Machine Metaphors in Biology I: Origins and Success

Introduction

Metaphors are not ornaments. Science does not advance only by accumulating facts. It also advances by changing the conceptual lenses through which facts become visible as facts. Before one can measure something, one must decide what kind of thing is being measured. Before one can explain something, one must decide what kind of explanation is appropriate. This is why metaphors matter. They are not decorative literary additions that scientists attach to an otherwise neutral description of the world. They are cognitive instruments. They organize attention, make unfamiliar domains tractable by translating them into familiar ones, and silently decide which similarities deserve to be emphasized and which differences can be ignored.

A metaphor is a controlled act of displacement. It carries a structure from one domain into another. When we describe DNA as a code, the cell as a factory, the brain as a computer, the heart as a pump, or in general organisms as machines, we are not merely choosing convenient words. We are importing a whole architecture of expectations: inputs and outputs, parts and functions, rules and execution, design and control, malfunction and repair. Some of these expectations have been extraordinarily fruitful. Others have distorted the phenomena they were meant to illuminate. The difficulty is that successful metaphors often disappear as metaphors. Once a metaphor becomes sufficiently productive, it hardens into common sense. What began as a comparison becomes a worldview.

This is particularly true of machine metaphors in life sciences. Today, biological language is saturated by terms borrowed from engineering, mechanics, and computation. We speak of molecular motors, genetic circuits, cellular machinery, regulatory networks, transcriptional programs, developmental toolkits, biological chassis, synthetic devices, information processing, signaling pathways, and organic factories. These terms are not accidental. They are the sedimented result of a long intellectual history in which biology repeatedly borrowed its leading images from the most advanced machines of its age. In the seventeenth century, organisms were compared to clocks, fountains, and automata. In the nineteenth century, they were compared to engines. In the twentieth century, they became control systems, cybernetic machines, and of course, computers.

Today, the dominant metaphor is no longer simply that organisms are machines, but that living systems are information-processing systems whose activity can be simulated, programmed, optimized, and perhaps even fully reproduced in silico. There is an obvious reason why this happened: machine metaphors work! They made biological phenomena intelligible. They gave scientists a way to decompose organisms into parts, assign functions to those parts, trace causal pathways, build experiments, identify failures, and intervene. The machine metaphor helped make physiology, molecular biology, genetics, biotechnology, synthetic biology, and systems biology possible in their modern forms. It gave biology a language of explanation that was precise, manipulative, and experimentally productive. It helped transform life from an object of wonder into an object of systematic investigation.

But precisely because it worked, it also became dangerous. Not dangerous because it is false in every respect, but because it can be mistaken for the whole truth. A metaphor illuminates by selective distortion. It does not merely reveal; it filters. It highlights analogies while suppressing disanalogies. Makes visible some aspects while leaving others invisible. Machine metaphors are powerful because organisms and machines really do share important features. Both are bounded physical systems. Both contain differentiated parts. Both transform energy. Both exhibit functional organization. Both can break down. Both can be understood, at least partially, by analyzing how components contribute to the performance of the whole. These similarities explain machine metaphors’ success.

Yet organisms are also unlike machines in ways that are not superficial, starting from the fact that human-made machines are assembled from parts that usually exist before and outside the whole, while organisms produce and maintain their own parts. Similarly, machines depend on external designers and users for defining their purposes, but organisms generate their own norms of persistence through metabolism, repair, development, and reproduction. Furthermore, machines can be turned off while their hardware remains. In contrast, living cells cease to exist as cells when the processes sustaining them collapse. Last but not least, machines operate within predefined spaces of possible states. Conversely, living systems transform the spaces in which future possibilities become relevant.

The aim of the present trilogy is not to reject mechanism, celebrate methodological vitalism, or retreat into a romantic opposition between life and machine. That would be too easy, and scientifically useless. The more interesting task is dialectical. The first movement, developed in this essay, reconstructs the origins and success of machine metaphors in biology. The second movement will examine their limits when they are extended into metaphysical doctrines such as biological pancomputationalism. The third movement will ask whether a transformed concept of mechanism is possible; one that retains the experimental power of mechanistic explanation while incorporating self-produced constraints, historical contingency, endogenous anticipation, and organizational closure.

The first step of this synthesis is therefore historical. The focus of today’s essay is to address many relevant questions regarding the origins and success of mechanical philosophy. How did machines become the dominant metaphors for life? Why did this metaphor become so successful? What kinds of machines were projected onto organisms at different moments? How did a metaphor that began with clocks and automata end up shaping contemporary claims that organisms, minds, and perhaps the universe itself are computers? To answer these questions, we must begin before modern machines, with a conceptual challenge that Western philosophy never fully escaped: the problem of self-motion.

Aristotle and the first division of life

Aristotle (384–322 BCE) did not think of organisms as machines. The ancient world lacked the technological environment that would later make the machine metaphor seem natural. Yet Aristotle formulated the problem that machine metaphors would later try to solve: how can a natural body move, act, and develop as if it were organized from within?

For Aristotle, an adequate explanation required four causes. The material cause tells us what something is made of. The formal cause tells us what kind of thing it is, the organization by virtue of which matter becomes this rather than that. The efficient cause tells us what brings a process about. The final cause tells us the end toward which a process is directed. In modern science, this fourfold structure was largely flattened. Explanation came increasingly to mean the identification of efficient causes: prior events producing later events through contact, force, or lawful transition. But for Aristotle, living bodies could not be understood in this reduced way. An organism is not merely pushed from behind. It is organized toward activities characteristic of the kind of being it is.

