I. The problem with a single ground
Cognitive linguistics achieved something rare in the human sciences. It took an intuition that everyone had half-noticed, that we speak of abstract things in concrete terms, and turned it into a research programme with predictive reach. George Lakoff and Mark Johnson showed that when we say an argument has no foundation, that a relationship is going nowhere, that prices are rising or that someone is feeling down, we are not decorating literal thought with optional ornament. We are thinking through the ornament. The abstract concept is structured by the concrete one. Argument is understood as war, love as journey, quantity as verticality, mood as elevation. The metaphors are systematic, they cluster into families, and they organise inference. If time is a resource then it can be spent, saved, wasted, borrowed, and run out. The entailments travel together because the mapping travels together.
Nothing in what follows disputes that finding. The dispute concerns the answer to a question that conceptual metaphor theory never posed with enough precision. Where does the concrete come from? What, exactly, is the reservoir from which metaphor draws, and why does the reservoir have the contents it has rather than others?
Lakoff and Johnson gave an answer that has hardened, over four decades, into something close to a slogan. The source is the body. Abstract concepts are grounded in embodied experience. Image schemas such as container, path, source-path-goal, balance, force, up and down, centre and periphery are said to arise from the recurring structure of bodily life, from having an inside and an outside, from moving through space, from standing upright against gravity, from pushing and being pushed. In the later work this becomes an explicit metaphysical commitment. Philosophy itself, on their account, is embodied, and the failure of the Western tradition is precisely its refusal to see that reason is shaped by the flesh.
This article argues that the answer is wrong, and wrong in an instructive way. It is not wrong because the body is unimportant. It is wrong because the evidence that Lakoff and Johnson themselves assembled, the largest and most carefully analysed corpus of conceptual metaphors in existence, does not show what they concluded it shows. Read with the right question in mind, their corpus demonstrates a different principle altogether, one that they had no conceptual vocabulary to see. The present article supplies that vocabulary and states the principle. Put as a claim about the history of the field: Lakoff and Johnson discovered one of the deepest empirical regularities of ordinary language, and misidentified the principle that generates it.
The governing thesis can be given at the outset, because everything that follows is derived from it and then tested against it. The source of ordinary conceptual language is the nonrecursive organisation of the mesocosm. Ordinary language evolved to coordinate human life in the mesocosm, the middle-scale world of bodies, others, places, and things in which all human coordination takes place. Its conceptual resources can therefore come from nowhere else. But the mesocosm contributes those resources asymmetrically. Its recursive processes, the self-modifying and mutually modifying processes of living, feeling, and coordinating with others, cannot serve as stable symbolic anchors, precisely because they are what symbolisation is trying to stabilise. Its nonrecursive organisation can. Containers, paths, boundaries, gravity, seasons, tools, pressure, fire, and the invariant topology of the human body itself supply the anchors, and recursive life supplies the targets. Symbolisation recruits stability in order to think about change.
The thesis is stated without hedging because it does not describe a statistical tendency. It follows from the ontology of Living Value Theory, within which this article is written, and specifically from a principle that theory has established independently in economics, medicine, law, and the analysis of institutions: recursive processes exceed complete symbolisation, and symbolic systems succeed to the degree that they anchor themselves in what does not recurse. The thesis is nevertheless falsifiable, and the article ends by saying exactly what would refute it. A single well-attested family of conceptual metaphors whose source domain is genuinely interrecursive, a metaphor that renders one dynamic coordination intelligible through another dynamic coordination, would break the principle. The article predicts that no such family will be found, in Lakoff and Johnson's corpus or in any other.
The argument proceeds in stages. The next two sections set out the architecture: what the mesocosm is, why its five mediations are the necessary and complete inventory of its organisation, why ordinary language belongs to the fifth mediation and can therefore draw only on the other four, and why, within those four, only nonrecursive organisation can serve as a symbolic source. The middle sections then take conceptual metaphor theory's own flagship evidence, the image schemas, the emotion metaphors, the argument-is-war family, and the metaphors of time, and show that every one of them confirms the principle against the theory that assembled them. A further section explains the most striking absence in the entire metaphorical record, the near-total silence of social coordination as a source domain, an absence that embodiment theory cannot explain and the present account predicts. The final sections turn the principle into a method for reading discourses by their symbolic anchors, restate the historical theory of metaphor that follows, and formulate the empirical prediction on which the whole account stakes itself.
II. The mesocosm as the source of ordinary language
The argument requires a brief statement of the ontology it draws on, and the statement can be brief because only a small part of the apparatus of Living Value Theory is needed here.
Human life is coordinated in the mesocosm, the middle-scale world that lies between the microphysical and the cosmological and in which every human being conducts every moment of an ordinary life. The mesocosm is not a container. It is the ongoing coordination of living beings with their world, and that coordination has exactly five irreducible forms, which the theory calls mediations. Human beings coordinate through multisensorial embodiment, the register of perceiving, moving, orienting, and sensing. They coordinate through multispecies being-with, the register of living among other recursive beings, human and nonhuman. They coordinate through multiversal dwelling, the register of inhabiting a place and a world, with its ground and sky, its seasons and cycles, its paths and boundaries. They coordinate through multimateriality, the register of handling substances, making and breaking things, working with tools and materials. And they coordinate through multisymbolism, the register of signs, words, images, notations, and every other symbolic technology.
Two features of this inventory matter for everything that follows, and both distinguish it in kind from the inventories of cognitive linguistics.
