I. The Social Construction of X
Arguments offered for the social construction of a phenomenon follow a certain pattern. Take sleep, for example. Is sleep socially constructed? Sleeping practices vary across history and between societies. Beds, bedrooms, sleeping arrangements, shift rotas, and doctrines of healthy sleep all have histories, some of them short. Sleep is regulated through norms and institutions, and its distribution follows the lines of class, occupation, gender, and age: who may sleep when, where, for how long, and beside whom is never settled by physiology alone. Interventions that are plainly social, an electrified street, a factory clock, a crying infant whose care falls to one parent rather than the other, alter the timing, duration, continuity, and depth of sleep. People learn to sleep in particular ways, and what they learn becomes an embodied disposition that feels like nature. By every criterion that has made the language of construction persuasive, sleep qualifies.
But something is clearly off with this line of argument. The “meaning” of sleep can be changed without putting anyone to sleep. A person can be required to sleep, scheduled to sleep, and given a bed in which to sleep, without anything ensuring that sleep occurs. The social classification of sleep can change while the physiological process remains as it was, and the physiological process can be changed by a hypnotic drug or a darkened room while its classification remains as it was. A person can also pretend to sleep, convincingly enough to be recognized as sleeping by everyone in the room, while remaining awake. Sleep therefore separates two things the vocabulary of construction tends to fuse: the social intelligibility of a phenomenon, and its occurrence.
The same separation bears directly on one of the most contested topics in which the vocabulary of construction has been used, the debate over sex and gender. In fact, I have avoided writing anything about sex/gender because there seems to be no way of doing this without causing offense to someone, somewhere. What follows are, therefore, my own tentative thoughts on an intractable subject.
My argument starts with a thought experiment: what happens if you substitute sleep for any object or activity that has been deemed to be socially constructed, and see what happens. For example, if we substitute every mention of sex or gender in the works of Judith Butler, her arguments largely survive. Sleep is accomplished through repeated practices; it is governed by learned conventions; it is normatively regulated and institutionally recognized. The stronger claim, that these practices constitute the phenomenon itself, does not survive the substitution. Without proving anything about whether sleep and gender are alike in their ontology, the analogy exposes a question that any claim of construction owes an answer to: what, exactly, is being constructed, and through what process?
This article answers that question through Living Value Theory (LVT), an ontology of living coordination organized by five mediations, five levels of recursivity, and three types of recursive counterpart (Ecks 2026). Its core principles can be stated briefly. How much of the world can be socially constructed cannot be determined by examining the historical variability of the categories through which the world becomes intelligible; it can only be determined by examining the processes through which living realities are constituted, maintained, and transformed. Sex and gender serve as the central case because they defeat both biological determinism and unrestricted constructivism, and sleep serves as the limiting comparison because it makes the difference between recognition and occurrence impossible to overlook.
II. The Social Construction of What, and of How Much?
The idea of social construction has been one of the great critical successes of twentieth-century thought. Across anthropology, sociology, feminist theory, science studies, and critical theory, it has shown arrangements once taken as natural or inevitable to be the outcome of human activity, and it has thereby made them answerable. Ian Hacking, who spent much of his career examining the idea without adopting it, identified its primary use with some precision. Constructionist talk, he argued, serves mainly to raise consciousness: its aim is rarely to refute an idea and usually to unmask it in Karl Mannheim's sense, showing that a present state of affairs need not have been as it is (Hacking 1997). Nowhere, he added, had the project been more successful than in feminist work on gender, and latterly on sex.
The achievement is real, and it is also more limited than its rhetoric. Showing that something need not have been as it is differs from showing that it is constituted by social activity, and both differ from showing that it can now be made otherwise. Four questions are habitually collapsed into one. Is the category historically produced? Is the phenomenon causally influenced by social practices? Are social practices necessary to the phenomenon's existence? Can the phenomenon be deliberately transformed through social intervention? A positive answer to any one of them entails nothing about the others. A correspondent responding to Hacking's essay made the point with some asperity: recognizing the forces that build a social construction confers no power to dismantle it, just as the realization that a political order was a human rather than a divine imposition did not, in the event, move its subjects to rise against it (Carpenter 1997).
Hacking's own contribution was to insist on the question that later gave his book its title, The Social Construction of What? (1999). Whenever construction is claimed, one must ask whether its object is an idea, a classification, a practice, a kind of person, or an object that would exist without any of these. He confessed to having conflated idea and object himself. Writing about child abuse, he had declared it both a real evil and a construct, and only later recognized that the real evil was the practice while the construct was the concept, and that the interaction between the two was the whole point of the matter. The title of his earlier essay, "The Making and Molding of Child Abuse" (Hacking 1991), had already preferred plasticity to assembly from parts, and his answer to the question of what had happened to what was that something had happened to both the idea and the practice, though not in the same way (Hacking 1997).