This is clearest in animal locomotion. A stone falls because it is acted upon or because it tends toward its natural place. But an animal pursues, avoids, feeds, mates, explores, and rests. A lion chasing a gazelle does not look like a passive body externally pushed through space. It seems to move itself. Aristotle’s solution was subtle, but it also established a pattern that would haunt later biology. According to him, strictly speaking, nothing can be the efficient cause of itself in the same respect. If a whole appears to move itself, the explanation must divide the whole internally. One aspect moves; another is moved. In animals, the soul functions as the organizing and animating principle, while the body supplies the material capacities through which motion is realized.

The crucial term here is orexis (appetite, desire, directed striving). The animal moves because something in the environment appears to it as relevant: food, danger, mate, shelter. The object, as apprehended by the animal, functions as a final cause. The soul’s directedness actualizes the body’s capacity to move. Self-motion is therefore explained by an internal differentiation between an active principle and a passive body. This was not yet mechanicism. Aristotle’s biology was teleological, qualitative, and deeply concerned with form. Yet the explanatory architecture already contained a division that later mechanistic thought would radicalize. If life appears self-moving, one can explain it by distinguishing the source of direction from the material apparatus being directed.

Whether this active principle is later named intention, will, volition, information, code, controller, algorithm, frontal cortex, corpus striatum, or anterior cingulate sulcus, the explanatory structure remains the same: something directs; something executes. This matters because machine metaphors did not emerge from nowhere. They inherited the old philosophical problem on how to explain organized activity without allowing the organism as a whole to become an irreducible cause. Aristotle kept the organism intelligible by multiplying causes. Modern mechanical thinking would keep it intelligible by narrowing causation.

From living cosmos to mechanical nature

The world of Aristotle and much of medieval natural philosophy was not primarily a world of inert matter. It was a world of forms, purposes, qualities, and natural tendencies. Even when this worldview was criticized, transformed, or Christianized, nature remained thick with internal principles. Things did not merely occupy positions in a neutral space; they had characteristic ways of being.

Modern machine metaphors arose by breaking this picture. The change was not purely philosophical. It was also technological. Mechanical clocks, mills, pumps, fountains, and automata transformed the imagination of European intellectual life. A clock is not just a useful device. It is a visible demonstration that complex, ordered, apparently purposive motion can arise from the arrangement of material parts. Automata sharpened the point. If a mechanical bird could sing, a mechanical figure could move, and a hydraulic fountain could produce lifelike behavior through hidden channels and pressures, then perhaps the activities of animals could be explained in the same way. Machines became an epistemic bridge between artifice and nature.

This technological transformation encouraged a new metaphysical image. Nature could be understood as matter in motion. Qualities could be replaced by quantities. Hidden forms could be replaced by visible arrangements. Purposes could be replaced by mechanisms. The world no longer needed to be read as an organism animated by inner tendencies; it could be analyzed as a machine governed by lawful interactions among parts. This shift was attractive because it promised clarity. A machine can be taken apart. Its parts can be isolated. Their arrangement can be studied. If the same arrangement is reconstructed, the same behavior should follow. Mechanicism therefore offered a new ideal of explanation: to know a system is to know how its parts generate its behavior.

In life sciences, this ideal was revolutionary. It suggested that living bodies, however complex, could be investigated without invoking occult qualities. Digestion, circulation, sensation, locomotion, and reproduction could be studied as physical processes. Organisms could be brought into the laboratory as complex physical systems. Yet this clarity came at a price. In order to make life mechanistically intelligible, modern science had to strip matter of intrinsic activity. Matter became passive, extended, movable, divisible, and governed by external laws. If organisms still appeared active, the activity had to be relocated, either outside the physical body, as in dualism, or inside a special arrangement of parts, as in reductionism. The drama of modern biology begins with this relocation.

Descartes and the genesis of machine metaphors

René Descartes (1596–1650) gave machine metaphors their classical philosophical form. In his natural philosophy, the bodies of animals could be understood as automata. Their movements did not require a rational soul. They could be explained by the arrangement of organs, fluids, nerves, and mechanical structures. The animal body became a machine more intricate than any human artifact, but a machine nonetheless.

Descartes did not simply compare bodies to machines. He made the comparison ontologically serious. The physical body belonged to res extensa, extended substance. It was spatial, divisible, and governed by mechanical laws. Mind belonged to res cogitans, thinking substance. The human being was therefore divided between a physical body that could be mechanized and a thinking mind that could not be reduced to extension. In non-human animals, Descartes denied the rational soul and treated behavior as bodily mechanism. In humans, he preserved thought by separating it from the machine. This is the decisive Cartesian move, transforming Aristotle’s internal distinction between soul and body into an ontological rupture.

The organism is thus no longer explained through a coordination of material, formal, efficient, and final causes. Scientific explanation is increasingly restricted to efficient causation, now interpreted as mechanical interaction. Formal and final causes are treated with suspicion. Purposes belong to theology or psychology, not to the proper study of matter. Forms become scholastic residues. The body becomes a machine. This transformation was enormously productive. Cartesian physiology encouraged researchers to look for material mechanisms behind life processes. The circulation of blood, muscular contraction, reflexes, sensation, and digestion could be investigated without appealing to vital spirits in the old sense. The body could be anatomized, represented, modeled, and repaired.

The genesis of machine metaphors made biological inquiry experimentally aggressive. But Descartes also created a problem that never went away. If the body is a machine and the mind is outside the machine, then organization, meaning, and purpose become difficult to naturalize. Either they are expelled from physical explanation, or they must later be reintroduced in mechanized form. In fact, much of the history of biology after Descartes can be read as a series of attempts to bring back what mechanical philosophy excluded, but without abandoning the mechanical clarity that made modern science powerful. The first great attempt was Newtonian. If Descartes made the animal body into a machine, Newton helped make the universe machine-like in a mathematically precise way.