The first is that the five mediations are ontologically derived, not empirically accumulated. Lakoff's catalogue of image schemas is explicitly open-ended. He presents container, path, link, force, balance, and the rest as the schemas that linguistic analysis has so far uncovered, and the list remains permanently hospitable to additions, because nothing governs it except recurrence in the data. The mediations are a different kind of claim. They are derived from the structure of mesocosmic coordination itself, and the derivation entails that there cannot be fewer, because each mediation is irreducible to the others, and there cannot be more, because any proposed sixth mediation will on examination either reduce to one of the five or turn out to be a particular combination of them. The five are not a taxonomy awaiting expansion. They are the complete architecture of the mesocosm, and the present article treats them as such throughout. Nothing in what follows is offered as provisional pending the discovery of further mediations, because on the theory being applied there are no further mediations to discover.
The second feature is that the mediations are irreducibly co-present. Actual mesocosmic coordination never occurs in one mediation at a time. Every moment of ordinary life is simultaneously embodied, situated among others, emplaced in a world, entangled with materials, and saturated with symbols. The mediations are not compartments among which experience is divided. They are the always-simultaneous dimensions of a single coordination, and any analysis that treats them as separate reservoirs from which language draws in turn has already falsified the ontology. What varies from case to case is not which mediations are present but which mediation is foregrounded. This will matter when the article turns to particular metaphors, because the right analytical question is never which mediation a metaphor comes from, as though the others were absent, but which mediation a metaphor foregrounds against the background of an irreducibly multimediated source.
From this architecture, the location of ordinary language follows immediately, and with it the first structural claim of the article. Ordinary language is a symbolic technology. It belongs to multisymbolism. It is, indeed, the oldest and most pervasive of the technologies that constitute that mediation. And this settles a question that would otherwise look like an arbitrary choice: why the account of metaphorical sources that follows involves four mediations rather than five. Multisymbolism cannot appear among the sources of ordinary conceptual language because ordinary conceptual language is what multisymbolism, in its linguistic form, is. The fifth mediation is not missing from the inventory of sources. It is the target. The architecture of the whole account is therefore asymmetrical by necessity. The mesocosm, coordinated through the four non-symbolic mediations, is the source. Ordinary language, belonging to the fifth mediation, is what gets generated. Conceptual metaphors, and with them the conceptual resources of abstract thought, are what ordinary language in turn makes possible.
Stated as a sequence: the mesocosm gives rise, through the four non-symbolic mediations, to ordinary language, and ordinary language gives rise to conceptual thought. Metaphor is one especially visible span of that sequence, the place where the derivation of the symbolic from the non-symbolic can still be watched happening, because in a live metaphor the source has not yet been forgotten.
This already reframes the question that conceptual metaphor theory asked. Lakoff and Johnson asked where metaphors come from. The prior question is where ordinary language comes from, and once the prior question is answered, the first answers itself. If ordinary language emerges from mesocosmic coordination, then metaphors, and equally analogies, classifications, exemplars, and explanatory models, will necessarily recruit mediational resources, because there is nowhere else for them to come from. A theory of metaphor is downstream of a theory of the mesocosmic origins of language, and the deepest error of the embodiment account was to build the downstream theory without the upstream one, so that a single mediation, glimpsed accurately but in isolation, was mistaken for the whole ground.
But the mediational architecture, by itself, is not yet the discovery this article reports. If the argument stopped here, it would say that metaphor draws on four registers rather than one, and that would be an improvement of detail rather than of principle. The discovery lies one level deeper, in a distinction that cuts across all the mediations, and to state it the ontology needs one further element.
III. The master asymmetry: nonrecursive source, recursive target
Living Value Theory distinguishes three types of process and entity within the mesocosm, according to their recursivity. Nonrecursive entities and processes do not modify themselves through their own operation. A stone, a path, a boundary, the fall of a heavy object, the boiling of water under heat, the alternation of day and night, the topology of a container: these behave the same way regardless of what has happened to them before, or change only on scales irrelevant to the coordination at hand. selfrecursive processes modify themselves through their own operation. A living body regulating its own states, a memory reshaping itself in being recalled, an emotion feeding on its own expression: these are processes whose present operation is continuously conditioned by their own past operation. Interrecursive processes are mutually modifying coordinations between recursive beings. A conversation, a friendship, an argument, a negotiation, a market, an apprenticeship: in each of these, every move changes the conditions of the next move for all parties, and the process has no state that is not already a response to responses.
The theory has argued, across its treatments of economics, medicine, law, and institutions, that recursive processes cannot be completely symbolised. A symbol stabilises reference. It picks something out and holds it still long enough for coordination at a distance, coordination across time, coordination among strangers. But selfrecursive and interrecursive processes are precisely what does not hold still. They are ongoing, self-modifying, and open-ended, and any symbol that captures their state has captured a state they are already leaving. This is why, in domain after domain, the theory has found the same solution being improvised. Money symbolises the interrecursive process of valuing by attaching to relatively stable nonrecursive artefacts. Diagnostic categories symbolise the selfrecursive process of living with illness by attaching to relatively stable signs, biomarkers, and anatomical structures. Legal forms symbolise ongoing relationships by attaching to fixed categories and documents. In every case the recursive process exceeds its symbols, and in every case symbolisation succeeds, to the degree that it succeeds at all, by anchoring itself in what does not recurse.
From these commitments the central principle of this article follows deductively, before a single metaphor has been examined. Recursive processes cannot be completely symbolised. Symbolisation therefore requires stable symbolic anchors. Only nonrecursive organisation provides stable symbolic anchors. Therefore the source of ordinary language must be the nonrecursive organisation of the mesocosm. Each step is supplied by the ontology, and the conclusion owes nothing to the metaphorical record, which matters for the logical standing of everything that follows. Lakoff and Johnson's corpus is not the evidence that establishes the principle. It is the empirical confirmation of a principle already derived, and the confirmation carries the particular weight it does because the corpus was assembled by researchers who did not possess the principle and who read their own data as demonstrating something else.
