Abstract
This article uses Living Value Theory (LVT) to ask a question that cognitive science, philosophy of mind, and narratology have each approached but none has asked in its sharpest form: what is "I"? The dominant answer across these fields, in its many versions, is that the self is at bottom a narrator, a structural position from which experience is organized into a story, and that the task of theory is to describe that narrator correctly. This article argues that the dominant answer is wrong at its root rather than merely imprecise in its details. Human living is not narratively organized in its basic architecture. Life consists in recursive mediation in a mesocosm, across five mediations, five levels of recursivity, and three types of recursive counterpart relation, and nothing in that architecture is, or requires, an additional entity called the self. What exists beneath any narration is organismic self-differentiation, available to animals and infants without symbols; selfrecursive coordination, one relation among several rather than the essence of selfhood; and interrecursive coordination with others, live and sedimented. "I" is a specifically human, late, and heterogeneous achievement: an L4 multisymbolic stabilization that cuts this heterogeneous living field into I and not-I, a binary that does not track any single line already drawn in the organism's recursive organization. Narration is later still, a selective and comparatively rare symbolic operation, minimally possible at L3 but characteristically an L4 achievement, through which some portion of a life already stabilized as "I" is arranged into a sequentially intelligible account. Narrative theories of the self mistake this rare and late operation for the structure of living itself. The article develops the consequences of this correction for cognitive science's search for the self, for the Cartesian cogito, for narratology's typologies, for the literary history of interiority, and for the politics of who is permitted to say "I" about whom.
I. Who Is I?
The word is said many hundreds of times a day, by every symbolically competent human being, and its strangeness is invisible for exactly that reason. I want tea. I am tired. I did not mean that. I have always been like this. Each use performs an act of extraordinary compression: it takes some fragment of a vast and heterogeneous field of living, a hunger, a memory, an intention, a limb, a debt, a reputation, and assigns it without apparent effort to one side of a distinction, I and not-I, that the utterance does not even need to state because it is already sedimented into the grammar being used. No comparable word exists for gravity, for a heartbeat, for a friendship, for a season. "I" is unique among the words a person uses for themselves in doing this work silently, continuously, and almost without cost. That silence is the phenomenon this article examines.
Living Value Theory's generative principle is that life is recursive mediation in a mesocosm. A living being coordinates its own continuing organization through multisensorial embodiment, other recursively responsive beings through multispecies being-with, found conditions through multiversal dwelling, and, where poiesis has arisen, made material and symbolic ground through multimateriality and multisymbolization. Coordination becomes available to itself across five levels, from L1 seamless coordination through L2 felt misalignment, L3 symbolic articulation, and L4 portable stabilization, to L5 recursive repositioning. And within any of these mediations, a counterpart may be nonrecursive, selfrecursive, or interrecursive, three types of relation rather than three kinds of entity. This architecture, the 5 × 5 × 3 architecture, is offered by LVT as a complete account of what living coordination is. Nowhere in it is there a slot marked self. There is no sixth mediation for selfhood, no privileged level at which selfhood resides, no fourth recursive type reserved for it. What existing theories call the self does not name an additional living process alongside embodiment, being-with, dwelling, and the rest. Human living, across every one of its mediations and levels, is already completely describable without that word.
This is the article's governing claim, and it requires that a widely shared assumption be abandoned rather than refined. Cognitive science, philosophy of mind, and narrative theory have each, in their own idiom, concluded that the self is fundamentally a narrator: a centre of narrative gravity, a narratively constituted person, a self-model, a generative model that predicts the organism's own states. The disagreements among these accounts, considerable as they are, all occur downstream of a shared premise, that there is some structural position, call it the self or the narrator, that organizes a life from within. This article's claim is that the premise itself, not merely its detail, is the source of the difficulty. Life is not organized by a narrator, resident or intermittent, fixed or fluid. It is organized by recursive mediation across heterogeneous counterparts, and only a small, late, specifically human portion of that mediation is ever narrated at all. Narrative theories of self have mistaken the rare and costly symbolic operation of narration for the structure of the living process it occasionally organizes.
The argument proceeds by installing three distinctions before the word "I" is allowed back into the discussion at all, and then two more once it returns. Section II asks what self-related living exists before any symbol is available, in animals, in infants, in the wordless coordination of adult bodies, and establishes that a great deal of it does. Section III examines the hidden binary that "I" performs and shows why the organism's skin, its recursive organization, and its symbolic self-boundary fail to coincide. Section IV dismantles the assumption that what ordinary language calls "me" is fundamentally selfrecursive, and distributes it instead across all three recursive types, including a substantial portion that is interrecursive with others who are not currently present. Section V follows that distribution into the specific case of so-called inner monologue and argues that much of what passes for talking to oneself is better described as sedimented dialogue with absent others. Sections VI and VII then define narration precisely and show where the major cognitive-science accounts of self went wrong in relation to it. The remaining sections rework Descartes, narratology, the literary history of interiority, and the politics of institutional self-attribution in light of the correction, before a final return to the title's question.
II. Before I
The place to begin is with what exists before any symbol is available, because it is here that cognitive science has most often smuggled the conclusion into the premise. A dog differentiates its own limb from a stick it is chewing. An infant, weeks old, differentiates the sensation of its own hand touching its face from the sensation of being touched by another. A wounded animal moves differently because the wound belongs to the process that must subsequently move, and it moves differently in ways that track the wound's own healing rather than an observer's report of it. None of this requires multisymbolization. It is enacted at L1, in the bodily coordination through which a living being determines, moment by moment and without representing the determination, which consequences belong to its own continuing organization and which arise elsewhere. LVT calls this recursive discernment, and its most elementary form, prior to any narrator and prior to any "I", is an enacted differentiation of organismic continuation rather than a proposition the organism silently holds. It should not be glossed, as an earlier version of this argument once did, with the phrase "this is me". That gloss borrows a symbolic pronoun to describe a discernment that occurs without one, and in doing so quietly projects the very achievement under investigation back onto the ground that is supposed to explain it.