That clause, not in the same way, is where the present argument begins. Hacking specified one mechanism by which such differences arise. People classified as being of a kind can become aware of the classification and alter their conduct, so that what was true of the kind ceases to be true; he called this the looping effect of human kinds, and he called kinds that loop in this way interactive (Hacking 1995, 1999). Quarks, by contrast, are indifferent: knowledge can act upon them, but they do not behave as they do because they are aware of how they are classified. Hacking added that indifference differs from passivity. Plutonium, which humans made and which kills humans, is singularly active and wholly indifferent to its name. The distinction, he suggested, might mark the deepest difference between the natural and the human sciences, and it would explain why social sciences that treat their objects as indifferent keep finding that those objects will not stay still under the microscope.
Sally Haslanger (2000, 2017) has supplied a complementary refinement, separating causal construction, in which social factors bring a thing into existence or shape its features, from constitutive construction, in which reference to social relations belongs to what it is to be the thing at all. Together, Hacking and Haslanger have removed much of the looseness from constructionist argument, and nothing proposed here claims that the difficulty has gone unnoticed. What remains underspecified in both is the organization of the processes themselves. If social practices causally produce changes, through what do they act? Under what conditions do they fail? And why does a classification loop powerfully around one phenomenon and leave another, classified just as forcefully, untouched?
Answering those questions requires three distinctions that the remainder of the article keeps apart. Symbolic variability is the capacity of a phenomenon to be described, divided, evaluated, or recognized differently. Historical contingency is the fact that a phenomenon emerged through particular circumstances and might have had another history. Mesocosmic transformability is the capacity of a phenomenon to be altered by intervention in the ongoing coordination through which it is lived. These are analytically distinct features of living processes, and they are not three degrees of a single property. An institutional arrangement can be historically contingent and extraordinarily hard to change; a bodily condition can be classified in many ways without any classification altering it; a bodily process can be transformed by interventions whose effect owes little to how the process is described. The historical contingency of a phenomenon reveals neither the magnitude nor the mechanism of its transformability.
III. From Gendered Practice to Materialized Sex
The distinction between biological sex and socially constructed gender was among the most productive moves in twentieth-century feminism. By assigning anatomy to one domain and roles, behaviour, and inequality to another, it broke the inference from bodily difference to social destiny. Yet the distinction carried a cost that became more visible the more it succeeded. It presupposed a body constituted in advance, onto which social meaning was subsequently laid, and so reproduced the division between nature and culture that feminist theory had set out to overcome.
Candace West and Don Zimmerman (1987) had already loosened the distinction from the side of interaction. Treating gender as a routine accomplishment of everyday conduct, they separated sex, the socially agreed criteria for classifying persons as female or male; sex category, the placement of a person in one category in everyday life, usually without any inspection of those criteria; and gender, the management of conduct in the light of normative conceptions of what befits one's sex category. Their analysis already found several processes inside what ordinary usage calls gender, each with its own way of operating.
Judith Butler's Gender Trouble (1990) pressed further. Gender, on Butler's account, is stabilized through the repetition of acts under norms, so that no identity precedes its expression, and the norms take part in constituting what can count as a recognizable identity at all. The performance has no prior performer. This is the argument most often caricatured as the claim that gender is a costume one may choose, and the caricature obscures how much the argument concedes to constraint. Repetition under norms is compulsory, and its sedimentation is what makes a gendered life feel natural to the person living it.
The decisive text for the present argument is Bodies That Matter (1993), where Butler extends the analysis from gender to sex. In its introduction she describes sex as "an ideal construct which is forcibly materialized through time" (Butler 1993: 1), and argues that the materiality of the sexed body cannot be treated as a given prior to the regulatory norms that govern its intelligibility. Butler is careful here, and the criticism must be equally careful. She does not claim that discourse causes sexual difference, and she distinguishes the claim that sex is always discursively mediated from the claim that discourse creates the body. What is at issue is the explanatory specificity of materialization, and the criticism has nothing to do with any alleged denial that bodies exist.
Three propositions need to be held apart. Norms constitute the recognition of bodily differences. Norms affect the treatment and development of bodies. Norms participate in the materialization of bodily sex. The first is close to uncontroversial; the second is well documented; the third is the one on which the theory rests, and it does not follow from the other two. Chromosomal inheritance, gonadal development, hormonal regulation, pubertal maturation, gametogenesis and reproduction, bodily habituation, medical intervention, interpersonal recognition, and legal classification are all grouped under sex. They interact, but they operate through different mechanisms, on different timescales, and with very different sensitivity to what anyone says or does about them. Subsuming them under a single mechanism of performative materialization gives the reiteration of norms a priority that the evidence for its influence does not establish.