State the principle in its general form. Symbolisation recruits nonrecursive organisation in order to render recursive organisation intelligible. Stability is the necessary enabling condition of a symbolic source. A source domain must hold still while the target is thought through it, and only the nonrecursive holds still. It follows that the source domains of conceptual metaphor will be drawn from the nonrecursive organisation of the mesocosm, wherever in the four mediations that organisation is found, and that the targets of conceptual metaphor will be the recursive processes of the mesocosm, the emotions, relationships, arguments, institutions, and temporal experiences that constitute recursive life. The asymmetry runs one way. Recursive life is what needs symbolising and cannot supply the anchors. Nonrecursive organisation supplies the anchors and rarely needs the symbolising, because what does not change does not generate the same pressure to be rendered thinkable.
Notice what kind of claim this is. It is not the claim that concrete things make good metaphors, which is a truism, and it is not the claim that source domains are usually physical, which is an observation. Concreteness is not the operative property. Stability is. A cloud is concrete and physical and supplies few conceptual metaphors, because clouds are unstable. A boundary is barely a thing at all and supplies an enormous family of them, because a boundary is maximally stable organisation. This substitution of stability for concreteness is what distinguishes the present principle from every version of the claim that abstract thought is grounded in concrete experience, and it is what allows the principle to explain a fact that the concreteness story cannot: mathematics, the least concrete domain human beings possess, is among the most productive sources of metaphor in existence, because formal structures are maximally stable. The principle is not that thought reaches down to the tangible. It is that symbolisation reaches out to the invariant.
And notice, finally, what the principle does to the notion of experience on which embodiment theory rests. Lakoff and Johnson say that image schemas arise from recurring embodied experience. But recurrence of experience explains almost nothing, because recursive experiences also recur, constantly. People repeatedly experience conversations, negotiations, embarrassment, teaching, deception, reconciliation, jealousy, mourning, and cooperation, every day of every life, and these recurring experiences almost never become source domains. If recurrence in experience made a source, the metaphorical record would be full of them. It is empty of them. What the record is full of is containers, paths, forces, pressures, boundaries, and verticality, and what those share is not that they are experienced more often but that they are organised more stably. The image schemas are not distillations of experience. They are the stable organisational features of the mesocosm that recursive experience continuously presupposes, and they enter language as sources because they, unlike the experience that presupposes them, can hold a symbolic anchor.
The remainder of the article takes the derived principle to the place where its confirmation ought to be hardest to obtain: the canonical evidence of conceptual metaphor theory itself. If the derivation is sound, then the corpus assembled to demonstrate embodiment should turn out, case by case and without exception, to demonstrate the recruitment of nonrecursive organisation instead. The embodiment metaphors should recruit the body's nonrecursive anatomy rather than its recursive life. The emotion metaphors should recruit physical processes rather than other emotions. The flagship argument-is-war family should recruit the material and spatial organisation of warfare rather than its social reality. The time metaphors should recruit celestial and seasonal invariants at least as readily as bodily motion. And the interrecursive world, the world of trust and friendship and negotiation, should appear on the target side of every mapping and on the source side of none. Each of these is what the following sections find.
IV. The body reread: anatomy, not experience
Begin where the received theory is strongest, with the metaphors it calls embodied, because if the principle holds here it holds everywhere.
Johnson's image schemas are presented as the recurring structures of bodily experience. Consider what they actually are. Inside and outside. Up and down. Front and back. Left and right. Balance. Containment. Contact. Support. Now ask, of each, whether it is a recursive bodily process or a nonrecursive bodily structure, and the answer is uniform. The body's inside and outside are not experiences. They are an anatomical topology of remarkable invariance. My heart does not migrate to my foot. My skin does not intermittently cease to be a boundary. When I say that a thought is inside me or that I have put something outside my mind, the metaphor works because the boundary it recruits is fixed, not because being bounded is vividly felt. Up and down are not bodily experiences either. They are the stable geometry of gravity and upright posture, invariant across every moment of terrestrial life, and it is exactly this invariance that lets happy be up and sad be down in language after language. Left and right, front and back, are anatomical constants. Balance is the one schema with a genuinely dynamic component, and even there what the metaphor recruits is the stable condition of equilibrium against which the dynamics play out, which is why a balanced argument is a settled one and not a wobbling one.
The point can be made most sharply through a simple observation. Nothing about the metaphorical use of these schemas changes the body, and nothing about the body's recursive life changes the schemas. I do not start walking on my head because my mood is described as down. My anatomy is indifferent to every symbolic use made of it, and that indifference is not incidental to the symbolic use. It is the precondition of it. The body serves ordinary language not as a sensing, self-regulating, recursively living organism but as a piece of invariant mesocosmic architecture, a portable set of stable orientations and boundaries that every speaker carries everywhere and that never lets the symbolism down by changing.
This is a different claim from the claim that Lakoff and Johnson overstate embodiment, and it needs to be distinguished from it carefully. The claim is that they misidentify what, in the body, does the symbolic work. They say the source is embodied experience. But experience, in any sense that phenomenology would recognise, is exactly what these schemas are not. The lived feeling of being inside a room, the vertigo of height, the strain of keeping balance: these vary enormously, moment to moment and person to person, and the metaphors are wholly insensitive to that variation. What the metaphors track is the organisation that the varying experience presupposes, the boundary that makes inside-feelings possible, the gravitational axis that makes vertigo possible, the equilibrium that makes strain possible. The source of the so-called embodiment metaphors is the nonrecursive organisation of the body, which is to say the least living thing about it.