This elementary differentiation is only the beginning of what a living being can do without symbols. Selfrecursive coordination, in which a process responds to its own changing state, is richly present in nonhuman life: a body regulating temperature, an immune system altering its later response after exposure, an animal favouring an injured limb, a bird recalibrating its flight after a gust. Interrecursive coordination, in which two or more recursively responsive beings respond to each other's responses, is equally rich and equally prior to symbols: a wolf pack's shared vigilance during sleep, a mother's attunement to an infant's cry, a dog's discernment of another dog's threat or invitation. Being-with is multispecies before it is sociological, and a dog's capacity to coordinate with a household, remember a person's return, anticipate a familiar routine, and become distressed at an unfamiliar disruption to it, is a rich, aspect-specific, continually revised exercise of exactly the recursive relations LVT describes. A dog can exhibit organismic self-differentiation, L1 and L2 recursivity, selfrecursive coordination in the relevant relations, and dense multispecies being-with. It can remember, anticipate, recognize, learn, coordinate, become distressed, and adjust. None of this is diminished by the fact that it lacks an I.
This is the clean test the animal comparison provides, and it should be stated as sharply as the evidence allows: a sense of self, in every sense of that phrase that matters for living, is available without an I. What the dog lacks is not organismic self-differentiation, not selfrecursivity, not being-with, but the further, specifically symbolic achievement by which a living being comes to hold a portable, generalizable, reusable stabilization of the distinction between what belongs to it and what does not, a stabilization that can be deployed across radically different domains, applied to a memory, a possession, an obligation, a future plan, and a bodily sensation with the same single word. That achievement is not a difference of degree from the dog's rich self-related coordination. It is a difference in kind, because it depends on a mediation the dog does not have in comparable form: sustained, cumulative multisymbolization, coupled to durable external material and symbolic deposit, of the kind that allows a distinction drawn once to be redeployed indefinitely across contexts its first use never anticipated. The relevant threshold is not multisymbolization considered loosely, since many species communicate, signal, and even use tools in ways that leave environmental traces. It is the specific, cumulative coupling of strong multisymbolism to an external archive of material and symbolic deposits, sedimented across generations, from which a portable self-stabilizing category like "I" becomes available to draw upon and refine.
The developmental sequence in human infancy confirms the ordering rather than complicating it. Primary intersubjectivity, the infant's earliest capacity for direct, embodied, turn-taking engagement with a caregiver's face, voice, and rhythm, precedes and is the ground for joint attention, the later capacity to coordinate one's own attention with another's toward a shared external object. Only well after both of these does the child begin to construct anything like an autobiographical narrative self, and this sequence, primary intersubjectivity before joint attention before autobiographical reasoning, should be stated in that order rather than reversed, since it is intersubjective attunement, not shared reference, that is developmentally first. The child's capacity for organismic self-differentiation, selfrecursive regulation, and rich interrecursive attunement to caregivers is operative from the earliest weeks of life, long before the child can use the word "I" correctly, and considerably longer still before the child can produce anything resembling a narrative about itself. Self-related living, in short, is not one thing that arrives all at once. It is a layered achievement, and the layers that matter most for keeping a living being coordinated with its world, embodiment, being-with, dwelling, are already substantially in place before the specifically symbolic layers, I and narration, are available at all.
III. The Hidden Binary
If organismic self-differentiation, selfrecursivity, and interrecursivity are already available without symbols, then the question becomes sharper rather than easier: what exactly does "I" add? The answer LVT proposes is that "I" is a hidden binary, an L4 stabilization that classifies the entire heterogeneous field of a life into two categories, I and not-I, and that does this classificatory work so efficiently that the classification itself disappears from view. It is hidden in the strict sense that ordinary usage never needs to name the excluded pole. "I want tea" asserts one side of a distinction without troubling to state the other, and the distinction's very invisibility is part of what makes "I" such a powerful and portable symbolic tool. This is the trapdoor of ordinary symbolization taken to its most consequential case. A disguised L4 is an abstraction that presents itself as though it named something directly and continuously present rather than something generalized and reusable. "I" is the most successful disguised L4 a human being will ever use, precisely because using it does not feel like using an abstraction at all. It feels like naming the one thing in the room that requires no naming.
The reason this matters, and the reason it belongs at the centre of this article rather than as a passing observation about theoretical concepts of self, is that the I and not-I binary cannot be mapped cleanly onto the actual recursive organization of a life. Three boundaries that ordinary thought tends to collapse into one turn out, under LVT's analysis, to be entirely independent: the skin, which marks the organism's biological limit; the shifting, aspect-specific pattern of recursive attribution, which marks what is currently being coordinated with as self, other, or found and made ground; and the symbolic I and not-I boundary, which marks what a given utterance or train of thought currently assigns to one side of the pronoun. These three boundaries do not coincide, and the gaps between them are not marginal curiosities. They are the ordinary texture of living.
Consider what falls inside the skin. A stomach, most of the time, functions as unremarked ground: it digests, absorbs, and signals satiety without becoming a recursive counterpart that anything in the person's living is actively coordinating with. It can suddenly become salient, an ache, a cramp, a wave of nausea, and at that moment it crosses from something like found ground into a selfrecursive counterpart demanding active coordination, without having moved an inch or changed its physical relation to the skin that supposedly marks it as unambiguously "me". A hand gripping a cup is, in the moment of absorbed use, transparent in exactly the sense LVT's account of embodiment describes: it does its work without becoming an object of attention. The same hand, examined after an injury, held up, turned over, assessed, has become something closer to an object under selfrecursive scrutiny, coordinated with in a manner not unlike the way one might coordinate with an external tool that has stopped working. Now consider what falls outside the skin. A scar, once healed, can become a piece of quasi-objective material evidence, consulted, shown, cited, in a way that treats it almost as multimaterial ground rather than as an unmediated part of felt embodiment. A diary, physically external to the body in every sense, can scaffold selfrecursive coordination more continuously and more reliably than most bodily processes ever do, holding a person's own patterns available for review in a form the body alone could never provide. An absent partner, nowhere in the room, can structure a person's ongoing interrecursive coordination, their choices, their rehearsed conversations, their anticipations of reaction, far more powerfully than many objects and even some people who are currently, physically present. A remembered criticism, spoken once years ago by someone now absent or dead, can continue to reorganize what a person does today with a force that exceeds most of what is happening inside the skin at this moment.
