Abstract
A meta-analysis published in JAMA Psychiatry in November 2026 reports that intolerance-of-uncertainty scores among North American college students rose substantially between 1992 and 2023, with a steeper increase after 2012. Because intolerance of uncertainty is associated with anxiety and other psychological symptoms, the finding has been received as evidence of an increasingly important mental-health risk factor. This article examines the ontological assumptions that make that interpretation possible. Drawing on Living Value Theory, it argues that uncertainty and intolerance are highly heterogeneous conceptual abstractions from multimediated, recursively constituted living coordination, and that their conjunction does not establish a discrete psychological property. Questionnaire operationalization nevertheless makes the compound appear measurable and historically stable. The difficulty intensifies when intolerance-of-uncertainty scores are correlated with anxiety scores, since two overlapping abstractions from the same coordination can then appear as independently measurable entities whose association calls for psychological explanation. The article distinguishes numerical reliability from ontological validity, develops the problem of recursive double-counting, reconstructs the historical findings through the five mediations, five recursivity levels and three types of recursive counterpart, and proposes a science of living coordination that investigates coordinative processes before deciding which psychological entities, if any, they contain.
Introduction: A Seemingly Compelling Discovery
On 18 November 2026, JAMA Psychiatry published a meta-analysis by Jeroen van Baar and colleagues on intolerance of uncertainty among young adults attending college in the United States and Canada. Bringing together 137 studies and 37,298 participants, the authors conclude that average scores on the two established versions of the Intolerance of Uncertainty Scale rose over three decades, and that the rise accelerated markedly after 2012 (van Baar et al. 2026). The finding has every quality that carries psychological research to a wide audience. It concerns a recognizable feature of contemporary experience. It connects with much-discussed increases in anxiety among the young. It offers a numerical historical trend with a turning point. It invites an easily grasped generational story. And it suggests that a psychological characteristic might be modified so that mental health improves. The accompanying explanation almost writes itself: smartphones and permanent connectivity have accustomed a generation to immediate information, and that generation has consequently become less able to endure not knowing.
The interesting question here is less whether the numbers were calculated correctly than what they are supposed to be numbers of. Three deceptively simple questions open it. What is uncertainty? What is intolerance? And what kind of thing is an intolerance of uncertainty? Far from being sceptical wordplay, these questions concern the most elementary requirement of any science, which is to identify the object under investigation before measuring it. An expression can be perfectly intelligible in ordinary language without designating an independently existing and generally measurable property. We understand someone who says they are uncertain whether their train will arrive, and we understand someone who says they cannot tolerate the noise from the flat upstairs. Whether these sentences pick out general entities called uncertainty and intolerance, which individuals possess in measurable amounts, remains to be shown.
Living Value Theory (LVT) approaches the question from a different starting point than psychology does. Psychology frequently begins with an ontology of individuals who possess discrete properties located inside them, whereas the processes it studies are constituted through ongoing, multimediated coordination between living beings and their mesocosms (Ecks 2026). That diagnosis was developed for constructs such as self-control, self-efficacy and growth mindset, where distributed coordination was converted into supposedly individual characteristics. The present article examines the next step in the same operation. Once several such characteristics have been constructed, measured and named, the relationships among them become objects of research in their own right. Intolerance of uncertainty is correlated with anxiety, with depression, with obsessive-compulsive symptoms, with worry and with avoidance, and the resulting network of correlations is taken to reveal the causal architecture of psychological suffering.
The central question of the article follows directly. What has psychology established when it demonstrates a correlation between two numerical measures, if neither purported object has first been adequately differentiated from the living coordination from which it was abstracted? The answer developed here is that a research programme can become steadily more precise about the statistical relationships among its classifications while remaining insufficiently precise about the phenomena those classifications are meant to describe. Without denying that the questionnaires record something real, or that people suffer when consequential matters remain unresolved, the argument asks what scientific work must be done before a stable score can be treated as the measure of a stable thing.