Newton, Laplace’s Demon, and the clockwork universe

Isaac Newton (1643–1727) did not simply provide a metaphor. He showed that the motion of bodies could be expressed through universal laws. Celestial and terrestrial motions could be brought under a common mathematical framework. Prediction became the mark of understanding. The heavens, once a domain of qualitative hierarchy, became a law-governed system of bodies moving according to mathematical relations.

The popular image that followed was the clockwork universe. Strictly speaking, Newton’s own thought was more complicated than later mechanistic caricatures suggest. His theory of gravitation involved action at a distance, and his natural philosophy remained entangled with theological concerns. But the cultural effect of Newtonian mechanics was unmistakable. Nature increasingly appeared as a system whose future states could be predicted from its present state plus universal laws. The ideal of explanation became the reconstruction of trajectories from initial conditions. This ideal reached its most famous expression in Laplace’s demon; an intellect that, knowing the positions and velocities of all particles and the forces acting on them, could compute the entire future and past of the universe.

Whether or not any real intellect could do this was secondary. The image crystallized a metaphysical dream: reality as a fully determinate state-space evolution. This dream profoundly shaped biology. If living organisms are physical systems, then in principle their behavior should also be derivable from the motions of their parts. The whole becomes a consequence of local interactions. Agency becomes a name for complicated efficient causation. Purposes become projections. The organism as a whole becomes explanatorily dispensable, except as a convenient shorthand for the activity of its components. In this way, the Newtonian machine metaphor therefore deepened the Cartesian one. Descartes mechanized bodies. Newtonianism mechanized nature.

Once the success of mathematical physics became the standard of science, biology inherited an aspiration of becoming a physics of living matter. The more biological explanation resembled mechanics, the more scientific it seemed. This aspiration generated both triumph and unease. The triumph was methodological. Biology could become exact, causal, and experimentally controlled. The unease was conceptual. Organisms did not behave like passive aggregates. They developed, repaired, reproduced, regulated, and pursued. They were not merely moved; they seemed to maintain themselves. The machine metaphor could describe their parts, but could it describe their organization? This question became unavoidable in the eighteenth century.

La Mettrie and Kant: the fork in the road

Julien Offray de La Mettrie (1709–1751) pushed the mechanistic view to its most provocative conclusion. In his L’Homme Machine, the human being itself becomes a machine. Thought, sensation, appetite, and moral life are not signs of an immaterial soul but outcomes of bodily organization. Descartes had mechanized animals while preserving the human mind. La Mettrie removed the exception. The human body does not merely house a machine; the human being is a machine.

This radical materialism was scandalous, but it also clarified the mechanistic ambition. If life and mind can be explained by the arrangement of matter, then biology needs no special principles. The same strategy that explains clocks and pumps may eventually explain sensation, behavior, and thought.

Machine metaphors then become not merely a heuristic but a declaration of ontological continuity between organisms and artifacts. Immanuel Kant (1724–1804) saw the difficulty. In his Kritik der Urteilskraft, he distinguished organisms from machines with remarkable precision. A watch contains parts that exist for one another, but the parts do not produce one another. A gear does not grow another gear. A spring does not repair another spring. The watch does not generate the conditions of its own maintenance. Its organization is imposed from outside by an artisan. An organism, by contrast, is a natural purpose: a system in which the parts exist for and through the whole, while the whole exists through the reciprocal activity of the parts.

In a living being, parts are not merely arranged; they are generated, maintained, repaired, and functionally integrated by the organism’s own activity. The organism is both organized and self-organizing. This is why Kant thought mechanistic explanation, although necessary, could never fully capture organisms as organisms. We must investigate organisms mechanistically, but we are also compelled to judge them teleologically. At this point, it is necessary to make an important clarification. Kant’s position is often misunderstood as a retreat into vitalism. It is more subtle than that. He did not deny the necessity of mechanical explanation. He denied that mechanism alone could make the organism intelligible as a self-organizing natural product.

The understanding of organisms thus forces a tension between two explanatory demands, two complementary modes of explanation: the demand for efficient causal mechanisms and the demand to understand the reciprocal organization of parts and whole. This tension defines the rest of machine metaphors history. La Mettrie represents the maximal expansion of mechanical thinking. Kant represents the realization that something in the organism resists the analogy. But remarkably, modern biology did not choose one side once and for all. It repeatedly moved forward by mechanizing more of life while leaving unresolved the deeper question of self-organization. The next transformation came not from clocks, but from engines.

From clocks to engines: the thermodynamic body

The clock was the emblem of early modern mechanicism because it represented order, precision, and regular motion. But the nineteenth century was the age of the engine. Steam engines, industrial machinery, railways, factories, and thermodynamic theory transformed the scientific imagination. The machine was no longer only a device of arrangement. It became a device of transformation.

This shift mattered for biology. A clock transfers motion. An engine converts energy. It consumes fuel, produces work, releases waste, and runs down. It must be maintained. It is not merely an arrangement of parts but a process embedded in flows of energy and matter. The engine therefore provided a more dynamic metaphor for living systems than the clock ever could. Thermodynamics emerged from attempts to understand the efficiency of heat engines. Sadi Carnot (1796–1832) analyzed the conditions under which heat could be transformed into work. Rudolf Clausius (1822–1888) and Lord Kelvin (1824–1907) developed the conceptual foundations of the second law and entropy. Hermann von Helmholtz (1821–1894) and others connected the conservation of energy to physiology.