The conclusion follows directly: the skin boundary is not the recursive boundary, and neither is identical with the I and not-I boundary that symbolic usage draws. What ordinary language gathers under "me" or "my body" is not a single, pre-given region that the word I merely labels. It is a heterogeneous, constantly shifting distribution of nonrecursive, selfrecursive, and interrecursive relations, some of them inside the skin, some of them outside it, some of them currently active, some of them sedimented from encounters long past, and the symbolic pronoun "I" performs a purification of this heterogeneity into a single reusable category. That purification is a remarkable achievement. It is not a discovery of a boundary that was already there waiting to be named.
IV. Most of Me Is Not Selfrecursive
One consequence of the argument so far must be stated with particular force, because it corrects an error that the earlier stages of this inquiry themselves made. It is tempting, once the priority of organismic self-differentiation over symbolic selfhood has been established, to conclude that what a person calls "myself" is fundamentally the selfrecursive domain: the body, coordinated with its own changing states, as opposed to the nonrecursive and interrecursive counterparts that surround it. This temptation should be resisted completely. Selfrecursivity is a type of recursive relationship, one among three, and it is emphatically not equivalent to selfhood. A human being is not predominantly or essentially selfrecursive. What a person experiences as their own living is distributed across all three recursive types, and recognizing this distribution is more important to the argument than any single claim made so far.
Consider first how much of "my body" functions, for most of any given day, as nonrecursive ground rather than as an actively coordinated selfrecursive counterpart. Teeth, fingers, posture, hair, the visual field, the particular gait a person has walked with for decades, a great deal of ordinary digestion: these are not, moment by moment, counterparts that a living being is selfrecursively monitoring and adjusting. They are simply there, the settled conditions within which living proceeds, functioning at L1 in a manner closer to dwelling than to active self-coordination. They can, of course, become recursively salient at any moment, and describing that transition precisely is one of the things LVT is well placed to do. The tooth that simply participates in eating can, without warning, become a throbbing object of concern, at which point it shifts from unremarked ground into an actively coordinated selfrecursive counterpart. The gait that has carried a person unreflectively for forty years can, after an injury, become something consciously monitored and relearned. But the everyday, unmarked condition of most of what falls inside the skin is closer to found ground than to active selfrecursive engagement, and treating the body as intrinsically selfrecursive simply because it is biologically the organism's own mistakes a possible relation for a constant one.
There are, second, genuine episodes of selfrecursive coordination, and these deserve to be named precisely rather than dissolved into the general background. Why am I angry. I need to calm myself down. Why did I do that. I am becoming tired and should stop. These are moments in which living folds back toward its own state as an object of active coordination, moments in which a person is genuinely responding to their own changing condition rather than to another being or to found or made ground. This is properly selfrecursive coordination, and it is real, important, and frequently symbolically articulated. But it is one configuration among several, not the default state from which all others depart.
The largest correction concerns a third category, and it is the article's major new contribution to the account of inner life. A substantial portion of what passes, in ordinary description, for a person's private, self-directed inner activity is not well described as selfrecursive at all. Much of the verbal stream that runs through waking life takes a form closer to this: she'll say this, then I'll tell her that, no, she'll object, I should have said, he would probably think. This is not a person coordinating with their own changing state. It is a person coordinating, in absentia, with another recursively responsive being, rehearsing, anticipating, revising, and arguing with a position that is not currently answering back but that continues to organize present coordination with real force. The crucial methodological principle this reveals is that a recursive relationship should never be classified according to where it is physically occurring. It should be classified according to the relational organization of the mediation. An event occurring inside a single skull is not thereby selfrecursive merely because it occurs inside a single skull. If its structure is dialogical, organized around anticipating and responding to another party's likely response, it is interrecursive in structure regardless of its physical location, and calling it selfrecursive because it happens "in my head" repeats exactly the collapse of the skin boundary onto the recursive boundary that section III has already shown to be unwarranted.
This distributed picture requires one further refinement before it can be used with precision, because full interrecursivity, in LVT's strict sense, requires a counterpart that is presently and actually responding. A person imagining a conversation with an absent colleague is not literally being answered by that colleague; nothing the colleague is currently doing is altering in response to the imagined exchange. It would therefore be too loose to call every internal rehearsal of another person's likely words interrecursivity without qualification. Three distinctions are needed. Live interrecursivity is coordination with a counterpart that is presently and actually responding, the conversation happening now, the argument in progress. Sedimented interrecursivity is coordination organized around a counterpart who is not currently present or responding, but whose prior relation with the living being has deposited enough expectation, memory, tone, and probable response that the deposited position continues to organize present mediation in the counterpart's absence, much as a road continues to organize movement long after the coordination that built it has ended. Simulated interrecursivity is the deliberate rehearsal of a future encounter that has not yet occurred at all, a conversation prepared for rather than remembered from. Sedimented interrecursivity is the term that matters most here, because it preserves the genealogy of the internal position in actual prior being-with, rather than treating it as a free-floating fiction the mind has simply generated.
The value of the distinction becomes vivid at the moment sedimented interrecursivity meets live interrecursivity again and is corrected by it. I thought she would be furious. She laughs. That single moment displays the difference the theory has been building toward with unusual clarity: the internal position that had been organizing a person's dread was a deposited, sedimented other, built from past encounters and carried forward in the other's absence, and it has just been overridden by the actually responding, live other, who was never obliged to conform to the deposit that had been standing in for her. Existing research in a number of adjacent fields independently supports the reality of this distributed inner life, from studies of dialogical and other-directed inner speech, to the literature on imagined interactions with significant others, to Dialogical Self Theory's account of the self as a multiplicity of internally available positions organized around significant others as well as around the first person. LVT's contribution is not to discover that inner life is dialogical, which others have already documented, but to explain why it should be expected to be dialogical: because what gets bundled inside the skin and called "myself" was never selfrecursive by nature, and a theory that assumes otherwise will misclassify a large portion of ordinary waking experience.
V. There Is No Inner Monologue
The previous section's distinction has a direct casualty, and it is worth making that casualty explicit, because the term it removes has done a great deal of quiet theoretical damage. "Inner monologue" names, in ordinary usage and in a good deal of literary theory, the private verbal stream that a person is presumed to be conducting with themselves. The term should be retired, or at least handled with far more suspicion than it usually receives, because it silently assumes exactly the conclusion the previous section has shown to be false: that verbal activity occurring inside one organism must be a monologue, a single voice addressing itself, rather than what a great deal of it actually is, an internally sedimented dialogue with one or several absent or anticipated others.