The difficulty therefore lies less in any neglect of matter than in a framework whose central mechanism remains normative and discursive, so that every bodily process appears in the theory to the extent, and in the manner, that norms bear upon it. Butler is right that there is no socially unmediated standpoint from which sex can be comprehensively known. That observation is epistemological and methodological, and it does not establish that the phenomena known are constituted through the same mediation by which they are known. If the materiality of bodies is produced through regulatory practice, there must be some way of telling which aspects of bodily existence are materially transformed by such practice, which are partly affected, and which remain unaffected or resistant. Performativity, as a general account, offers no such way. The architecture set out in the next section is designed to supply one.
IV. From Construction to Recursive Coordination
Living Value Theory begins in the mesocosm, and the mesocosm has no binaries, anywhere. Whenever there is a binary, it is bound to be a symbolic classification designed to make living coordinations quicker, easier, and more reliable, but whatever has been classified does not become objectively “there” as a thing. Living phenomena are continuously constituted and maintained through recursive coordination in a mesocosm, the lived field of available beings, processes, materials, and conditions within which any living being proceeds, rather than first divided into biological and social components that subsequently interact. In human life that coordination runs through five mediations. Multisensorial embodiment is coordination with one's own ongoing living organization. Multispecies being-with is coordination with other living beings who respond in turn. Multiversal dwelling is coordination with found conditions that do not respond at all: terrain, climate, season, the alternation of light and dark. Multimateriality deposits prior coordination into durable material form, a path, a garment, a hospital ward. Multisymbolization deposits it into durable symbolic form, a name, a register, a diagnostic manual, which persists across absence, delay, and death (Ecks 2026). The first three mediations are zoetic, belonging to living process as such; the last two are poietic, and their coupling in human life vastly enlarges both the scope of coordination and the scope for deliberately transforming it.
It follows that the phenomenon is the coordination itself, with no biological interior subsequently exposed to social influence and no cultural exterior constructing an otherwise inert body. What are called the biological and the social are aspects of that coordination picked out for particular purposes, and they do not name two regions of being that exist before their encounter. It is equally important that the five mediations are not five causal factors. LVT offers no multifactorial model supplementing biology with social, cultural, and environmental determinants, each acting as an independent variable on a subject already constituted, because the mediations are implicated in the very occurrence of whatever lives. The question of construction is accordingly reformulated. Instead of asking whether social factors influence a biological reality, one asks how a particular change in one or more mediations reorganizes an existing coordination, and how far the reorganization reaches.
The five levels of recursivity add a second axis. At L1, coordination proceeds without becoming separately salient: a person dresses, moves, works, and sleeps without gender being at issue. At L2, something becomes salient before it can be named: an expectation chafes, a garment feels wrong, a gesture from another person registers as attraction or affront. L2 contains no latent sentence waiting to be released, and it is better understood as the condition from which articulation may arise (Ecks 2026). At L3 the experience is articulated for the first time, as when someone asks why a particular behaviour is expected of women, or says aloud what gender they are. At L4, articulations are abstracted and stabilized into portable form: diagnostic categories, statutes, registration systems, scientific theories, political programmes. At L5, those stabilizations themselves become objects of repositioning, as in the debate this article joins. The levels are not stages through which every experience must pass, and several are typically active in the same situation.
The third axis concerns the counterparts with which coordination proceeds. Some aspects of a counterpart do not alter course in response to being engaged: the cut of a garment once sewn, the shelving of a shop, a statute once enacted. These are nonrecursive, and every poietic deposit is nonrecursive in this sense, however much recursive activity went into producing it. Selfrecursive counterparts alter in relation to their own ongoing state, as habits, anticipations, and bodily regulation do. Interrecursive counterparts respond to one another's responses, as in recognition, flirtation, refusal, and dispute. Recursivity is settled aspect by aspect: a single entity, a body for instance, presents nonrecursive, selfrecursive, and interrecursive aspects at once, and no entity has a single recursive character as a whole.
Hacking sorted kinds into the interactive and the indifferent, and the sorting generated his most stubborn puzzle: how could schizophrenia be both a moving target that loops with its classification and a neuropathology indifferent to what is known about it? His proposed resolution was semantic, allowing the term to refer to an as yet unknown pathology while its enriched stereotype remained socially constructed, and he admitted to finding his own position uncomfortable (Hacking 1997). The puzzle arises from assigning recursive character to kinds. Settled aspect by aspect, the dilemma loses its grip, since a condition can present aspects that answer to classification, aspects that answer only to their own ongoing state, and aspects that answer to nothing, all within one coordination. Hacking's binary also folds together two variables. One is whether an aspect answers back at all; the other is whether it answers back through awareness of a classification, which is the specific route of L3 and L4 uptake. Hacking himself noted, almost in passing, that behavioural modification of depressive states alters serotonin levels. That observation, which belongs to neither of his kinds, describes a selfrecursive aspect, answering to its own engagement without any awareness of a name. His remark about plutonium anticipates a further point: recursive category varies independently of stability, predictability, controllability, and durability, so that a nonrecursive counterpart may be violently active and a recursive one remarkably inert.