Their own favourite phrase gives the game away. Lakoff and Johnson repeatedly describe the ground of the image schemas as the body in space, and the phrase deserves more scrutiny than it has received, from them or from their critics. First, the moment one says in space, one has exceeded embodiment. A body in space is already a body dwelling, already oriented within an inhabited world, and their examples confirm it: the schemas are bodies in containers, bodies on paths, bodies crossing boundaries, bodies entering and leaving places. The evidence points beyond the body at every step, and the ontology keeps folding it back in. In the terms of the present account, the body in space is already a multimediation, embodiment and dwelling operating simultaneously and inseparably, exactly as the co-presence of the mediations requires. But second, and more deeply, neither embodiment nor dwelling, as registers of coordination, explains the symbolic productivity of the schemas. What explains it is the property the two registers share at exactly the points the schemas recruit: invariance. The body contributes its anatomy, not its metabolism. The dwelt world contributes its paths and boundaries, not its weather. The phrase the body in space names, without recognising it, the intersection of two mediations at their most nonrecursive, and that intersection, not the experience of either, is where the image schemas live.
V. The image schemas as evidence
Once the schemas are reread this way, the catalogue as a whole becomes evidence of a kind that Lakoff never claimed for it, because it answers a question he never asked.
His question was where the schemas come from, and his answer, recurring embodied experience, has already been examined. The unasked question is why the catalogue has these members and no others, why it is so narrow, and why its narrowness has the particular shape it has. Set out the canonical inventory: container, path, source-path-goal, link, force, blockage, balance, support, contact, surface, centre and periphery, verticality, cycle. Thirteen or so schemas, endlessly recombined, carrying the metaphorical weight of entire languages. Every one of them is a stable organisation. Not one of them is a recursive process. There is no conversation schema, no negotiation schema, no friendship schema, no teaching schema, no mourning schema, although conversations, negotiations, friendships, teaching, and mourning are among the most frequent, most consequential, and most intensely attended experiences in any human life. The inventory is not a fair sample of experience. It is a census of invariants.
The embodiment account has no explanation for this shape. If experience were the source, the recursive experiences, which are the ones that matter most to people and recur most insistently, should dominate the inventory, and they are absent from it. The present account predicts the shape exactly. Recursive processes cannot serve as symbolic anchors, so they cannot become schemas, however often they are experienced. Stable organisations can, so the inventory consists of nothing else. The catalogue that cognitive linguistics assembled empirically, schema by schema, without a governing principle, turns out to have had a governing principle all along. It is a list of the mesocosm's most reliable nonrecursive structures, filtered out of the flux of experience by the requirements of symbolisation itself. Lakoff could describe the inventory but not derive it. The recursivity asymmetry derives it.
Put the point at its full strength: under the principle, the shape of the corpus becomes inevitable. Why containers? Why paths? Why bridges, foundations, battlefields, pressure vessels, seasons, celestial bodies? Because these are precisely the kinds of stable nonrecursive organisation that the mesocosm provides, and a symbolic technology in need of anchors had nowhere else to go. Given the ontology, one could have written out the inventory in advance, without opening a single text: the invariant topology of the body, the enduring structures of the dwelt world, the reliable behaviour of materials and made things, and nothing from the flux of coordination itself. That is what the catalogue contains, and all that it contains. Lakoff could catalogue the inventory. The principle explains why the catalogue has exactly this shape and could have had no other.
It also derives the internal proportions of the metaphorical record, which the embodiment account likewise leaves unexplained. Among the four source mediations, productivity is not equal, and the inequality follows the distribution of nonrecursive organisation. Multiversal dwelling is enormously productive, because the dwelt world is dense with invariants: paths, boundaries, grounds, horizons, enclosures, heights, centres, edges, and the celestial and seasonal cycles to which the article returns. Multimateriality is equally productive, because the made world is dense with them too: tools, buildings, foundations, bridges, containers, liquids, fires, pressures, machines. Embodiment is productive precisely to the extent of its anatomy and no further, which is why its contribution, on inspection, keeps reducing to the same short list of orientational constants. And being-with, the mediation richest in recursive coordination and poorest in stable structure, contributes almost nothing, an absence so systematic that it receives its own section below. The proportions are not an accident of linguistic history. They are the map of where, in the mesocosm, stability is to be found.
VI. The emotion metaphors: pressure vessels, not feelings
The metaphors of emotion are the received theory's showcase of embodiment, because emotion is bodily if anything is, and so they make a clean test.
He is about to burst. She is letting off steam. He is boiling with rage. She was frozen with fear. That blew my mind. He exploded. She simmered for days. He erupted. Their anger evaporated. Take the sources in turn. A container failing under internal pressure. Steam escaping a pressure vessel. Liquid at a phase transition under heat. Matter at a phase transition under cold. A detonation. Another detonation. Sustained sub-boiling heat. A volcanic discharge. A liquid passing into gas. Not one of these sources is an emotion. Not one is a bodily experience. Every one is a nonrecursive material process, drawn from the physics and domestic technology of heat, pressure, and phase change, processes whose course is fixed by their conditions and identical at every occurrence. Water boils the same way every time. That is why boiling can anchor rage, which never happens the same way twice.
The pattern of the mapping is the principle in miniature. The target, emotional life, is selfrecursive through and through: feeling feeds on its own expression, dread compounds itself, anger rehearsed is anger enlarged. The source is drawn from the least recursive processes the mesocosm offers. And the direction never reverses. No one renders the behaviour of steam intelligible through jealousy. The recursive is symbolised through the nonrecursive, never the other way around, because the asymmetry of stability runs only one way.
It is worth pausing on what the embodiment reading has to say about these cases, because the contrast is instructive. On that reading, boiling with rage is grounded in the felt heat of anger, the physiological arousal that emotion involves, so the source is bodily experience after all. But this gets the anatomy of the metaphor wrong. The vocabulary is not the vocabulary of arousal. It is the vocabulary of vessels, lids, valves, pressure, and boiling points, a technology of containment that no body possesses. The felt heat of anger may explain why heat was available as a mapping at all, but what the metaphor recruits and elaborates is the behaviour of heated liquids in containers, a nonrecursive material process, and the elaboration is where the conceptual work happens: the pressure builds, the lid comes off, the contents spill, the vessel bursts. A theory that stops at the felt heat cannot explain the vessel, and the vessel is the metaphor.