The assumption has a traceable source, and naming it is instructive. The I and not-I purification described in section III has already trained ordinary thought to classify everything occurring within the boundary of one organism as belonging to a single, unified symbolic centre. Once that classification is in place, any verbal activity detected inside the skull is automatically assigned to "my" voice, "my" monologue, even when its structure, its turns, its anticipations, its argumentative shape, is unmistakably dialogical. The word monologue does not describe a phenomenological finding. It describes the output of the I and not-I binary applied, without further scrutiny, to verbal activity that happens to occur inside a single head.
Literary theory, properly read, does not in fact support the monological assumption as firmly as it is often taken to. Dorrit Cohn's Transparent Minds, the foundational text for this domain of narrative analysis, distinguishes psycho-narration, quoted monologue, and narrated monologue as its central triad of techniques for representing fictional consciousness, with free indirect discourse understood as closely related to narrated monologue rather than as a separate fourth category standing alongside it, and "interior monologue" reserved for a narrower, more specific technique of directly quoted, syntactically unmediated inner speech, not for consciousness representation in general. Even within this narrower technical category, however, what a quoted monologue represents on the page need not be, and in the strongest instances of the technique frequently is not, a single unbroken first-person voice. It is often organized around anticipated interlocutors, remembered accusations, imagined replies, exactly the dialogical structure that section IV has described as sedimented interrecursivity. The literary technique conventionally called interior monologue may therefore be a defensible description of a textual method while being a poor description of the actual inner verbal life it is taken to represent, since the texture of that inner life is very often conversational rather than monological even where the page presents it as a continuous first-person stream.
A phenomenological observation can be offered here in the spirit of a hypothesis rather than a settled finding, because the empirical literature on inner speech documents substantial individual variation in its frequency, form, and dialogicality, and no single introspective report should be generalized beyond its own evidential weight. In at least some people's ordinary waking experience, internally conducted exchanges with significant others, rehearsed arguments, imagined explanations, anticipated objections, appear to occur with a frequency that considerably exceeds anything that would naturally be described as talking to oneself in the strict, single-voiced sense. This should not be inflated into a claim that most people spend most of their day in internal dialogue with others; the evidence available does not support a frequency claim of that strength, and existing research on inner speech itself demonstrates wide variation between individuals in exactly this respect. What the observation does support is a research question that LVT is well positioned to generate: what would experience-sampling studies of inner speech find if they classified inner verbal episodes not by whether they occurred inside or outside the head, but by the type of counterpart, live, sedimented, simulated, or genuinely selfrecursive, around which each episode was organized? That reclassification, applied to data already being collected in the cognitive science of inner speech, would test the claim this section has made directly, and it would do so without requiring any new theoretical commitment beyond the one LVT is already proposing: classify a mediation by its relational organization, never by its location.
VI. What Is Narration?
The argument has so far established that self-related living, including a great deal of what feels like private inner speech, is possible without any narrative organization at all. It is now necessary to say precisely what narration is, because the literature this article engages has frequently proceeded as though the answer were obvious, and the imprecision has allowed narration to be credited with organizing far more of human life than it plausibly does.
A minimal LVT definition can be stated directly. Narration is the multisymbolic organization of differentiated events or states into a temporally, causally, perspectivally, or evaluatively connected sequence. This definition has the effect of excluding a great deal of symbolic activity that is sometimes loosely called narrative but is not. L1 coordination does not narrate; there is nothing at L1 available to be organized into a sequence, because nothing has yet become separately salient. L2 questionability does not narrate either; a felt misalignment, in LVT's careful phrase, is questionability rather than a question, let alone a story, and treating it as an incipient narrative is exactly the retrospective colonization of L2 by language that the theory elsewhere warns against. Even a considerable amount of L3 articulation, the level at which a distinction is first deposited into symbolic form, does not narrate. I hurt. Where is she. I don't know. That's wrong. Each of these is a genuine symbolic articulation, portable, shareable, capable of being returned to, and none of them arranges events into a connected sequence. They name a state or ask a question. They do not tell a story.
A minimal narrative can begin at L3: I got angry when he said that links an emotional state to a triggering event through a simple causal and temporal connective, and this is enough to count as narration under the definition offered above, even though it remains close to the ground of the disturbance it reports. Developed narration, however, the kind that cognitive science and narrative theory have generally had in mind when they describe the self as narratively constituted, is characteristically an L4 achievement, because it does more than link one state to one trigger. It selects among the available material, organizes it into a beginning, middle, and end, imputes motives to the parties involved, connects the present event to a pattern drawn from other events, and produces an account portable enough to be told again, to someone else, in circumstances the original event did not anticipate. I became angry because his remark repeated a pattern in which my contribution is consistently dismissed is doing all of this work: selecting, patterning, generalizing, evaluating, and stabilizing the result into a portable account. The boundary between minimal L3 narration and developed L4 narration is deliberately left porous here, because the point of the definition is not to produce another rigid taxonomy. The point is that narration, at any level, requires symbolic organization, and therefore cannot exist at L1 or L2, and that a great deal of what does occur symbolically, greeting, commanding, pointing, labelling, joking, calculating, asking, swearing, naming, promising, describing, does not organize events into a sequence and is therefore not narration either, however symbolic and however central to human life it plainly is.
The consequence for the theory of self is direct. Most human life is not narrated. Most human symbolic activity is not even narration. Narrative is a specific, comparatively demanding, and comparatively rare symbolic operation, not the natural format of consciousness, and a theory that makes narration the organizing structure of selfhood has quietly generalized from a small, late-arriving, effortful capacity to the whole of living, in exactly the way section III has already shown "I" itself to generalize from a genuine but partial symbolic achievement to an apparently seamless and continuous presence.
VII. How Narrative Theories Mistook the Product for the Process
With narration precisely defined, the major cognitive-science accounts of self can be examined individually rather than treated as a single convergent tradition, because they are considerably more heterogeneous than a survey organized around "the narrative self" tends to suggest, and the heterogeneity matters for how each should be corrected.