I. What the Historical Study Establishes
A critique of this kind must begin from an accurate reading of the evidence, since its force depends on accepting what the evidence shows. Van Baar and colleagues searched for published studies of college students in the United States and Canada that reported mean scores on either the original 27-item Intolerance of Uncertainty Scale (IUS-27), developed by Freeston and colleagues (1994), or its 12-item short form (IUS-12), developed by Carleton, Norton and Asmundson (2007). Of 294 full texts reviewed, 92 articles met the inclusion criteria, yielding 137 samples, 62 using the long form and 75 the short form. The analysis then related sample means to the years in which the samples were collected.
The headline results are precise. Scores on the long form, which range from 27 to 135, increased by an estimated 0.66 points per year between 1992 and 2023. Scores on the short form, which range from 12 to 60, increased by an estimated 0.60 points per year from its first included sample in 2005 through 2023. In standardized terms, the combined increase was about 0.04 standard deviations per year across the whole period, about 0.02 per year before 2012 and about 0.06 per year from 2012 to 2023, an acceleration of roughly threefold. The authors conclude that the persistent increase justifies further research into the role of uncertainty in the epidemiology of anxiety (van Baar et al. 2026). The senior author has described the construct as “an intermediary cognitive process” rather than an outcome or a root cause (Columbia Mailman School of Public Health 2026).
A distinction that recurs throughout this article must be made at once. The study has established a historical pattern in questionnaire responses, and that is a legitimate empirical finding. What it has not established, merely by obtaining the pattern, is that an independently existing capacity of individuals called tolerance of uncertainty has diminished, still less that its diminution explains rising anxiety. The stronger critique accepts the trend and investigates its interpretation.
Between the data and the interpretation lie four inferential transitions, each requiring a justification that the preceding statistical result does not supply. First, responses to questionnaire items are assigned numerical values, which is a matter of scoring. Second, these values are combined into a total taken to represent a proposed psychological construct, which is a matter of construct validity. Third, historical differences in the totals are interpreted as historical differences in the construct, which requires historical comparability. Fourth, changes in the construct are proposed as possible contributors to changes in mental health, which introduces causal interpretation. Each subsequent transition draws on assumptions that the arithmetic of the first cannot certify.
The design of the study matters at the third transition in particular. A cross-temporal meta-analysis compares different samples drawn at different times, rather than following the same people through time. The populations attending North American colleges changed between 1992 and 2023, as did recruitment practices, the settings in which questionnaires were completed (paper in a lecture theatre, later an online survey on a phone), and plausibly the ways in which respondents understood the items. A sentence about being upset by unforeseen events may not mean the same thing to a student in 1995 and a student in 2022. Psychometrics has a technical name for the requirement at stake, historical measurement invariance, which demands that a scale relate to its construct in the same way across the periods compared. Whether and how the authors examined this would need to be established from their complete methods and supplementary material, and nothing in the argument below depends on assuming that they failed to do so. The point is structural: even a perfectly invariant instrument would show only that the procedure behaves consistently across time.
The research therefore has a well-defined empirical object, namely the distribution of responses to a standardized questionnaire among a particular population across three decades. The open question is whether its theoretical object has been properly identified. This introduces the distinction on which the rest of the article turns, between the stability of a procedure and the ontological stability of its proposed referent.
II. What Kind of Objects Are Certainty, Uncertainty, Tolerance and Intolerance?
Begin with certainty, since uncertainty is defined against it. Someone can be practically certain that the mountains visible from their window will still be there tomorrow. This certainty, far from residing in the mountains as a property they carry, involves a particular anticipated development, a temporal horizon, an established history of coordination with that landscape, and some knowledge of the nonrecursive geological processes whose rates over that horizon are well understood. Change the horizon from tomorrow to several million years and the same expectation becomes much less secure. Change the question from whether the range will exist to whether a particular slope will slide after a week of heavy rain, and the relevant expectations change again. The mountain cannot be assigned a general quantity of certainty independently of what is being anticipated, over what span, and by whom. What exists is a particular coordination within which some anticipated developments can be relied upon and others cannot.