The living body could now be understood as a system that transforms chemical energy into work, heat, movement, and maintenance. This was a profound scientific advance. Metabolism became intelligible in energetic terms. Muscles could be studied as energy-transforming systems. Food became fuel. Respiration became combustion without flame. The body became an engine, and physiology gained a quantitative language for life as thermodynamic work. The engine metaphor also changed the meaning of mechanism. The organism was no longer merely a clock-like arrangement of stable parts. It was a flow-through system. It required inputs. It generated outputs. It dissipated energy. It maintained itself far from equilibrium.

The above made machine metaphors more biological by introducing metabolism, expenditure, and irreversibility. Yet the engine metaphor still lacked something essential. It explained power, but not steering. It showed how energy could be transformed into work, but not how work could be organized toward the maintenance of a living system. A steam engine does not decide where to go. It has no intrinsic norm of viability. Its constraints are externally designed. Its function belongs to the purposes of its builders and users. This distinction would become central later, especially for theories of biological organization. But in the nineteenth century, the engine metaphor was enough to deepen the mechanization of life. It allowed biology to incorporate dynamics without abandoning the Newtonian paradigm.

So far, life has become not just arranged matter, but energetic matter under constraint. The next challenge was to explain adaptation itself. A universe that runs like clockwork still needed a creator, even if it were a blind watchmaker. How could the appearance of design be mechanized without invoking a designer?

Darwin and the mechanization of design

Charles Darwin (1809–1882) transformed machine metaphors by changing the meaning of design. Before Darwin, organisms were often treated as artifacts whose exquisite contrivances pointed beyond nature to divine intelligence. The eye, the wing, the hand, and the flower seemed machine-like because they appeared fitted to ends. Machine metaphors therefore served natural theology: organisms were machines because they looked designed.

Darwin did not deny the appearance of design. He explained it historically. Natural selection showed how complex adaptations could arise through variation, inheritance, differential survival, and reproduction. The apparent finality of organs could be reconstructed as the cumulative result of a blind historical process. Wings are for flying not because they were planned for flight by an external designer, but because lineages possessing variations useful for locomotion, gliding, display, insulation, or other functions were differentially preserved and transformed over time. This was one of the most important mechanistic victories in the history of science. Darwin provided a mechanism for the origin of mechanisms. He showed how organized contrivances could arise without foresight.

In this way, teleology was not simply abolished; it was naturalized through history. Biological function could be explained by selection rather than design. Here, it is important to note that “mechanism” in Darwinian biology has at least two senses. It can mean a cause-law process, as in the mechanism of heredity or natural selection. It can also mean contrivance, as in the mechanism by which an orchid attracts an insect. In his contribution, Darwin used both senses. Natural selection is a mechanism in the causal sense. Adaptations are mechanisms in the functional sense. The power of Darwin’s theory lies in connecting them: functional contrivances can be explained by a historical causal mechanism.

This connection made biology more mechanistic while also making mechanism more historical. Newtonian mechanism explained trajectories from initial conditions. Darwinian mechanism explained forms through descent with modification. Thus, organisms were no longer only machines operating in the present. They were the product of a long sequence of inherited transformations. Machine metaphors acquired memory. This was a major conceptual achievement. It allowed biology to preserve function without external purpose. Hearts pump blood because pumping blood contributes to the persistence and reproduction of organisms with hearts. Eyes see because visual capacities mattered in ancestral environments. Functional language could remain scientifically legitimate because it could be grounded in evolutionary history.

Yet Darwin also complicated machine metaphors. Human-made machines are designed in advance. Biological “machines” are assembled retrospectively by selection acting on variation. They are not optimized from first principles. They are bricolages, compromises, jury-rigged arrangements, historical palimpsests. Evolution does not design like an engineer; it tinkers with what is available in the environment. This difference would later become one of the central criticisms of engineering metaphors in biology. But it definitely did not weaken mechanicism. It made mechanicism historical. The late nineteenth and early twentieth centuries then pushed in a more aggressively experimental direction.

Jacques Loeb and the engineering ideal

If Darwin mechanized biological design historically, Jacques Loeb (1859–1924) tried to mechanize biological behavior and development experimentally. Loeb’s ambition was not merely to describe organisms as machines but to control them as machines. He opposed vitalism and argued that biological phenomena should be explained in physico-chemical terms. Animal behavior, development, and reproduction could be manipulated by identifying the relevant physical and chemical conditions. Loeb’s work on tropisms is emblematic. A moth flying toward light need not be interpreted as choosing, intending, or seeking. Its movement could be explained as a forced orientation induced by external stimuli. Behavior becomes a lawful reaction to gradients, chemicals, light, and physical conditions.

Under Loeb’s lens, the organism becomes a physico-chemical system whose apparent purposiveness can be decomposed into local causal responses. His experiments on artificial parthenogenesis carried the same message. If development could be initiated by chemical or physical manipulation, then life processes were not mysterious. They were experimentally controllable and biologists could thus become engineers of living processes. In this way, the question was no longer simply whether organisms are machines. The question was whether one could intervene in them with the precision of an engineer. This ideal, imposed by the contemporary success of other fields such as physics, remains alive in contemporary synthetic biology, developmental bioengineering, and molecular medicine.

Loeb’s contributions also reveal why machine metaphors became so attractive to scientists. A machine is not only something one can understand. It is something one can manipulate. Mechanistic explanation promises control. If one knows the parts, the forces, the pathways, and the conditions, one can predict and intervene. The machine metaphor thus connects epistemology and technology: to know life mechanistically is to gain power over life. This does not mean that Loeb solved the problem of organismal organization posed by Kant two centuries back. Like many mechanists, he treated purposive activity as a surface appearance to be dissolved into efficient causes. But his program strengthened the experimental culture that would later make molecular biology possible.