Daniel Dennett's centre of narrative gravity is the account closest to the position this article is arguing against directly. The metaphor's own logic, a centre of gravity as a stable structural property of an object, always present, always available for calculation, implies that the self is always operating as a narrative centre, and this implication does not survive contact with the phenomenology of absorbed living. The surgeon absorbed in a complex procedure, the farmer reading the soil with their hands, the parent attuned to an infant in the first hours of life, are at their most intelligent and most fully coordinated with their situation precisely when no narrative centre is active at all. The corrected claim is not that the narrative capacity is intermittent rather than continuous, though that correction matters too, but that there need be no persisting narrative centre, active or dormant, waiting to be switched on. What is present in the surgeon's absorption is L1 coordination across embodiment, being-with, and multimaterial engagement with instruments; a narrative capacity may later be deployed to describe the procedure, but describing it as a centre that was there all along, merely quiet, restores exactly the permanence the correction is meant to remove.
Marya Schechtman's Narrative Self-Constitution View should not be represented, as an earlier stage of this argument represented it, as requiring continuous conscious autobiographical storytelling; her own articulation constraint explicitly allows for local, occasional narrative articulation rather than demanding perpetual self-conscious recounting, and the fairer criticism is more specific than the charge of assuming continuous narration. The more precise difficulty is that Schechtman's account still grants narrative intelligibility a constitutive role in personhood, treating the capacity for local narrative integration as what makes something a person at all, when the much larger field of non-narrative living, organismic differentiation, selfrecursive regulation, live and sedimented interrecursivity, dwelling, and absorbed multimaterial engagement, is doing the overwhelming majority of the coordinative work that keeps a human being alive and oriented, with or without any narrative integration occurring at all.
Thomas Metzinger's account of the transparent Phenomenal Self-Model comes closer than the others to the position this article defends, and it deserves to be credited precisely rather than folded into a generic narrative-self critique, since Metzinger's self-model is explicitly not a narrative construct but a representational one, and his insistence on its transparency, that the organism looks through the model to the world rather than at the model itself, already implies that the model is not continuously foregrounded as anything resembling a narrator. Where Metzinger's account remains incomplete, from LVT's point of view, is that it analyses the self-model primarily through the moments when its transparency breaks down, in depersonalization and related phenomena, leaving the healthy organism's ordinary condition, in which the model is invisible because it is working well, comparatively undertheorized. This is the principle of invisible value applied to selfhood: the coordinations most fully meeting a stake are the most likely to disappear from view, and a theory built around breakdown will systematically underweight the well-functioning case it cannot see.
Karl Friston and Andy Clark's predictive processing account should also be treated with more care than a narrative-self critique allows, and one claim from an earlier version of this argument should be withdrawn outright: it is not true that because the predictive model is always running, predictive processing has no room for the oscillation between absorbed engagement and narrator-position activity that this article has been describing. Continuous subpersonal predictive processing is entirely compatible with intermittent, effortful, explicit symbolic activity built on top of it; the two are not in competition. The more defensible criticism is narrower and, for that reason, more useful. Predictive processing's computational architecture does not itself discriminate between absorbed L1 living, L2 recursive opening, L4 symbolic I-stabilization, and L4 narration. All of these are, at the level the framework describes, instances of hierarchical prediction and error minimization, and a framework that can equally well redescribe absorbed coordination, felt disturbance, symbolic self-classification, and narrative construction in the same computational vocabulary has achieved a certain kind of intersymbolic fit with the rest of computational neuroscience without thereby establishing mesocosmic fit with the specific, qualitatively different achievements this article has been at pains to keep separate.
Galen Strawson's argument against narrativity, distinguishing diachronic individuals who naturally experience their lives as continuous stories from episodic individuals who experience life as a sequence of disconnected moments, is largely vindicated by the account developed here, though it can now be given a firmer foundation than Strawson's own phenomenological appeal provides. Strawson is correct that continuous self-narration is neither a universal psychological reality nor a normative requirement. LVT can now say why: narration is a costly, comparatively rare, L4-characteristic symbolic operation performed on top of a living process that does not require it, so of course some people perform it more often, over longer temporal spans, than others, without either group thereby living a less complete or less healthy life. What matters, on this account, is not whether a person narrates continuously, occasionally, or barely at all, but whether the organismic, selfrecursive, and interrecursive coordination beneath any narration remains capable of proceeding well, since that coordination, and not the narrative overlay upon it, is what living actually consists in.
VIII. Descartes and the Purification of I
The Cartesian cogito can now be read with considerably more precision than a narrator-position account allows. René Descartes, subjecting all his beliefs to systematic doubt, discovers that the one thing he cannot doubt is the existence of the doubting itself, and installs the thinking, doubting "I" as the bedrock of philosophical certainty. The earlier stage of this argument described this as a moment of pure selfrecursive narrator awareness, a special but genuine narrator configuration. That description credited the cogito with too much: it still assumed that what Descartes achieves is a privileged glimpse of the self's true narrator-nature, rather than something considerably stranger and, once seen, more interesting.
What Descartes has actually done is strip away, one after another, the changing contents of experience, the body, the world, other people, mathematical certainties, everything that might be doubted, in order to see what symbolic pole remains available for those contents to be attributed to, once attribution is stripped down to its bare minimum. I doubt. I think. I am. What survives the doubt is not a discovery of the self beneath all mediation. It is the I and not-I distinction itself, already available in Descartes's Latin and French as an ordinary feature of the language he thinks in, now purified of every particular content that had previously been assigned to its marked pole, and mistaken, in that purified state, for direct contact with the ontological ground of selfhood. Descartes does not invent the I and not-I binary; ordinary language already supplies it, and has supplied it to every speaker who has ever used a first-person pronoun. What Descartes does is radicalize the binary philosophically, by treating the marked pole of an already available symbolic distinction, emptied of content until only the bare act of attribution remains, as though it were ontological bedrock rather than an extraordinarily portable L4 stabilization operating, in this instance, at the very limit of its own portability.
Seen this way, the cogito is not evidence that the self is fundamentally a narrator, available in a special configuration to a philosopher willing to cultivate it. It is evidence of something more specific and, for this article's purposes, more useful: evidence of just how far the I and not-I purification can be pushed, and of how easily that purification, pushed far enough, can be mistaken for the discovery of a substance. The error that has structured three centuries of subsequent philosophy of mind is not the error of taking a rare narrator-position too seriously. It is the error of taking an emptied symbolic pole for an ontological ground, and mistaking the extreme portability of "I" for evidence that "I" names something that was there all along, prior to and independent of the symbolic operation that produced it.