VII. Argument is war: the flagship confirms the principle
Argument is war is the first extended example in Metaphors We Live By and the most famous conceptual metaphor in the literature. If the present account is right, it should be confirmable here, at the centre of the theory it corrects, and it is.
Rehearse the familiar examples. He shot down my argument. Your claims are indefensible. I attacked every weak point. His criticisms were right on target. I demolished his argument. We defended our position. He came under fire. We had to retreat. He ceded ground. On the received reading, the abstract domain of argument is structured by the concrete domain of war, and the case demonstrates that even our most intellectual activities are conceptualised through physical conflict.
Now apply the recursivity analysis, and begin with the target. An argument is interrecursivity in nearly its purest ordinary form. Two or more recursive beings, each move changing the space of possible next moves for everyone, no state of the exchange that is not already a response to responses, no script survivable past the first reply. If any everyday phenomenon epitomises what Living Value Theory means by interrecursive coordination, argument is it. The target could not be more recursive.
Then look, item by item, at what actually does the symbolic work on the source side, and a remarkable fact emerges. It is not war as a social phenomenon. War, after all, is itself interrecursive, a mutually modifying coordination between opposed parties, and if war in that sense were the source, the metaphor would be stabilising one interrecursion with another, which the principle says cannot work. And indeed that is not what the vocabulary recruits. To come under fire is to be in the path of physical projectiles. To cede ground is to give up territory. To defend a position is to hold a fortification on stable ground. The inventory of the mapping is the battlefield, the front, the trench, the fortress, the weapon, the projectile, the target, the territory, the line of advance and retreat. Every item is a piece of war's material and spatial organisation, its multimateriality and its dwelling-structure, and not one item is war's interrecursive reality, the mutually modifying contest of commanders and armies. The metaphor does not borrow warfare. It borrows warfare's furniture.
This resolves a question the received theory never asks: why war, of all social institutions, should be so metaphorically productive when other social institutions are so barren. Friendship, marriage, teaching, and negotiation are as familiar as war and far more frequently experienced, and they supply almost no source domains. The answer is that war is unusual among social phenomena in trailing an enormous inventory of stable nonrecursive organisation, fronts, territories, fortifications, weapons, supply lines, and it is that inventory, not the fighting, that language raids. War gets into the metaphorical record for the same reason buildings and rivers do, and its social reality gets in no further than theirs. The flagship example of conceptual metaphor theory, examined at the level of what its mappings actually recruit, is a demonstration of the recursivity asymmetry, executed forty years before the asymmetry was formulated.
VIII. Time: dwelling's cycles, not the moving body
The metaphors of time occupy a special place in the received theory, because Lakoff and Johnson use them to argue that some domains cannot be conceptualised except metaphorically. Time, they hold, is not directly experienced as an object, so it must be understood through something else, and the something else is motion through space. Christmas is coming. We are approaching the deadline. The weeks flew by. Time is a moving object, or we are moving observers, and either way the ground is the body's locomotion along a path. The treatment of time is thus embodiment theory at its most explicit: even temporality reduces to the moving body in space.
Half of this is a direct confirmation of the present account, and the confirmation should be registered before the correction. Temporality as lived is deeply recursive. The experienced present is continuously reshaped by retention and anticipation; memory rewrites itself in being consulted; the felt pace of a life is a self-modifying process if anything is. And, exactly as the principle predicts, this recursive target is never symbolised through another recursive process. It is symbolised through nonrecursive structure: paths, motion of objects, fixed landmarks ahead and behind. On the question of what kind of source time recruits, Lakoff and Johnson's finding is the finding of this article.
The correction concerns which nonrecursive structures, and here their reduction to the moving body is not merely too narrow but demonstrably false to the record, including the ordinary record of English. Consider the temporal metaphors that any speaker commands. Spring into summer. Winter is coming. The year is turning. Three moons ago. Before the harvest. After the thaw. Under a full moon. The dawn of a new era. The twilight years. The dawning of the Age of Aquarius. None of these recruits bodily locomotion. None requires a source-path-goal schema. Several do not involve the body's organisation at all, even as orientation. What they recruit are the stable cycles of the inhabited world: the alternation of seasons, the phases of the moon, the transit of the sun, the rhythms of cultivation, the configurations of the sky. These are the invariants of multiversal dwelling, and they anchor temporal language directly, without passing through the body.
The last example repays a moment's attention, because it is a small demolition in itself. The dawning of the Age of Aquarius recruits two dwelling-structures simultaneously, the solar cycle in dawning and the zodiacal order in Aquarius, and nothing else. There is no body in the mapping and no path. A theory obliged to trace every temporal metaphor to bodily motion has nothing defensible to say about it, and the same is true of three moons ago, where the source is a celestial cycle whose regularity was, for most of human history, the very infrastructure of temporal coordination.
That historical observation opens onto something larger, which section X develops: expressions like three moons ago are linguistic fossils, survivals from mesocosms in which lunar and seasonal cycles were the recursively relevant invariants of everyday life. Their existence shows that temporal metaphor tracks whatever stable structure a given world makes relevant, not a universal bodily schema. But even synchronically, within present-day English, the correction stands. Time recruits nonrecursive organisation, exactly as Lakoff and Johnson found. It recruits it from wherever in the mesocosm such organisation is found, from the moving-object schemas they analysed, from the seasonal and celestial cycles they overlooked, and, in carving out time, spending time, and running out of time, from the material and economic artefacts of multimateriality as well. The uniformity is in the recursivity type of the source, not in its mediation, and mistaking the second for the first is precisely the error that a theory equipped only with the category of embodiment was bound to make.