Life would then be explained not by contemplating its essence but by manipulating its mechanisms. The decisive “new mechanicism” arrived when biology found a way to talk not only about energy, but about information.

Turing, Shannon, and Wiener: the computer enters biology

The twentieth century did not abandon machine metaphors. It updated the meaning of machine. The dominant machine was no longer the structural clock or the metabolic steam engine, but the computer. This transition changed the biological imagination more radically than any previous metaphorical shift. The engine gave biology a way to talk about power. The computer gave biology a way to talk about control. Living systems do not merely transform energy; they regulate, signal, remember, correct, encode, decode, and respond. To understand these activities, biology needed a language of instructions, communication, feedback, and symbolic order. That language emerged from computation, information theory, and cybernetics.

Alan Turing (1912–1954) formalized the notion of effective procedure as rule-governed symbol manipulation. A Turing machine reads and writes symbols according to finite rules. It abstracts from the material substrate and focuses on the structure of procedure. This abstraction was revolutionary because it clarified what a machine could do in principle. Mechanism no longer meant gears, levers, and pistons. It meant finite rules operating over discrete symbols.

Claude Shannon working in a different context, mathematized communication. Information could be quantified independently of meaning. A message could be analyzed in terms of uncertainty, probability, coding, transmission, and noise. This was not a theory of biological meaning, but it provided a powerful formal language for signals. Once information could be measured, it could travel across domains: telegraphy, electronics, genetics, neuroscience, and ecology.

Norbert Wiener (1894–1964) then fused communication and control in cybernetics. His book title already announced the ambition: control and communication in the animal and the machine. Positive and negative feedback became key concepts. A system could regulate its behavior by sensing its output and modifying its input. Purposeful behavior could be modeled without invoking an immaterial purpose. Goal-directedness could be redescribed as error correction.

This was a real turning point. Aristotle’s orexis, Descartes’ excluded active principle, Kant’s natural purpose, and Loeb’s tropisms could now be reformulated in informational terms. The organism acts because it receives signals, processes information, compares states, and adjusts behavior through feedback. The active principle becomes neither soul nor vital force but control architecture. The computer metaphor therefore differed from earlier machine metaphors in one crucial respect. It did not merely mechanize the body. It mechanized direction. A clock explains regular motion. An engine explains energy transformation. A computer explains rule-governed control. Once this metaphor entered biology, the organism could be imagined as a hierarchy of programs, signals, codes, and circuits, as nowadays we do.

This shift also reconfigured what we mean by causation. In classical mechanics, causes are forces. In thermodynamics, they are energy gradients and constraints. In information theory and cybernetics, causes can be messages, differences, instructions, and feedback loops. The language of information and computation allowed biology to speak of specificity without returning to Aristotelian forms and purposes. It allowed biologists to say that one molecular sequence could determine another, that signals could control pathways, that genes could encode products, and that regulatory networks could compute cellular decisions. This was the conceptual environment in which molecular biology, a transformative period between the 1950s and 60s, flourished.

Schrödinger, von Neumann, and the logic of self-reproduction

In his book, What Is Life?, Erwin Schrödinger (1887–1961) addressed the problems of genetics, looking at the phenomenon of life from the point of view of physics. Said manuscript proved to be of great influence among the scientists of the era, and it is often remembered because it helped inspire molecular biology. Its central question was physical: how can living systems maintain order in a world governed by thermodynamic degradation?

Schrödinger’s answer involved two persuasive ideas. First, organisms feed on negative entropy, maintaining order by drawing upon environmental flows. Second, hereditary stability requires a molecular structure capable of storing complex information without simple periodic repetition. He called this an aperiodic crystal and described it through the image of a code-script. This was a metaphor of enormous power. Heredity became textual. The organism’s future was imagined as written in molecular structure. A gene was not merely a material particle; it was a bearer of instructions. Schrödinger did not invent the idea that heredity involves organization, but he gave physicists and biologists a vivid image: life is order maintained by molecular information.

John von Neumann (1903–1957) approached the problem from the side of automata theory. He asked what kind of machine could reproduce itself while allowing open-ended increases in complexity. His answer required a distinction between a constructor and a description. A self-reproducing automaton must contain instructions that are used in two ways: interpreted to build the machine and copied as data for the next generation. This dual use of description anticipated, at an abstract level, one of the most important features of genetic systems. DNA is both transcribed and replicated. It is both interpreted and copied.

When the structure of DNA was discovered, the computational metaphor found its material anchor. The double helix suggested a mechanism for replication. The sequence of bases suggested a digital alphabet. The cracking of the genetic code showed how nucleotide triplets correspond to amino acids. Molecular biology now had what looked like a code, a tape, a program, a translation system, and a cellular machinery for executing instructions.

This was the golden age of the information metaphor in biology. The central dogma, the term “genetic code”, including messenger RNA, transcription, translation, regulation, and genetic programs, all became part of a new explanatory vocabulary. The cell was no longer only a chemical factory. It was an information-processing system. DNA was the archive, RNA the messenger, ribosomes the reading machinery, proteins the effectors, and regulatory circuits the control logic. This vocabulary was not merely rhetorical. It guided research. It suggested experiments. It made gene expression intelligible. It allowed biologists to search for signals, regulatory elements, feedback loops, switches, repressors, promoters, operators, and networks. It helped transform biology into the molecular science we know today!