Two further concepts complete the toolkit. Mediational bandwidth names the range of mediations across which a category, practice, or institution organizes consequential coordination; recursivity bandwidth names the range of levels at which it does so. Both must be distinguished in their claimed and their demonstrated forms. A theorist may assert that gender organizes every aspect of a person's embodied life. The assertion is a claim about bandwidth, and it demonstrates nothing about how much of that life gender in fact organizes.
V. Sex and Gender: What Exactly Can Be Constructed?
Sex, male, female, man, woman, masculinity, femininity, and gender do not identify a single object, and they do not consistently identify one kind of object. Depending on context they refer to reproductive processes, anatomical features, personal identities, social expectations, administrative classifications, power relations, or political solidarities, and these referents cannot be assumed to share one underlying mechanism. The naturalization of sex and the generalization of performativity commit the same fault from opposite sides: each projects a broad L4 stabilization onto a heterogeneous field and then reads the field as though it were as unified as the word. LVT requires instead that the coordination corresponding to each claim be established. Four cases show what this involves.
The first case is sex as a collection of developmental and reproductive processes: chromosomal constitution, gonadal differentiation, hormonal regulation, genital anatomy, the organization of the body around the production of one or other type of gamete, and the secondary characteristics that emerge at puberty. Two points hold together. These dimensions do not invariably align in one simple configuration; bodies combine sex characteristics in many ways, and some configurations change across a lifetime. Yet the existence of variation does not make reproductive physiology a product of whichever classification a society adopts. The chromosomal constitution of a body is, for every practical purpose, a nonrecursive aspect of it, altering course under no description and no treatment. Gene expression, hormonal regulation, and gonadal function are selfrecursive, answering to the body's own ongoing state and to substances that enter it. Neither kind of aspect answers to a classification as such.
Nothing here implies biological immutability, since sexed bodies are materially transformed by nutrition, by environmental exposures, by exogenous hormones, by surgery, and by the long habituation of movement and posture. The point concerns the specificity of transformation. Changing a classificatory system is one intervention; changing hormone levels is another; changing reproductive anatomy is a third; altering the interpersonal and institutional conditions under which these differences matter is a fourth. Each acts through different mediations, at different levels, on different counterparts, and each has its own possibilities and limits.
Anne Fausto-Sterling has done more than anyone to bring this specificity into view. Sexing the Body (2000) refused the dualisms of sex and gender, nature and nurture, real and constructed, and her later work uses dynamic systems theory to show how gender/sex becomes embodied through development, beginning in infancy with repeated sensorimotor and relational experience (Fausto-Sterling 2019). Her most recent work proposes methods for studying identity in infants and toddlers that can register gender/sex indifference alongside gender/sex insistence, so that development need not be presumed binary from the outset (Fausto-Sterling 2025). Without claiming to have discovered that biology and culture interact, and sharing Fausto-Sterling's commitment to process, LVT adds a specification of where a dynamic system is recursive, at which level, and with which counterparts. The hyphenated term gender/sex records entanglement; the architecture of mediations, levels, and counterpart types makes it possible to say which strand of the entanglement will respond to which intervention.
The second case is the classification of a newborn as male or female, which shows how a single symbolic act can initiate pathways of coordination that run for a lifetime. Within minutes it organizes a legal record, a name, a set of medical expectations, the clothing the infant will wear, the toys that will be bought, the tone in which the infant is addressed, and the kinship terms by which relatives place it. Here Hacking's account of looping needs extending. On his criterion, an interactive kind loops because the classified become aware of their classification. A newborn is aware of nothing of the kind, and yet few classifications have more far-reaching effects. The loop runs through others: through the being-with of parents and kin who respond to the classification, through the multimaterial deposits of nursery, wardrobe, and toy aisle, and through the infant's embodiment, which takes up those responses at L1 and L2 long before anything can be articulated at L3. Recursion here lies in uptake by the many parties around the classified person, and the classified person's own awareness arrives late, if at all.