IX. The missing source: why being-with supplies almost nothing
Every account of metaphorical sources must face not only what the record contains but what it lacks, and the record has one absence so large and so systematic that it constitutes, once noticed, the single strongest piece of evidence in this article.
Interrecursive life supplies almost no metaphorical source domains. The domain of human coordination with others, of trust and betrayal, friendship and rivalry, teaching and following, negotiation and reconciliation, is the domain human beings know most intimately, attend to most constantly, and care about most. On any experiential theory of metaphor it should be the richest source in the language. It is nearly the emptiest. Speakers do not render arguments intelligible through friendships, or time through negotiations, or emotions through apprenticeships. There is no significant metaphor family of the form X is trust, X is a conversation, X is teaching, where the social phenomenon does the anchoring. The interrecursive world appears on the target side of the metaphorical record constantly, as the thing rendered thinkable through bridges, paths, wars, containers, and storms, and on the source side it barely appears at all.
Two apparent classes of counterexample dissolve on inspection, and the manner of their dissolving is itself instructive. The first class is ordinary social vocabulary: trust, promise, authority, recognition, betrayal. These are sometimes gestured at as evidence that social experience feeds language, and of course it does, but feeding language is not supplying metaphors. Trust is not a metaphor. It is a word, the direct name of an interrecursive phenomenon, and the distinction between naming a domain and serving as a source domain is the distinction on which the whole of conceptual metaphor theory rests. That interrecursive life is richly named shows only that people talk about it, which was never in doubt. The question is whether it anchors the understanding of other domains, and it does not. The second class is expressions like following a leader or standing by someone, which look social until their mappings are examined, at which point they resolve into the body in space once again: locomotion behind a moving figure, stable co-location. The apparent social sources are dwelling and anatomy in social costume, and the costume comes off under the provenance test that section XI states.
The embodiment account cannot explain this absence and, revealingly, has never tried; the social world simply does not figure in its inventory of sources, and the omission passes without comment, as though it were natural. The present account predicts the absence as a theorem. Interrecursive coordination is the maximally recursive region of the mesocosm. It is therefore the region least capable of anchoring a symbol, for exactly the reason it is the region most in need of symbolisation. A metaphor that recruited a genuinely interrecursive source would be attempting to stabilise one open-ended mutual modification by means of another, and there is nothing stable in the transaction for the mapping to grip. The mediation of being-with is not thereby diminished. It is indispensable to the mesocosm, and it is present, co-present, in every coordination from which language arises. But its contribution to symbolisation is made from the target side. Being-with is what the metaphors are for, not what they are made of, and the one-sidedness of its position in the record is the recursivity asymmetry writing itself across the entire lexicon.
It should be conceded, because the concession sharpens the claim, that being-with is not absolutely without stable structure. Ritual and institution exist precisely to introduce nonrecursive fixity into social life, and where they succeed, language notices: an argument can be a trial, a courtship a dance with steps. But observe what has happened in such cases. What gets recruited is the fixed form, the procedure, the choreography, the office, which is to say the nonrecursive deposit that ritual has precipitated out of interrecursive flux, and never the flux itself. The exception obeys the principle, and obeys it conspicuously: even when social life does supply a source, it supplies it only through whatever in it has stopped recursing.
X. The historical dimension: recursive relevance and the drift of repertoires
A theory of where metaphors come from ought to explain not only the synchronic record but its history, and here the two accounts come apart completely, because the embodiment account entails that the deep sources of metaphor are historically constant, the body being everywhere and always the same, while the present account entails that they drift, tracking the transformations of the mesocosm itself.
The relevant concept is recursive relevance. At any historical moment, certain nonrecursive structures of the mesocosm are load-bearing for coordination: life is scheduled by them, work is organised around them, survival depends on attending to them. These recursively relevant invariants are the ones that saturate attention, and they are therefore the ones that symbolisation recruits. In mesocosms coordinated by the sky and the seasons, the moon, the dawn, the harvest, and the thaw were the great invariants of everyday life, and the temporal language of section VIII is their deposit. In mesocosms coordinated by craft and cultivation, the forge, the loom, the field, and the yoke supplied the anchors, and forging alliances, weaving narratives, sowing doubt, and shouldering burdens are theirs. The industrial mesocosm made engines, pressure, production, and circulation recursively relevant, and the languages of feeling and of knowledge alike filled with pressure vessels, outputs, and flows. The digital mesocosm is performing the same substitution now, as bandwidth, storage, platforms, and networks harden into the anchors through which minds, institutions, and relationships are rendered thinkable.
The principle is constant across all of these; the inventory is not. Symbolisation always recruits nonrecursive organisation, and it always recruits it from among the structures that the current mesocosm has made relevant. This yields a historical theory of metaphor with real predictive content: metaphorical repertoires will shift when and only when the recursively relevant invariants of the mesocosm shift, and the lag between the two, the survival of three moons ago into a world of clocks, gives the lexicon its archaeological stratification, live metaphors above, fossils below. Embodiment theory has no purchase on any of this. A source fixed in the universal body cannot drift, and the manifest drift of the record is, on that account, noise. On the present account it is the signal, and it explains cross-cultural variation by the same stroke: different mesocosms, different relevant invariants, different repertoires, one principle.
XI. From principle to method: reading discourses by their anchors
A principle of this generality would remain philosophy unless it issued in a procedure, and it issues in one directly, because the principle identifies, for any body of language, a determinate question with a determinate answer: from which nonrecursive structures does this discourse draw the anchors through which it thinks its objects?