Yet the metaphor is also smuggled in a hierarchy: DNA as software, organism as hardware; genes as instructions, cells as execution environment; genotype as plan, phenotype as output. This hierarchy was often too simple, but it was productive. It allowed the molecular biology of the twentieth century to proceed with extraordinary speed. The cell became legible as a machine that reads, writes, copies, edits, repairs, and executes.

A gene-centered view of evolution

By the second part of the twentieth century, François Jacob (1920–2013), Jacques Monod (1910–1976), and André Lwoff (1902–1994), helped make gene regulation a central part of biology. The lac operon showed that genes are not merely structural units but regulated components of cellular control. Cells respond to environmental conditions by turning genes on and off. Molecular biology thus became cybernetic; not just genes making proteins, but regulatory systems controlling expression.

The genetic program metaphor had obvious appeal. Programs specify operations. They can be executed. They can contain conditional instructions. They can regulate behavior. They can be copied with modifications. They can generate complex outcomes from simple rules. In development, the metaphor suggested that the organism unfolds according to instructions encoded in DNA. In evolution, it suggested that selection modifies programs. In medicine, it suggested that disease can be treated by identifying errors in code. As Sydney Brenner (1927–2019) emphasized over a decade ago, Turing and von Neumann provided conceptual tools for thinking about biological information. Molecular biology discovered a new logic of life based on sequences, codes, messages, and machines that interpret them.

Jacques Monod’s image of the cell as a chemical automaton gave the metaphor philosophical force. Life could be understood as lawful molecular machinery shaped by chance and necessity. The organism no longer required vital forces, or Aristotelian formal and final causes. It required molecular mechanisms, regulatory circuits, and evolutionary history. This was an extraordinary scientific achievement. The successes are too large to dismiss. Molecular genetics made heredity experimentally tractable. It enabled recombinant DNA technology, genome sequencing, genetic engineering, molecular diagnostics, gene therapy, and eventually CRISPR-based genome editing. The informational machine metaphor did not merely describe biology. It helped build modern biotechnology.

But even here, the metaphor worked best when used with discipline. DNA is not a blueprint in the same way an architectural drawing is a blueprint. A blueprint is externally interpreted by builders who exist independently of it. DNA is interpreted by cellular machinery that is itself produced through the history of the system. The “program” does not run in a neutral machine; it runs in a living, metabolizing, historically constructed cell. Genes do not dictate organismal form line by line. They participate in regulated, context-sensitive, multilevel processes. Still, the metaphor’s productivity was undeniable. It made biological complexity seem programmable, manipulable, controllable. This was the condition for the next step: creating life from scratch.

Cells as factories, circuits, and chassis

Synthetic biology represents machine metaphors in their most explicit contemporary form. Its language is unapologetically engineering-oriented. Cells are chassis. Genes are parts. Regulatory pathways are circuits. Biological functions are devices. Metabolic pathways are production lines. Organisms can be redesigned, refactored, standardized, optimized, and assembled. As we learnt above, this language did not arise by accident. It emerged from the success of molecular biology and the increasing ability to synthesize, edit, and recombine DNA. Once DNA became writable as well as readable, the cell could be treated as a programmable platform. The goal was no longer only to understand living systems but to build new ones or redesign existing ones.

This engineering metaphor has real advantages. It encourages modularity. It demands explicit design principles. It supports standardization. It makes biological construction teachable to interdisciplinary teams of biologists, engineers, computer scientists, and entrepreneurs. It links biological research to industrial production, medicine, environmental sensing, and materials science. It makes it possible to imagine bacteria that detect toxins, cells that manufacture drugs, microbes that produce fuels, tissues that repair damage, and biosensors that respond to environmental signals. This is why criticisms of machine metaphors must be careful. The problem is not that engineering language is useless. It is often useful precisely because it forces clarity.

To build a genetic circuit, one must specify inputs, outputs, regulatory logic, failure modes, and measurable behavior. To redesign metabolism, one must track fluxes, bottlenecks, yields, and constraints. To make biology tractable, one often needs machine-like abstractions. The deeper issue is whether the abstraction is mistaken for the organism. A chassis in engineering is a stable platform onto which components can be mounted. A cell is not a passive platform. It grows, mutates, regulates, repairs, adapts, competes, responds to stress, and changes the conditions under which engineered parts function. Biological parts are context-sensitive. The same gene circuit can behave differently across strains, environments, histories, and physiological states. Evolution can break engineered design. Noise can become functional. Interdependencies can defeat modular assumptions.

Yet even these difficulties have not killed the metaphor. They have made it evolve. Modern synthetic biology increasingly acknowledges burden, host context, retroactivity, stochasticity, resource allocation, evolutionary instability, and ecological interaction. In other words, biology is forcing engineering to become less naive. Machine metaphors are not a one-way imposition from technology onto life. It is reciprocal. As we use machines to understand organisms, organisms force us to rethink what machines can be. This reciprocity is important. It prevents history from becoming a simple morality tale in which mechanists are wrong and organicists are right. Machine metaphors have repeatedly changed because machines themselves changed!

Clocks, engines, computers, robots, large language models, soft machines, biohybrid devices, and self-organizing materials do not support the same analogies. The question is not whether organisms are machines in some timeless sense. The question is what concept of machine is being used, what it reveals, and what it hides. This question became especially dramatic at the end of the twentieth century.