The act of assignment nevertheless does not produce the bodily characteristics to which it refers. Intersex variations make the point from the other side. They show that categorical conventions can fail to describe the diversity of sexed development, and the history of their medical management shows how much L4 classification can do once it is coupled to institutional authority. Classification has determined whether infants underwent genital surgery, hormonal treatment, and lifelong surveillance, often without any possibility of informed consent (Kessler 1998; Karkazis 2008). Few demonstrations of the material consequences of classification are clearer. Yet those consequences did not occur through classification alone. They required surgeons, protocols, instruments, anaesthesia, hospital hierarchies, parental trust, and bodies that heal, scar, and respond in their own selfrecursive ways. The causal explanation is mesocosmic throughout, and an institutional classification can be causally powerful without constituting the reality it classifies. Nor does intersex variation in itself establish a third gender; sex characteristics and gender categories must be distinguished analytically even while the ways in which they become connected are investigated.
The third case is the gendering of everyday things. Pink clothing, flared trousers, sheepskin boots, and body wash marketed separately to men and women are cases of historically variable symbolic association in an almost pure form. Nothing in pink fabric establishes femininity, and in the United States the association of pink with girls and blue with boys settled only in the middle decades of the twentieth century, after a period in which advice to parents ran in both directions (Paoletti 2012). Flared trousers have passed from naval uniform to counterculture to fashion for both sexes, and sheepskin boots made for surfers became, within a few years, the sign of a young feminine style. To call these associations socially constructed is accurate, and it says very little about how they persist.
Their persistence runs through the mediations unevenly. In embodiment, preferences for colours, textures, and cuts are habituated until they feel like taste. In being-with, they are anticipated, recognized, praised, and teased by others. In multimateriality, they are sedimented into product design, manufacturing runs, retail categories, and shops that place girls' and boys' clothing on different floors. In multisymbolization, they are carried by advertising, naming, and the classifications through which products are sold. Dwelling plays almost no part. Mediational bandwidth explains why such conventions are easy to challenge and hard to change. A single L3 articulation can contest their symbolic component, but their multimaterial component, a supply chain and a retail architecture, answers to no articulation whatever and changes only when something reorganizes the deposits themselves. A person can reject the association between pink and femininity while continuing to shop in stores organized around it.
A person can also wear pink without the act being organized by gender at all, and this possibility matters more than any other in the case. The same action can belong to coordinations that are strongly gendered and to coordinations in which gender is idle: a shirt chosen because it was clean, because it was a gift, because it matched a team's colours. Fausto-Sterling's notion of gender/sex indifference names this condition at the developmental level. A theory of performativity that treats every act as a citation of gender norms makes gender omnipresent by stipulation, whereas LVT requires the significance of gender in any particular coordination to be shown, and allows it to be absent.
The fourth case, gender identity, is the most demanding, and it is the one in which theoretical error is most costly to actual people. A gendered sense of self emerges through heterogeneous coordinations involving embodiment, being-with, material surroundings, and symbolization. It can involve embodied experience that no articulated category exhausts: a felt fit or misfit between one's body and the ways others respond to it, registered at L2 long before it finds words. Symbolic recognition, social relationships, clothing, legal classification, and medical intervention can then profoundly change which forms of life become possible. The division between sex as biological and gender identity as psychological or cultural captures none of this, because the phenomenon crosses the division at every point.
LVT can accommodate substantial transformations without claiming unlimited plasticity, and it can recognize limits without deriving them in advance from an abstract definition of biological sex. Both the possibilities and the limits of gendered transformation are mesocosmic. They must be investigated in the processes concerned and cannot be deduced either from biological essentialism or from performativity in general.
Taken together, the four cases show that the question whether gender is socially constructed has no meaningful general answer until the phenomenon at issue has been specified. A commercial convention, a reproductive process, a personal identity, a legal status, and an institutional inequality cannot all be equally socially constructed merely because each is associated with gender. The cases point instead to variable constructibility, understood as the empirically ascertainable scope of particular interventions within particular coordinations, and in no sense as a numerical property that a phenomenon possesses. The question is never how much of gender is biological and how much cultural, but which parts of an ongoing coordination can be altered through which interventions, and which remain relatively continuous. Repeated norms certainly participate in making bodies and identities what they are. Establishing how far, and in what way, requires tracing the full coordination, and performativity cannot serve as an overarching explanation in place of that tracing.
VI. The Sleep Test
Sleep shares many properties with sexed embodiment. It is physiologically organized and deeply mediated by history and society. It can be regulated, interpreted, classified, disciplined, facilitated, interrupted, and materially transformed. And it continues to exhibit properties that resist both symbolic and intentional control. No one becomes physiologically asleep by declaring themselves asleep, and the difference between an unconscious person, a sleeping person, and a person pretending to sleep can matter a great deal whatever their social classification. The comparison therefore tests whether a theory can distinguish the recognition of a phenomenon from its occurrence.