The joke about Descartes's dinner survives this correction and does useful work within it. Having completed the Meditations, Descartes could presumably eat his dinner without thinking therefore being for the duration of the meal, and the joke's point is no longer that Descartes returns from an exotic narrator-position to an embedded one, but something plainer: the I and not-I purification he performed in his study was an extraordinary and effortful symbolic achievement, not his permanent condition, and its absence from the dinner table does not mean his I had vanished, only that the philosophical work of emptying it out and holding it up for inspection was, for the evening, over.
A related and more cautious point concerns depersonalization, which an earlier stage of this argument identified too confidently with permanently extended Cartesian narration. That identification claimed more than the theory can support, and it should be scaled back to a hypothesis about one phenomenological feature rather than a definition of a clinical syndrome. Depersonalization can involve, for at least some who experience it, a felt inability to return from a state of recursive self-observation to absorbed, unreflective, first-person engagement with one's own living, and if that is so, LVT's account of the difficulty of moving back from an emptied, purified I-position to organismic and interrecursive coordination may illuminate part of what is happening. This should be offered as one possible way of conceptualizing part of a complex clinical phenomenon, not as its definition, and the same caution applies to any suggestion that therapy consists in restoring a general fluidity between narrator positions. LVT can suggest that restoring the capacity to move between symbolic self-stabilization and absorbed coordination may be one useful way of describing certain therapeutic change. It should not be advanced as a full account of what therapy, in its many forms and for its many purposes, is.
IX. Narratology After the Narrator-Self
Literary narratology, developed by Gérard Genette, Wayne Booth, Dorrit Cohn, and James Phelan with extraordinary technical precision, has often been read, by cognitive science and by an earlier stage of this article alike, as supplying a typology of narrator-position configurations that a living self can occupy. That reading credits narratology with more than it can deliver, and the correction here is significant rather than cosmetic.
A literary narrator is an L3 or L4 symbolic construction: a made deposit, produced by an author's recursive coordination and sedimented into a text, available to be taken up by readers who did not participate in producing it and who encounter it only as a stabilized artefact. Genette's focalization categories describe ways in which narratives, as symbolic constructions, organize access to a storyworld's participants. Zero focalization, the classical omniscient narrator with simultaneous access to every participant's inner states, is a textual technique with no direct psychological equivalent; no living human being has, or has ever had, the kind of access to other minds that an omniscient narrator is conventionally granted, and treating zero focalization as a configuration a living self can "achieve" when reflecting on a past interaction, as an earlier stage of this argument proposed, overstates the analogy considerably. What a person achieves in retrospective reflection on a past interaction is, at most, an inference about another party's likely stakes and motives, built from memory, sedimented interrecursive knowledge of that party, and whatever the person said or did at the time; it is an inference under uncertainty, not the god's-eye simultaneous access that zero focalization grants a fictional narrator by textual convention.
The corrected claim should therefore be considerably more modest and, for that reason, more defensible. Narratology gives an unusually precise description of what human multisymbolism can do once experience has already been narrativized into a text. It does not supply the missing ontology of a living self that narratology's categories were secretly describing all along. Genette's focalization types, Booth's reliability spectrum, and Cohn's modes of consciousness representation should be used, in any account that draws on them, as comparative symbolic techniques, evidence of the range of things a multisymbolic being can do with narrative once it chooses to narrate, rather than as a hidden psychological taxonomy of configurations available to living selves whether or not they are narrating at all.
Booth's distinction between reliable and unreliable narrators retains a genuine, if narrower, application to self-narration once this scaling-back is observed. When a person does narrate their own past, factually, evaluatively, or in terms of motive, the account produced is characteristically unreliable in Booth's specific senses, because self-narration takes up sedimented traces across multiple mediations, embodied, affective, relational, spatial, habitual, only some of which were ever previously symbolized, and produces from them a new symbolic organization rather than retrieving a stored and complete original story. A memory is not, as an earlier stage of this argument claimed, simply a symbolic deposit; memories can be embodied, sensory, affective, relational, spatial, habitual, and scaffolded by multimaterial objects long before, or instead of, being symbolized at all. What self-narration does is perform a new act of symbolic organization on this heterogeneous material, selecting, connecting, and evaluating traces that were never a story to begin with, which is precisely why the self that narrates itself alters the living process it narrates, since narrating a memory is producing a new L3 or L4 artefact from older, only partly symbolic sediment, not replaying an unmediated original.
X. Homer, Caesar, and the Convention Error
The historical argument that literary conventions of narration reveal, or fail to reveal, changes in the underlying structure of selfhood remains one of the strongest sections of this inquiry, and the correction developed above makes it stronger still rather than weaker, because it replaces an overreaching conclusion with a more precisely targeted one.
The standard view in intellectual history, shared by thinkers from Hegel to Charles Taylor and by literary historians from Erich Auerbach to Ian Watt, holds that something changed in the structure of selfhood between the ancient world and modernity, that pre-modern persons had a different, less inward, less psychologically complex form of self-experience, and that the rich interiority associated with modern selfhood is a historical achievement rather than a biological given. This view draws its evidence primarily from literary sources. Homer's characters are externally narrated, their inner states rendered through speech and action rather than through reported inner experience; Julian Jaynes read the divine machinery of the Iliad as evidence that Homeric persons literally heard their own impulses as external divine voices, lacking a unified inner narrator; Bruno Snell argued that the fragmented vocabulary for mental states in Homer reflects a genuinely fragmented Homeric psychology; Auerbach distinguished the Homeric style, which brings everything to the surface, from the biblical and modern styles that permit depth and concealment.
An earlier stage of this argument concluded that these claims are simply mistaken, that literary convention tells us about convention and nothing more. That conclusion should be softened from irrelevance to underdetermination, because the stronger and more defensible claim is not that literary evidence tells us nothing about the persons it represents, but that it cannot, by itself, settle the question either way. A literary convention is a multisymbolic deposit, produced by the recursive coordination of poets, scribes, audiences, and genres, and sedimented into a form that later users can take up without reconstructing its origin. Like every deposit, it preserves constraints and affordances rather than the living recursion that produced it. Homeric conventions can tell us a great deal: about publicly performable forms of articulation, about what a given genre and its material carrier made cheap or expensive to represent, about which aspects of a life were considered worth depositing in a form intended for public recitation. What Homeric conventions cannot do, on their own, is establish that Homeric persons lacked organismic self-differentiation, selfrecursive coordination, or the capacity for what this article has called sedimented interrecursivity with absent others. The absence of a literary technique for representing interior narration underdetermines the question of whether interior experience, in the rich sense this article has developed, was present. It does not settle that question in either direction.