Artificial Life and the temptation of realization

By the late twentieth century, the computational machine metaphor had become powerful enough to generate a new ambition: not merely to model life, but to synthesize life-like behavior in artificial media. Christopher Langton’s Artificial Life program defined the field as the study of human-made systems exhibiting behaviors characteristic of living systems. The methodological shift was significant. Traditional biology analyzes life as it exists. Artificial Life attempts to synthesize life-like processes in order to understand life’s general principles. This was an exciting move. It liberated biology from carbon chauvinism. If life is defined by organization rather than material substrate, then perhaps the same living principles can be instantiated in different media, whether chemical, robotic, digital, or hybrid.

Artificial Life made it possible to study evolution, self-organization, adaptation, swarm behavior, cellular automata, artificial chemistries, and virtual ecologies. It gave researchers a laboratory for exploring possible life, not merely actual life. Machine metaphors here become almost metaphysical. If life is a pattern of organization, and if computation can instantiate that organization, then a sufficiently rich simulation might not merely represent life. It might realize it. Strong Artificial Life took this possibility seriously. A digital organism would not be a model of an organism but an organism in another medium. However, as Howard Pattee has argued since the very first workshop on Artificial Life, this is where the distinction between simulation and realization becomes unavoidable.

A simulated hurricane does not make the room wet. A simulated fire does not burn. A simulated digestive system does not metabolize. A simulated cell does not maintain itself through physical exchange with a real environment unless the relevant material processes are actually implemented. The question is not whether simulations are useful. They are indispensable. The question is whether formal organization alone is sufficient for life, or whether life requires a particular kind of material-semiotic closure in which symbols, constraints, energy flows, and construction processes are physically coupled. In living systems, symbolic descriptions do not float freely. They are physically embodied and interpreted by molecular machinery.

DNA functions as a symbol system only because the cell contains constraints capable of reading, translating, and using it. The epistemic cut—as Pattee termed it—between symbol and dynamics is not an arbitrary philosophical distinction; it is physically realized in the organization of the cell. In a computer simulation, by contrast, the interpretation of symbols is supplied by the pre-existing hardware, operating system, and human-defined semantics. The simulated organism does not generate the conditions of its own interpretation. Artificial Life therefore extends the machine metaphor to its limit. It asks whether life is ultimately independent of its material realization, casting aside the material, final, and formal causes. This question will return in the second part of the trilogy.

For now, what matters historically is that Artificial Life was possible only because the computer metaphor had already transformed biology. Once life was understood as an information-processing organization, digital life became thinkable. The final expansion of the current machine metaphor was pancomputationalism.

From digital organisms to computational universes

The contemporary machine metaphor no longer stops at organisms. In its most expansive form, it treats the entire universe as computational. If every physical process can be described as state transition, and if computation is defined broadly enough as rule-governed state transition, then everything computes. Particles compute. Rocks compute. Cells compute. Brains compute. Societies compute. Evolution computes. The universe computes its own future. Stephen Wolfram’s work popularized one version of this image. Simple programs, iterated over time, can generate astonishing complexity. Cellular automata show that simple rules can produce patterns that look unpredictable, structured, and lifelike. The lesson drawn by pancomputationalists is that nature itself may be understood as the unfolding of computational rules. Physics becomes software. Reality becomes execution.

Seth Lloyd and David Deutsch extended the metaphor into quantum computation. If physical interactions can be interpreted as quantum information processing, then the universe is not merely like a computer. It is a quantum computer. In each case, machine metaphors evolve: from clockwork determinism to computational process, from gears to bits, from force to information, from prediction to transformation. There is a reason this view is seductive. Computation is the most general machine metaphor we currently possess. A clock is a specific mechanism. An engine is a specific energy converter. A computer, by contrast, is a universal formal device. It can simulate clocks, engines, organisms, economies, ecologies, and perhaps universes.

This universality, in the philosophical and computational sense, makes it tempting to promote computation from model to ontology, or even to metaphysics. But this is exactly where success becomes overreach. The fact that a process can be modeled computationally does not show that the process is literally computation in the same sense. The map is not the territory. A computational description may be the best tool we have for prediction, compression, or simulation, but that does not mean the world is made of computations. The danger is especially acute in biology, where the language of code, program, and information can obscure the material, historical, and organizational conditions that make biological symbols meaningful in the first place. I will revisit this discussion in depth during the second chapter of the present trilogy.

The main point here is that pancomputationalism is historically intelligible. It is the latest stage in a long series of projections from dominant technologies onto nature. The seventeenth century saw clocks and automata. The nineteenth century saw engines. The twentieth century saw computers and control systems. The twenty-first century sees networks, algorithms, artificial intelligence, and programmable matter. Each age mistakes its most powerful machines for the deep grammar of reality; I wouldn’t be surprised if there were someone out there claiming that the universe is a massive large language model… This does not make machine metaphors worthless. It makes them historically situated.

Why machine metaphors succeeded

At this point, it is tempting to tell the story as a sequence of errors. Aristotle divided the organism. Descartes mechanized the body. Newton mechanized nature. La Mettrie mechanized life. Thermodynamics mechanized metabolism. Darwin mechanized design. Loeb mechanized behavior. Turing mechanized information. Molecular biology mechanized heredity. Synthetic biology mechanized construction. Pancomputationalism mechanized the universe. But that would be too simple. The machine metaphor did not succeed because scientists were naive. It succeeded because it repeatedly delivered results.

It delivered decomposition. Complex living systems could be broken into organs, tissues, cells, organelles, molecules, pathways, and interactions. This made biological complexity experimentally manageable.

It delivered localization. Functions could be associated with parts: the heart pumps, the lungs exchange gases, enzymes catalyze reactions, ribosomes synthesize proteins, ion channels regulate excitability, genes contribute to traits. Even when such assignments were oversimplified, they generated successful research programs.