The working criterion is provenance. For any metaphorical term, do not ask what it could be reduced to if one were determined to reduce it, which is the question that let embodiment theory talk everything back to the body. Ask where the source term does its literal work, in what activity a competent speaker uses it non-metaphorically, and identify the nonrecursive structure operating there. Forge does its literal work at the smithy; its anchor is the metallurgical process. Navigate does its literal work in wayfinding; its anchor is the stable relation of route and landmark. Boiling does its literal work at the hearth; its anchor is a phase transition. The provenance question is answerable intersubjectively for the great majority of cases, the straddlers can be recorded as straddlers without damaging the distribution, and the output, for any discourse, is a profile: the set of nonrecursive anchors through which that discourse habitually symbolises its recursive objects, weighted by use.
Profiles make discourses comparable, and comparison exposes a pathology that individual metaphors never could. Call a discourse captured when its profile has collapsed onto a single narrow family of anchors, so that one region of nonrecursive organisation has become the only structure through which the discourse can imagine its object. Capture is invisible from inside, because a monopolised anchor stops reading as a metaphor at all and starts reading as plain description, and the plainness is the symptom. The diagnosis is testable by counting, and the alternative renderings that other anchors would supply sound strange in proportion to the completeness of the capture, which makes the strangeness itself a measure.
A worked case shows the method operating, and shows what the recursivity analysis adds to it. Take the vocabulary in which the contemporary history of science speaks of its central object, knowledge, as sampled by the programme of any large conference in the field. Knowledge is produced, circulated, exchanged, transferred, transmitted, stored, accumulated, distributed, mobilised; it moves through networks and flows. Apply the provenance test term by term and the profile is stark. Every anchor is drawn from a single region of multimateriality, and a historically specific region at that: the production, logistics, and warehousing of goods in the industrial economy. The first finding is capture, and it carries an exact irony, because this is a field that has spent decades decentring Europe and recovering the knowledge of non-industrial peoples while describing everything it recovers in the conceptual grammar of the factory and the freight network, reflexive to its core about its objects and wholly unreflexive about its anchors. But the recursivity analysis locates the deeper finding beneath the irony. Knowing is a recursive process, selfrecursive in the knower and interrecursive among knowers, and like every recursive process it exceeds its symbols. The industrial vocabulary is one way of anchoring it, legitimate as far as it goes, illuminating what in knowledge does behave like movable stock, and blind to everything in knowledge that behaves like trust, like skill, or like inhabitation. Capture, understood through the principle, is not the use of a wrong anchor, since anchors are not true or false. It is the narrowing of a discourse's symbolisation of a recursive object to a single stable structure, when the object's recursivity is precisely what no single structure can render, and the remedy is not substitution but repluralisation: a field that could say, without strain, both that knowledge circulates and that knowledge dwells would be symbolising its object less incompletely than a field that can only say the first.
The method generalises without modification, because every discipline that studies recursive life faces the same representational problem and solves it with anchors that can be profiled. Economics thinks value through flows, equilibria, and bubbles; psychology thinks the mind through containers, mechanisms, and processors; medicine thinks illness through invasions, defences, and burdens; management thinks organisations through machines, pipelines, and levers; the discourse of artificial intelligence, with unusual density, thinks its object through training, weights, and architectures. In each case the profile can be computed, the capture diagnosed, and the analytical question asked in its sharpened form: not which mediation a discourse draws on, but which stable structures it has recruited to hold still an object that does not, and what those structures make it unable to see.
XII. Notation and demediation: why the exact sciences had to leave ordinary language
The principle has a further consequence, and it reaches beyond metaphor into the history of science, because it explains a fact about scientific notation that has otherwise only been observed: genuine notation systems are free of mesocosmic metaphor, and they are free of it by design.
Begin with what the principle entails about ordinary language as such. Ordinary language is mesocosmic through and through, and irreducibly so. It did not merely arise in the mesocosm and retain some traces of the origin. It is made of the mesocosm, its every conceptual resource recruited from the nonrecursive organisation of the four non-symbolic mediations, and this provenance cannot be purged from it, because it is not a contaminant in ordinary language. It is what ordinary language is. For the coordination of mesocosmic life this is no defect at all; it is exactly why ordinary language serves that coordination so well. But it creates a determinate and unavoidable problem for any field of enquiry whose domain lies outside the mesocosm, or whose ontology is nonrecursive in ways the mesocosm never exhibits. Chemistry and physics are the clearest cases. As long as such a science thinks in ordinary language, it smuggles mesocosmic metaphor into its domain with every sentence, not through carelessness but through the constitution of its medium. Particles attract. Forces act. Currents flow. Electrons occupy shells and form clouds. Atoms bond. Each of these imports the organisation of bodies, dwellings, and materials into a domain where none of it obtains, and no discipline of usage can stop the importing, because ordinary language has nothing else to offer. A speaker cannot draw from a well that contains only mesocosm and bring up something else.
Scientific notation systems emerged precisely to solve this problem, and the solution is demediation. A notation severs the coupling between a science's working symbols and ordinary language, and in severing it cuts out the metaphorical baggage wholesale rather than word by word. The equation, the chemical formula, the structural diagram do not translate the ordinary-language description of their objects. They replace it, with symbols whose meanings are stipulated within the system rather than inherited from the mesocosm, and that is why notation is free of mesocosmic metaphor: not because its inventors were unusually careful writers, but because demediation is what notation is for. I have argued elsewhere that paradigm shifts in the sciences are enabled by exactly such demediation events, formal notations that sever a field's vocabulary from its ordinary-language coupling, and the present principle supplies the reason those events are so consequential. What a demediation removes is not vagueness. It is the mesocosm itself, the entire inherited anchor-set through which the field had been imagining, and mis-imagining, its objects. It also explains the division of labour that the previous section relied on: capture lives in a field's ordinary-language self-description, where the mesocosmic anchors operate, and a field's formal work can proceed untouched by it, because the formal work is conducted in a demediated system where those anchors no longer exist.