Imagine importing the distinction between sex and gender into sleep. Biological sleep would be the physiological process; cultural sleep would encompass learned practices, expectations, meanings, and perhaps identities, the early riser and the night owl. The distinction might be heuristically useful. There would not, however, be two kinds of sleeping, because actual sleep always unfolds in a particular embodied and mesocosmic situation, and the social organization of sleeping directly affects its physiology. Shift work alters its timing and architecture, a shared bed alters its continuity, and the infant who wakes alters the sleep of whoever has been made responsible for waking. Nobody alternates between biological and cultural sleep. The same holds for sex and gender, and rejecting their ontological separation no more requires the conclusion that sex is constituted by gender norms than rejecting the separation of biological and cultural sleep requires the conclusion that sleep is constituted by bedtime conventions.
The levels of recursivity show why. At L1, sleep is coordination that requires no reflective engagement at all; it is perhaps the purest case of muted coordination in human life. At L2, tiredness, restlessness, or an unfamiliar room make it salient. At L3, sleep becomes an object of articulation, worry, and planning. At L4, it is stabilized in diagnoses, working-time regulations, scientific classifications, and cultural expectations of what a good night is. At L5, its relation to productivity, consciousness, and personhood becomes an object of theory. These levels influence one another, but their causal capacities differ. A diagnosis of insomnia changes how a person understands and responds to the condition without constituting the physiological disturbance it names. Conversely, the prospect of not sleeping can itself produce the arousal that prevents sleep. Clinical models of insomnia describe precisely this loop, in which worry about sleep and deliberate effort to sleep sustain the very wakefulness they target (Harvey 2002; Espie et al. 2006). Here an L3 articulation has embodied consequences, and the architecture that explains its efficacy also explains its limits: articulation can obstruct sleep far more easily than it can produce it.
Hacking's distinction between interactive and indifferent kinds again proves too coarse. Insomnia loops with its classification, in that people who come to regard themselves as insomniacs monitor and fear their sleep in new ways. The most consequential segment of the loop, however, passes through the selfrecursive regulation of arousal, which responds to engagement of any kind, named or unnamed, and which no awareness of a diagnosis is needed to set in motion.
The sleep test finally turns on the difference between intending and accomplishing. A person can intend to sleep without being able to accomplish sleep, and trying too deliberately to fall asleep interferes with the process. The difference extends well beyond sleep to social and institutional transformation in general. A future can be articulated, legislated towards, imagined, desired, and campaigned for without thereby coming about, and it becomes realizable only through the coordination capable of sustaining it. The past, in the form of deposits, can be laid down as ground, whereas the future can only be anticipated, and every anticipation remains open to revision by what the future brings (Ecks 2026). Construction talk tends to treat an articulated future as half accomplished, as though declaring that something could be otherwise were already a step toward its being otherwise. The sleep test shows why that inference fails.
VII. The Inflation of Social Categories
The problem is wider than gender theory, and class provides the instructive parallel. In its early uses, class distinguished economic positions: owners and non-owners of the means of production, wage earners and rentiers. Over the following century the category acquired consciousness, political agency, ideology, identity, taste, social reproduction, and historical destiny, until it could be invoked to explain nearly every domain of social life. Each expansion had its reasons. What was never established, and could not be assumed, was that economic position, political consciousness, identity, and collective action are in fact coordinated as tightly as the expanded category implied. The mediational bandwidth attributed to class grew faster than any demonstration of it.
Gender has undergone a comparable expansion. A category first used for patterned differences in social position and conduct became a framework for explaining the constitution of embodied identity, the materiality of sex, the classifications of science, the micro-order of interaction, and the architecture of institutional power. The expansion need not be mistaken, and some of these dimensions are closely connected. The problem arises when their connection is assumed and left undemonstrated, and the distinction between claimed and demonstrated bandwidth is the instrument for telling the two situations apart.
A hypothesis follows about academic theorizing itself. Once a concept becomes the dominant L4 stabilization within a field, researchers become disposed to recognize new phenomena through it. Findings generate elaborations that extend its reach, and the extension is rewarded because it displays the concept's fertility. The result can be insight. It can also be a self-reinforcing framework in which almost any observation counts as further evidence of the concept's centrality. Class comes to explain every social differentiation, gender every embodied practice, power every arrangement. In each case the explanatory reach of a concept is mistaken for the organization of reality. In LVT's terms, intersymbolic fit, the coherence of a body of concepts with one another, is taken for mesocosmic fit, the degree to which those concepts track the coordination they claim to describe, although intersymbolic fit can never validate itself (Ecks 2026).