It delivered intervention. Machines can be repaired, modified, controlled, and redesigned. Mechanistic biology made medicine more precise. It supported pharmacology, surgery, biotechnology, and genetic engineering.

It delivered formalization. Mechanisms can be represented in diagrams, equations, circuits, networks, algorithms, and models. This made biology compatible with physics, chemistry, mathematics, computer science, and engineering.

It delivered anti-vitalism. By showing that more and more life processes could be explained materially, machine metaphors protected biology from vague appeals to mystic force. It insisted that living systems are natural systems.

It delivered technological power. Modern biotechnology, synthetic biology, genomics, and computational biology would be unimaginable without mechanistic abstractions. To sequence, edit, model, engineer, and simulate biological systems, one needs machine-like descriptions.

These achievements should not be minimized. A critique of machine metaphors that cannot explain its success is not a critique but a gesture. The machine metaphor became dominant because it made biology productive. It turned the organism into something that could be investigated, manipulated, and rebuilt. But productivity is not the same as universality. A metaphor can be indispensable within a domain and misleading when inflated into metaphysics. This is the central distinction. Machine metaphors are one of the most powerful epistemic tools in the history of science. It is not therefore the final ontology of life.

The hidden metaphysics of machine metaphors

Machine metaphors carry a metaphysics, even when scientists do not intend it to. It tends to assume that systems are composed of parts whose identities are stable, that relations among parts are external, that functions can be assigned locally, that behavior follows from rules, that organization can be decomposed, that purposes come from outside, and that explanation consists in reconstructing how components generate outputs.

This metaphysics fits many artifacts because artifacts are built that way. A clock is assembled from parts that can be manufactured independently. A piston can exist outside an engine. A circuit component can be stored in a drawer. A line of code can be copied from one program to another. A machine’s purpose is imposed by its designer or user. Its parts have functions because of the role they play in an externally defined task.

Organisms complicate each of these assumptions. Their parts are produced within the organization they help sustain. Their identities depend on context. A heart outside an organism is not a pump in the same biological sense; it is tissue losing the conditions of its function. A gene does not have a fixed meaning independent of regulatory, developmental, cellular, ecological, and evolutionary context. A cell is not assembled from independently manufactured parts; it continuously produces and replaces the components that sustain it.

This is why machine metaphors have always oscillated between illumination and distortion. They reveal local mechanisms but struggle with self-production. They explain pathways but struggle with historical novelty. They capture control but struggle with intrinsic normativity. They formalize information but struggle with meaning. They support intervention but often underestimate adaptation, evolution, and context-dependence.

The history reconstructed here shows that these problems are not accidental. They are inherited from the earliest narrowing of causation. Aristotle’s four causes allowed material, formal, efficient, and final aspects of explanation to coexist. Modern mechanicism gained power by privileging efficient causes. Later machine metaphors had to reintroduce the excluded dimensions in mechanized format: form as program, finality as feedback, materiality as substrate, purpose as control, history as stored information, development as execution. This strategy worked, but only partially.

The problem is not that machine metaphors fail to capture anything about life. The problem is that they often capture formerly excluded dimensions by translating them into the grammar of externally designed artifacts. Biological organization then appears as if it were a machine whose designer has disappeared. The question for the last stage of the present trilogy is whether “mechanism” can be expanded without returning to vitalism.

Can we develop a concept of mechanism in which constraints are not merely imposed from outside but produced and maintained by the system itself? Can we understand biological organization as materially open, efficiently closed, historically plastic, and syntactically open to novelty? Can we build a science of life that remains experimentally rigorous while refusing to reduce organisms to devices with fixed parts, fixed rules, and externally assigned purposes?

Conclusion

Machine metaphors have shaped biology because they offered something life sciences desperately needed: intelligibility. They translated living processes into forms that could be analyzed, measured, controlled, and rebuilt. The clock metaphor made order mechanical. The engine metaphor made metabolism energetic. The computer metaphor made control informational. Darwin made design historical. Molecular biology made heredity programmable. Synthetic biology made life engineerable. Artificial Life made possible life thinkable beyond carbon. Pancomputationalism turned the metaphor into a cosmogony.

This trajectory is not merely a history of mistaken analogies. It is a history of scientific power. Without machine metaphors, modern biology would not be what it is. We would not have the same molecular genetics, physiology, genomics, biotechnology, systems biology, or synthetic biology. Machine metaphors earned their authority. But authority is not finality. The same history also shows that machine metaphors survive by changing the meaning of machine and mechanism.

Organisms were not always computers. Before that, they were engines. Before that, they were clocks. Before that, the central problem was not machinery at all but self-motion, form, and purpose. Every machine metaphor has an expiration date because every machine expresses the technological imagination and sophistication of its time. When the dominant machines change, the metaphors change. What appears today as the final language of life may tomorrow look like another historical projection.

The task, then, is not to abolish machine metaphors. That would be impossible and undesirable. Science needs metaphors because complex phenomena do not interpret themselves. The task is to prevent metaphors from becoming metaphysics by inertia. We must remember that organisms are not simply devices waiting to be reverse engineered. They are self-producing, historically constituted, environmentally embedded, norm-generating processes whose parts exist through the organization they help maintain.

Machine metaphors remain one of biology’s most successful conceptual tools. But it is a tool, not a destiny. Its success explains why it became the current paradigm of science. Its limits will explain why biology now needs a transformed notion of mechanism: one capable of preserving the clarity of machines while doing justice to the strange, precarious, self-maintaining, and open-ended organization of life. That is where the second part of this trilogy must begin.