The consequence for the present argument should be stated precisely, because notation might look like a counterexample to the principle and is in fact its strongest corroboration. Notation does not violate the stability requirement. The written mark is as nonrecursive as symbolic material gets, and formal systems are prized for exactly the invariance the principle identifies. What notation abandons is the mesocosmic provenance of its anchors, replacing recruitment with stipulation. And the fact that this abandonment required the invention of new symbolic technologies, rather than the reform of the old one, is itself the principle's signature. Ordinary language could not be cleaned of mesocosmic metaphor, because the metaphors are not in it. They are it. Where enquiry needed symbols without mesocosmic provenance, it had to demediate, and the history of the exact sciences is in significant part the history of it doing so.
XIII. The empirical prediction
The argument has been derived from the ontology, but its authority does not rest on derivation alone, because the principle entails a prediction that is exposed to the record wherever the record is examined, and the exposure should be stated as plainly as the thesis was.
The prediction is this. In every substantial corpus of conceptual metaphor, the source domains will be nonrecursive structures and processes of the mesocosm, and genuinely recursive phenomena, selfrecursive and above all interrecursive, will appear as targets and not as sources. The prediction covers the canonical corpus of Metaphors We Live By and The Body in the Mind, whose systematic tabulation, source domain by source domain, mediation by mediation, recursivity type by recursivity type, this article has begun and which can be completed by anyone; it covers popular song, whose renderings of love, memory, and loss through seasons, currents, bridges, and distances instantiate the asymmetry with textbook clarity; and it covers political oratory, scripture, fiction, psychotherapy transcripts, and the technical vernaculars of every discipline named above. The refuting observation is equally definite. A well-attested metaphor family whose source is a genuinely interrecursive process, one that renders some target intelligible through the open-ended mutual modification of recursive beings as such, and not through that process's material apparatus, spatial organisation, or ritualised fixed forms, would break the principle. The prediction is that no such family exists in any language.
The prediction can also be given a second, procedural form, and this is the form an analyst will actually use. Whenever an apparent counterexample is proposed, provenance analysis will reveal that the actual symbolic work is being done by the nonrecursive organisation embedded within the proposed source. That is exactly what happened to argument is war, whose apparently social source dissolved, mapping by mapping, into battlefields, projectiles, territory, and fortifications, and it is what happened to following a leader, which dissolved into locomotion behind a moving figure, and to the ritual sources of section IX, which dissolved into procedure and choreography. The procedural prediction is that every apparent social source, every apparent emotional source, and every apparent temporal source that is put forward will dissolve the same way, into the stable structure it carries, and that the dissolving is not a defensive manoeuvre available to the theory but the correct analysis of the metaphor, checkable by the provenance test in each case.
It is worth being precise about the logical situation, because it is unusual and it is a strength. If the survey of Lakoff and Johnson's corpus comes out as predicted, then the principal body of evidence ever assembled for the embodiment of thought becomes, without the alteration of a single datum, evidence for a different ontology than the one it was collected to support. Their database does not need to be contested. It needs to be read with the recursivity distinction in hand, at which point the regularity they discovered, one of the great empirical regularities of ordinary language, detaches from the explanation they gave it. That is the strongest position from which a theory can be corrected: not by counterexamples from outside, but by its own exemplary cases.
XIV. Conclusion: the law of symbolisation
Lakoff and Johnson discovered something true and important, and this article has not disputed it for a moment. Abstract, dynamic, recursive life is thought through concrete mappings, the mappings are systematic, and they organise inference. What this article has disputed is the identification of the source. The source of ordinary conceptual language is not embodied experience. It is the nonrecursive organisation of the mesocosm. Even where the body supplies the anchor, it is recruited through its invariant anatomy and not through its recursive life; even where warfare supplies the anchor, it is recruited through its territory and its projectiles and not through its interrecursive reality; even where time is the target, the anchors are paths, moons, seasons, and dawns, the stable structures of dwelling and matter, and never another temporal flow. And where the mesocosm is most recursive, in the coordination of beings with beings, it supplies no anchors at all and stands wholly on the target side of the record, an absence that experiential theories cannot explain and the present account entails.
The finding can now be given the form it has turned out to deserve, which is the form of an explicit law. The law of symbolisation: every successful symbolic system recruits relatively nonrecursive organisation to render recursive processes intelligible. Metaphor is the oldest and most pervasive manifestation of the law, but it is only that. Money manifests it, anchoring the interrecursive process of valuing in stable artefacts. Diagnosis manifests it, anchoring the selfrecursive process of living with illness in stable signs and structures. Legal form manifests it, anchoring ongoing relationships in fixed categories and documents. And notation manifests it at the limit, where the recruitment of mesocosmic anchors gives way to the stipulation of demediated ones, stability retained while provenance is deliberately severed. One law, appearing independently wherever recursive life is made thinkable.
The law does not stand alone within Living Value Theory, and its position in the larger theory is the final thing to state. The theory has argued throughout its development that recursive life exceeds complete symbolisation, and it has traced the consequences of that excess through money, diagnosis, law, and the forms of institutional life, showing in each case that symbolic systems hold only by anchoring themselves in what does not recurse. What the present article adds is the complementary discovery at the foundation of all of these: ordinary language itself, the medium in which every such symbolic system is built and explained and lived with, obeys the same law, and obeyed it first. Symbolisation succeeds, where it succeeds, not by capturing recursion directly, which cannot be done, but by anchoring recursive life in the nonrecursive organisation of the mesocosm, and metaphor is that anchoring caught in the act. Conceptual metaphor theory assembled the largest demonstration of this law ever compiled and read it as a demonstration of embodiment. Read rightly, it demonstrates where ordinary conceptual language, and with it every theory that has ever borrowed its metaphors, actually comes from: not from the flesh, but from what holds still.