisms that are not only alive but also have experiences (feels something to be them). Also debated for AI (machine consciousness).
Definitional Skepticism: The position that attempts to strictly define life may be misguided or premature【53†L367-L376】. Proponents (e.g., Cleland) suggest focusing on developing a theoretical understanding or using tentative criteria, as current definitions might be overly influenced by Earthly life’s contingent traits.
Dissipative Structure: A term from thermodynamics for an ordered system that exists in far-from-equilibrium conditions by dissipating energy. Examples: a cyclone, the Belousov-Zhabotinsky chemical oscillation, or a living cell【34†L3-L10】. Life is often considered a dissipative structure (maintains order by releasing entropy【29†L270-L278】).
Entropy (Thermodynamic Entropy): A measure of disorder in a system. The Second Law of Thermodynamics states entropy in an isolated system tends to increase. Living systems maintain local order (low entropy) by expelling entropy to their environment【29†L270-L278】 (hence Schrödinger’s ‘negative entropy’ consumption).
Evolution (Darwinian Evolution): The process by which populations of organisms change over generations via variation, inheritance, and natural selection. Often considered a hallmark of life – NASA’s definition requires capability of Darwinian evolution【39†L1-L5】. Key terms: heritable variation, selection, adaptation.
Family Resemblance Concept: A concept (from Wittgenstein) defined not by strict criteria but by overlapping similarities among examples【12†L575-L583】. Applied to life, it means no single property may be necessary to all life, but life forms share a network of common features (growth, metabolism, etc.).
Homeostasis: The ability of an organism or system to maintain internal stability (temperature, pH, etc.) despite external changes. A feature of life (e.g., humans sweating to cool down). Indicates self-regulation.
Metabolism: The set of chemical reactions in an organism that maintain life, including energy-harvesting (anabolism) and breakdown processes (catabolism). Many definitions of life require metabolism (energy and material flow) as a core component – a criterion for autonomy.
Natural Kind: In philosophy, a category of things that reflects a real, fundamental grouping in nature (like elements, or species if one views them that way). The question ‘Is life a natural kind?’ asks if ‘life’ has an underlying essence (like H₂O for water) or is just a cluster of properties we defined.
Negentropy: Short for ‘negative entropy’. Schrödinger’s term for the order or information that living systems import to compensate for entropy increase【29†L270-L278】. Equivalent to saying organisms feed on free energy to maintain order.
Operational Definition: A definition framed in terms of observable operations or criteria. For life, an operational definition might be a checklist of observable traits (e.g., ‘if it moves, grows, reproduces, responds to stimuli, then it’s alive’). Useful for practical identification, even if not theoretically tight.
Origin of Life (OOL): Scientific field studying how life might have arisen from prebiotic chemistry. Competing theories (RNA world, metabolism-first, etc.) often rely on different implicit definitions of what minimal life is (genetic replication vs metabolic cycles).
Panpsychism/Biopsychism: Philosophical view that mind or experience is a fundamental feature of all things (panpsychism) or all living things (biopsychism). In our context, the enactivist hint that even simple life has ‘mind-like’ qualities is a mild form of biopsychism, suggesting life and mind continuum.
Prion: A misfolded protein that can induce other proteins to misfold in the same way, effectively replicating without genes. Causes diseases like mad cow disease. Prions are often cited in life-definition debates: they reproduce (in a sense) but have no nucleic acids or metabolism.
Self-Organization: The spontaneous formation of ordered structures and patterns in a system. Life is sometimes described as self-organising matter. Autopoiesis is a specific kind of self-organisation (with self-maintenance). But self-organisation also occurs in non-life (crystals, convection cells), so additional criteria are needed to pinpoint life.
Self-Replicator: Any entity or system capable of making copies of itself. A central concept in origin-of-life (e.g., the first self-replicating molecules) and in artificial life (von Neumann’s self-reproducing automata). Viruses are self-replicators (with help), cells are self-replicators. Some consider this the defining trait (with variation leads to evolution).
Sentience: The capacity to have sensations or experiences (often used to mean the capacity to feel pain/pleasure). A subset of consciousness. Used in ethics to decide moral standing (sentient beings deserve moral consideration). Many animals are sentient; plants and microbes are not (as far as we know).
Substrate-Neutral/Independent: Not tied to a particular physical medium. For example, saying ‘life is substrate-neutral’ means life could in principle be made of other materials (not just carbon/water) if organised properly. Similarly, ‘consciousness is substrate-independent’ implies a mind could arise in a computer, not just a brain. We aimed for a substrate-independent ontology – valid for any material basis.
Synthetic Biology: A field that combines biology and engineering to design and construct new biological parts, devices, and systems, or redesign existing biological organisms. Example feats: synthetic genomes, minimal cells, organisms with expanded genetic codes. It tests definitions by creating borderline life forms (like a cell with synthesized DNA was called ‘synthetic life’).
Technosignature: A detectable sign of technology or intelligent life (e.g., radio signals, city lights on an exoplanet). Relevant in astrobiology: searching for intelligence via its artifacts. This term relates to our discussion on contact with extraterrestrial intelligence (which might be recognised via technosignatures rather than biology per se).
Vitalism: The discredited doctrine that living organisms possess a non-physical inner force or element (élan vital) distinguishing them from inert matter. Historically significant (18th–19th century) but largely abandoned after chemical and physiological processes of life were explained without invoking a special ‘vital force’【50†L227-L236】. Modern science treats life as explainable by physics and chemistry, though emergent properties are recognised.
Viability (Fetal Viability): In medical/legal context, the stage of fetal development when a fetus can potentially survive outside the womb (with medical support). Used as a threshold in abortion law and tied to a concept of independent life – albeit still biologically dependent on intensive care. Shows an applied definition of when ‘a life’ becomes its own entity.
These terms collectively clarify the interdisciplinary vocabulary used in discussing life. They span physics (entropy), chemistry (autocatalysis not explicitly listed but implied), biology (metabolism, evolution), cognitive science (consciousness, sentience), philosophy (natural kind, panpsychism), and emerging tech (biosignature, technosignature, synthetic biology). Having these definitions at hand helps avoid confusion when these concepts overlap (for instance, distinguishing metabolic self-organisation from full autopoietic life, or distinguishing being alive from being a sentient person). Each term anchors a key part of our ontology and discussion.