Julius Caesar's Gallic War, in which Caesar refers to himself consistently in the third person, "Caesar decided," "Caesar advanced," "Caesar considered the terrain," remains the cleanest available demonstration, and it becomes sharper once "I" itself is understood, per section III, as a symbolic stabilization rather than a transparent report of lived standpoint. If grammatical person were a transparent guide to lived experience, Caesar's third-person self-reference would suggest that he experienced himself in a permanent condition of external observation of his own life, which is absurd for a general making rapid, high-stakes decisions under extreme pressure across thousands of soldiers, allies, and enemies. The third-person convention is a genre feature of Roman military and historical writing, associated with authority and a specific political gravitas appropriate to a text written for a Roman public with its own stakes in how a campaign narrative should be presented. Caesar can decline to use the first-person pronoun within that genre without thereby losing, or ever having lacked, the ordinary symbolic capacity for the I and not-I distinction that every competent speaker of his language possessed and used elsewhere. Grammatical person, in short, cannot be equated with lived standpoint, and Caesar's text demonstrates this dramatically precisely because we have such strong independent grounds, from the sheer cognitive demands of his situation, for believing that his lived standpoint was not what his grammar suggests.
The claim that oral epic composition "required" external narration, and that interior narration was simply too costly for its material carrier to bear, should likewise be presented as a hypothesis about the affordances of a particular carrier and genre rather than as an established causal law. A performed epic, held in memory, metre, and formula, and delivered before a responsive audience, plausibly makes extended interior narration expensive to sustain in performance; written prose, sedimented on a page that can be read alone and reread, plausibly lowers that cost considerably. This is a reasonable hypothesis about why certain techniques appear when and where they do, consistent with the broader LVT point that multimateriality and multisymbolization elaborate together, each making new operations available to the other. It should not be overstated into a claim that oral tradition strictly required, rather than merely made comparatively costly, one particular representational convention.
XI. Who Gets to Say I?
The correction developed in this article deepens, rather than displaces, an argument that belongs at its centre: the question of who is permitted to stabilize the "I" of another person, and on what terms. An earlier stage of this inquiry described parents, clinicians, courts, and schools as narrators of other selves. That description undersells what is actually at stake, because these parties do not merely tell stories about other people. They participate, often with institutional authority that the person concerned cannot easily contest, in stabilizing what counts as that person's I in the first place: what belongs to it, what counts as its action, which intentions are attributed to it, where responsibility for a given outcome lies, and which parts of a person's living can be safely disregarded because no available category recognizes them.
A diagnosis, in this light, is not only offering an account of a person's history. It is asserting, with institutional force behind it, that a given symptom belongs to a given I, that this is the kind of person the diagnosed party is, and that this is what a given experience means, all prior to and independent of any narrative that might subsequently be constructed from the diagnosis. A court's verdict performs comparable work at a different site: this intention belonged to this defendant, this loss belongs to this claimant, this account, and not that one, is authoritative. In each case, institutional power is operating through something closer to a pronoun-like partitioning of a recursive field than through storytelling as such: the institution is deciding, before any narrative is told, where the boundary of I and not-I falls for the person being classified, and what belongs on which side of it.
This distinction, between symbolic classification of an I and narrative organization of that I across time, needs to be kept sharp, because forced misarticulation, the specific harm LVT identifies when a person is required to speak but only in a symbolic form that distorts what they are trying to say, does not require forced narration to occur. A person can be required to accept a classification, I am depressed, I have attachment issues, I am noncompliant, I am a victim, without yet being asked to produce any narrative at all. Only at a further, and often later, stage does an institution typically demand a narrative organized around the classification: when did this begin, what caused it, why did you act this way, what happened next. Distinguishing the classificatory imposition from the narrative demand allows a sharper diagnosis of where a given harm is actually occurring. Sometimes the damage is done entirely at the level of classification, before any story has been asked for; sometimes a classification that a person could accept is compounded by a demand for a narrative shape that does not fit their actual history at all.
The recursive burden of reopening such a closure remains asymmetrical in exactly the way LVT's more general account of recursive opening predicts. The party who benefits from an existing institutional classification of another person's I typically does not bear the cost of contesting it; the party classified must produce standing, evidence, and a hearing simply to have the classification reconsidered. This asymmetry is a dimension of power, and the ethical response LVT proposes, conceptual justice, requires that a symbolic account of a person, whether a classification or a narrative built upon one, remain accountable to the coordination it claims to describe rather than overriding the self-articulation of the person whose living is being classified. Conceptual justice, applied here, does not privilege a person's own account of their own I over an institution's; a person is, as section IX has shown, an unreliable narrator of their own life in Booth's precise senses, and no account, one's own included, is more than a provisional closure. What conceptual justice requires is that the capacity to reopen any such closure, whoever produced it, remain genuinely available to the person whose living it purports to classify.
XII. Who Is I?
The question the title poses can now be answered, and the answer requires setting aside every candidate this inquiry has examined along the way. The answer is not the embedded narrator, a living agent whose narrator-position capacity expands and contracts with circumstance; that formulation, useful as a corrective to a still cruder fixed-narrator view, still leaves a narrator quietly in place across the vast stretches of living in which no narration is occurring, waiting, as it were, for its cue. The answer is not the self as selfrecursive organism, since section IV has shown that selfrecursivity is one recursive relation among three, and that a great deal of what a person experiences as their own inner life is, in its actual relational structure, interrecursive rather than selfrecursive. The answer is not the brain's self-model, the centre of narrative gravity, or the predictive brain's most robust generative model, each of which names a real and valuable finding about a mechanism, without settling the question of what that mechanism is a mechanism for.
What exists, beneath every one of these candidate answers, is living recursive mediation: organisms, bodies, other beings, found and made places, materials, and symbols, coordinated across five mediations, five levels, and three types of recursive relation, continually opening and resettling, continually distributing what is currently ground and what is currently counterpart. Organismic self-differentiation exists, available without symbols to any sufficiently complex nervous system. Selfrecursive coordination exists, one relation among the three a living being maintains. Live and sedimented interrecursivity exist, and a considerable portion of what feels like private inner life belongs here rather than to the selfrecursive domain. Symbolization exists, and among its achievements is the specific, portable, endlessly reusable stabilization of a distinction between I and not-I that no other species has been shown to produce in comparable form. Narration exists, later and rarer still, a selective symbolic organization of some portion of an already-stabilized I into a sequentially intelligible account.