The remedy is methodological. A theory of class must be able to identify situations in which class is not consequential, and a theory of gender must be able to identify situations in which gender is not the relevant organizing principle. Without negative cases there is no way to distinguish the significance of a phenomenon from the expectations built into the theory that looks for it. Because mediational analysis does not begin by requiring every coordination to be explained through one privileged category, it can find gender idle where gender is idle, and its findings are all the more credible where it finds gender decisive.
VIII. Toward a Science of Mesocosmic Transformability
If the power of social construction is an empirical question and no ontological premise, claims about it can be assessed through a sequence of analytical steps. First, the phenomenon must be identified independently of the category used to discuss it, as far as the evidence allows. Second, its coordination across the five mediations must be established. Third, the levels of recursivity at which it occurs, becomes salient, is articulated, and is institutionally stabilized must be distinguished. Fourth, the nonrecursive, selfrecursive, and interrecursive counterparts through which the coordination continues must be identified. Fifth, the mechanisms through which particular interventions could alter the coordination must be specified. Finally, the actual consequences, limits, and unintended effects of those interventions must be investigated. The object of inquiry then becomes the ongoing coordination itself, rather than an abstract category or an isolated biological process.
Consider the gendered division of caregiving. A symbolic intervention might set out to change expectations about who should care for children and elderly relatives. Such a campaign can matter, but it does not by itself reorganize a household. Actual change may require different employment arrangements and income security, a redistribution of time and domestic responsibility, different material arrangements within the home, new embodied habits and practical competences (the parent who has never soothed a feverish child at three in the morning must learn to do so), changed expectations among partners, children, relatives, and employers, and institutional rules backed by durable legitimation. These are not separate factors to be added together, since their interlocking coordination determines which changes become possible and which can be sustained. A campaign can succeed in changing attitudes while leaving the division of labour almost untouched, as the persistent gap between egalitarian belief and domestic practice has long shown (Hochschild 1989). Conversely, an institutional intervention such as a period of parental leave reserved for fathers can reorganize caregiving practice before any change in articulated identity or belief, and the practice may then carry the belief along behind it. Treating social transformation primarily as a change in discourse or recognition misses both possibilities.
Intersectionality has rightly rejected the assumption that separately defined social identities simply add together (Crenshaw 1989). Demonstrating that several categories interact, however, does not yet explain the mechanisms through which an inequality is produced. Consider the documented under-treatment of pain in Black patients in the United States, which has been linked to false beliefs among medical trainees about biological differences between Black and white bodies (Hoffman et al. 2016). Mediational analysis begins with the clinical encounter and traces the coordination it involves: bodies in pain, the communication of pain across social distance, assessment protocols, prescribing rules, institutional priorities, professional expectations, beliefs sedimented in training and textbooks, and the histories that deposited them. Race, gender, and class may be highly consequential within this coordination, and the analysis will find them consequential where they are; what must be shown is their role and the routes by which they act. Far from weakening intersectional critique, this gives it a more precise account of how inequalities are enacted and sustained, and therefore of where they can be interrupted.
Rejecting unlimited constructibility does not entail political conservatism. Recognizing the material and historical constraints on transformation is a condition of effective change. A programme that assumes symbolic reclassification to be sufficient may leave untouched the conditions it set out to transform, while biological determinism wrongly presents historically contingent arrangements as inevitable. LVT treats transformation as a real but empirically variable possibility of mesocosmic coordination. The aim of critique should then extend beyond revealing that something could have been otherwise, to understanding what would actually make it otherwise.
IX. The World Is Not Waiting to Be Constructed
The long opposition between biological essentialism and social constructivism has encouraged the assumption that the central task is to apportion phenomena between nature and culture. LVT shows why that task is misconceived. There are no two independent domains awaiting division, only continuously changing forms of living coordination whose mediations differ in their causal properties, their histories, and their susceptibility to transformation. Sex and gender are symbolic stabilizations within that reality, L4 operations that bundle heterogeneous processes, and they are not its underlying components.
Butler is right that social identities do not simply express fixed inner essences, and the social and institutional constitution of recognizable identities is a real and consequential process. The materiality of embodied existence, however, cannot be explained by the reiteration of norms alone, and the task is to establish the concrete ways in which norms participate in coordination, together with the conditions under which they fail to produce what is attributed to them. Hacking is right that "the dynamics of classification is where the action is" (Hacking 1997). LVT adds that classification is one route of transformation among several, that its loops run through the uptake of others and through material deposits as well as through the awareness of the classified, and that a great deal of transformation proceeds without classification at all. The three sticking points Hacking identified in constructionist debate, contingency, nominalism, and the explanation of stability (Hacking 1999), each receive an answer. Contingency divides into the separate questions of variability, history, and transformability. The choice between nominalism and inherent structure gives way to recursive discernment, in which attributions are mesocosmic and contestable while the coordinative feedback that sustains, revises, or dissolves them is at no mesocosm's disposal. Stability is explained by sedimentation and redundancy, and assumed from neither inherent structure nor social convention.
Return, finally, to sleep. A person can be recognized as sleeping without being asleep. A person can be physiologically asleep without being recognized as such. Sleep can be transformed by social practices without being reducible to them. The same distinctions apply, in different ways, to sexed embodiment and gendered existence. The sleep test leaves open whether gender and sleep are alike, and it shows that social recognition, social regulation, and the occurrence of a phenomenon must be distinguished before any claim that the phenomenon is socially constructed can be assessed.
The world is neither an inert biological substrate on which culture inscribes meaning nor an indefinitely plastic reality called into being by symbolic practice. It consists of ongoing coordinations whose continuity and transformability depend on specific mediational and recursive processes. Social practices can alter these coordinations, sometimes profoundly, but their powers are limited, and they cannot be inferred from the historical contingency of the categories through which phenomena are understood. The question is therefore less how much of the world is socially constructed than how the world is continuously made and remade, which interventions take part in that making, and why some transformations become possible while others do not. Social construction, on this account, names certain accomplishments within living coordination rather than a general explanation of how the world comes into being, and LVT makes it possible to investigate how those accomplishments happen, how far they reach, and where they stop.
Selected References
Butler, Judith. 1990. Gender Trouble: Feminism and the Subversion of Identity. New York: Routledge.
Butler, Judith. 1993. Bodies That Matter: On the Discursive Limits of "Sex". New York: Routledge.
Carpenter, Malcolm. 1997. Letter. London Review of Books 19(18), 18 September.
Crenshaw, Kimberlé. 1989. "Demarginalizing the Intersection of Race and Sex: A Black Feminist Critique of Antidiscrimination Doctrine, Feminist Theory and Antiracist Politics." University of Chicago Legal Forum 1989(1): 139–167.
Espie, Colin A., Niall M. Broomfield, Kenneth M. A. MacMahon, Leanne M. Macphee, and Lynne M. Taylor. 2006. "The Attention–Intention–Effort Pathway in the Development of Psychophysiologic Insomnia: A Theoretical Review." Sleep Medicine Reviews 10(4): 215–245.
Fausto-Sterling, Anne. 2000. Sexing the Body: Gender Politics and the Construction of Sexuality. New York: Basic Books.
Fausto-Sterling, Anne. 2019. "Gender/Sex, Sexual Orientation, and Identity Are in the Body: How Did They Get There?" Journal of Sex Research 56(4–5): 529–555.
Fausto-Sterling, Anne. 2025. "Rethinking Gender/Sex Identity." American Journal of Human Biology 37(4): e70044.
Hacking, Ian. 1991. "The Making and Molding of Child Abuse." Critical Inquiry 17(2): 253–288.
Hacking, Ian. 1995. "The Looping Effects of Human Kinds." In Causal Cognition: A Multidisciplinary Debate, edited by Dan Sperber, David Premack, and Ann James Premack, 351–383. Oxford: Clarendon Press.
Hacking, Ian. 1997. "Taking Bad Arguments Seriously." London Review of Books 19(16), 21 August.
Hacking, Ian. 1999. The Social Construction of What? Cambridge, MA: Harvard University Press.
Harvey, Allison G. 2002. "A Cognitive Model of Insomnia." Behaviour Research and Therapy 40(8): 869–893.
Haslanger, Sally. 2000. "Gender and Race: (What) Are They? (What) Do We Want Them to Be?" Noûs 34(1): 31–55.
Haslanger, Sally. 2017. "The Sex/Gender Distinction and the Social Construction of Reality." In The Routledge Companion to Feminist Philosophy, edited by Ann Garry, Serene J. Khader, and Alison Stone. New York: Routledge.
Hochschild, Arlie Russell, with Anne Machung. 1989. The Second Shift: Working Parents and the Revolution at Home. New York: Viking.
Hoffman, Kelly M., Sophie Trawalter, Jordan R. Axt, and M. Norman Oliver. 2016. "Racial Bias in Pain Assessment and Treatment Recommendations, and False Beliefs about Biological Differences between Blacks and Whites." Proceedings of the National Academy of Sciences 113(16): 4296–4301.
Karkazis, Katrina. 2008. Fixing Sex: Intersex, Medical Authority, and Lived Experience. Durham, NC: Duke University Press.
Kessler, Suzanne J. 1998. Lessons from the Intersexed. New Brunswick, NJ: Rutgers University Press.
Paoletti, Jo B. 2012. Pink and Blue: Telling the Boys from the Girls in America. Bloomington: Indiana University Press.
West, Candace, and Don H. Zimmerman. 1987. "Doing Gender." Gender & Society 1(2): 125–151.