But the self is not another item to be added alongside these, a further mediation, level, or counterpart type discovered once the inventory is complete. The I is the spectacularly successful, almost entirely invisible L4 operation that makes this whole heterogeneous, constantly shifting field of living appear, to the person living it, to possess one naturally bounded, continuously present symbolic centre, when what actually sustains that appearance is a purification performed by, not prior to, multisymbolization. This is not the eliminativist claim that nothing is there. A great deal is there: living, mediated, recursive, distributed across nonrecursive, selfrecursive, and interrecursive relations that exceed, on every side, the tidy boundary the pronoun draws around them.
"I" is what becomes possible when a multisymbolic living being stabilizes the distinction I and not-I across a field of mediation whose actual recursive organization continually exceeds that distinction, and narration is what becomes possible, later and more occasionally still, when some portion of what has already been stabilized as I is selected, connected, and arranged into a sequence intelligible enough to be told. Neither achievement should be mistaken for the ground it operates on. The narrator was never underneath, waiting. What was underneath, and remains underneath every time the word is used, is life itself, coordinating, mostly without a name for what it is doing, mostly without needing one.
Selected References
Alderson-Day, Ben and Charles Fernyhough. "Inner Speech: Development, Cognitive and Clinical Implications." Psychological Bulletin 141, no. 5 (2015): 931-965.
Auerbach, Erich. Mimesis: The Representation of Reality in Western Literature. Translated by Willard R. Trask. Princeton: Princeton University Press, 1953.
Augustine. Confessions. Translated by Henry Chadwick. Oxford: Oxford University Press, 1991 [c. 397-400].
Austen, Jane. Emma. London: John Murray, 1815.
Bakhtin, Mikhail. The Dialogic Imagination: Four Essays. Translated by Caryl Emerson and Michael Holquist. Austin: University of Texas Press, 1981.
Booth, Wayne C. The Rhetoric of Fiction. Chicago: University of Chicago Press, 1961.
Bruner, Jerome. Acts of Meaning. Cambridge, MA: Harvard University Press, 1990.
Caesar, Julius. The Gallic War. Translated by Carolyn Hammond. Oxford: Oxford University Press, 2008 [c. 50 BCE].
Clark, Andy. Surfing Uncertainty: Prediction, Action, and the Embodied Mind. Oxford: Oxford University Press, 2016.
Cohn, Dorrit. Transparent Minds: Narrative Modes for Presenting Consciousness in Fiction. Princeton: Princeton University Press, 1978.
Dennett, Daniel C. Consciousness Explained. Boston: Little, Brown, 1991.
Dennett, Daniel C. "The Self as a Center of Narrative Gravity." In Self and Consciousness: Multiple Perspectives, edited by F. Kessel, P. Cole, and D. Johnson. Hillsdale, NJ: Erlbaum, 1992.
Descartes, René. Meditations on First Philosophy. Translated by John Cottingham. Cambridge: Cambridge University Press, 1986 [1641].
Dreyfus, Hubert L. Being-in-the-World: A Commentary on Heidegger's Being and Time, Division I. Cambridge, MA: MIT Press, 1991.
Fernyhough, Charles. The Voices Within: The History and Science of How We Talk to Ourselves. New York: Basic Books, 2016.
Friston, Karl. "The free-energy principle: a unified brain theory?" Nature Reviews Neuroscience 11, no. 2 (2010): 127-138.
Gallagher, Shaun. How the Body Shapes the Mind. Oxford: Clarendon Press, 2005.
Genette, Gérard. Narrative Discourse: An Essay in Method. Translated by Jane E. Lewin. Ithaca: Cornell University Press, 1980 [1972].
Hermans, Hubert J. M. "The Dialogical Self: Toward a Theory of Personal and Cultural Positioning." Culture & Psychology 7, no. 3 (2001): 243-281.
Homer. The Iliad. Translated by Robert Fagles. New York: Viking, 1990.
Honeycutt, James M. Imagined Interactions: Daydreaming About Communication. Cresskill, NJ: Hampton Press, 2003.
Hume, David. A Treatise of Human Nature. Oxford: Oxford University Press, 2000 [1739].
Jaynes, Julian. The Origin of Consciousness in the Breakdown of the Bicameral Mind. Boston: Houghton Mifflin, 1976.
McAdams, Dan P. The Stories We Live By: Personal Myths and the Making of the Self. New York: Guilford Press, 1993.
Metzinger, Thomas. Being No One: The Self-Model Theory of Subjectivity. Cambridge, MA: MIT Press, 2003.
Metzinger, Thomas. The Ego Tunnel: The Science of the Mind and the Myth of the Self. New York: Basic Books, 2009.
Montaigne, Michel de. The Complete Essays. Translated by M. A. Screech. London: Penguin, 1991 [1580-1595].
Phelan, James. Reading People, Reading Plots: Character, Progression, and the Interpretation of Narrative. Chicago: University of Chicago Press, 1989.
Ricoeur, Paul. Oneself as Another. Translated by Kathleen Blamey. Chicago: University of Chicago Press, 1992.
Schechtman, Marya. The Constitution of Selves. Ithaca: Cornell University Press, 1996.
Snell, Bruno. The Discovery of the Mind: The Greek Origins of European Thought. Translated by T. G. Rosenmeyer. Oxford: Blackwell, 1953.
Strawson, Galen. "Against Narrativity." Ratio 17, no. 4 (2004): 428-452.
Taylor, Charles. Sources of the Self: The Making of the Modern Identity. Cambridge, MA: Harvard University Press, 1989.
Tomasello, Michael. The Cultural Origins of Human Cognition. Cambridge, MA: Harvard University Press, 1999.
Trevarthen, Colwyn and Kenneth J. Aitken. "Infant Intersubjectivity: Research, Theory, and Clinical Applications." Journal of Child Psychology and Psychiatry 42, no. 1 (2001): 3-48.
Woolf, Virginia. Mrs Dalloway. London: Hogarth Press, 1925.
Zahavi, Dan. Subjectivity and Selfhood: Investigating the First-Person Perspective. Cambridge, MA: MIT Press, 2005.