Uncertainty is still more various. In ordinary use the word can refer to an absence of information, to the recognition that one’s information is incomplete, to an inability to anticipate an outcome, to a decision not yet taken, to apprehension about what is coming, or to a felt disruption in coordination that has not yet been named. These phenomena need not coincide. Someone may lack information without being aware of the absence. They may recognize a gap in their information without being disturbed by it. They may feel apprehension while possessing excellent information. They may face an open future and continue ordinary coordination without any difficulty at all. And they may positively enjoy not knowing, as when travelling without an itinerary or reading a detective novel whose pleasure depends on the outcome being withheld. That all these situations can be described with one word does not make them degrees or variants of one thing. Mathematical uncertainty is a different matter again. The probability of a defined event within a specified model can be quantified rigorously, but the success of probability theory does not establish a unitary psychological entity corresponding to every ordinary use of the word.
Tolerance is more complex still. What counts as tolerating something? Feeling no discomfort? Feeling discomfort and carrying on? Refraining from action? Resisting an impulse? Accepting a situation? Ceasing to notice it? Revising one’s expectations? Finding another way of coordinating? These are distinct processes, with different conditions and different consequences. Tolerance also carries a normative charge that its psychometric use tends to conceal. To call someone intolerant implies that they ought to endure something they do not endure. Consider a tenant who refuses to stay in a flat after noticing cracks spreading across a load-bearing wall, and who will not wait for the landlord’s surveyor to say whether the building is safe. Is this intolerance of uncertainty, or appropriate recognition of a potentially lethal problem? A neighbour who stays might be exceptionally tolerant, poorly informed, unable to afford to leave, or simply accustomed to the risk. The same conduct supports radically different classifications, and the classification chosen depends on a judgement about what the situation warranted, which the word tolerance silently imports.
These analyses reveal a familiar operation that is unproblematic in itself. Ordinary language routinely stabilizes heterogeneous coordinations into concepts that are extraordinarily useful for communication. In LVT terms, such concepts are L4 stabilizations: abstractions that gather many particular coordinations under a portable name that can be carried into new situations. L4 stabilization is no mistake, since without it there would be no shared language at all. The error arises when a useful classification is taken to designate an independently existing, unified object without anyone testing whether its instances share the common structure that such unity requires.
The old phrase “weasel words” is instructive here, provided it is stripped of any implication of evasiveness. The difficulty lies less in an intention to deceive than in the way familiar words give the appearance of clear reference while concealing unresolved questions about what they refer to. The weasel of the proverb sucks out the egg and leaves the shell intact; the word uncertainty arrives with its shell of everyday intelligibility unbroken, and nobody thinks to check what is inside. The principle that follows can be stated plainly. The intelligibility of an expression in ordinary language is no evidence of the ontological unity of the scientific object subsequently constructed from it.
III. Compound Abstraction and Recursive Proliferation
The expression “intolerance of uncertainty” has an apparently straightforward grammar. There is something called uncertainty, there is something called tolerance, and individuals are presumed to possess different degrees of capacity to tolerate the first. The construction is linguistically ordinary, and its ordinariness is precisely what gives it authority. Its simplicity nevertheless conceals several unresolved relationships. Uncertainty is an abstraction from particular relations between present coordination and anticipated developments. Tolerance is an abstraction from the different ways in which coordination continues, changes or fails when something becomes disruptive. Their conjunction presupposes that uncertainty can be specified well enough to serve as the object towards which a general disposition of tolerance is directed. That is exactly what remains unestablished. A compound concept does not acquire a stable referent merely because its constituent words are familiar; if anything, joining two heterogeneous abstractions multiplies the points at which the compound’s reference can drift.
Two processes should be distinguished here. Recursive accumulation occurs when a concept combines abstractions that already depend upon recursive processes of anticipation, interpretation and adjustment. Uncertainty already involves a living being’s anticipation of its own future coordination, and tolerance already involves its relation to its own ongoing response, so that the compound folds one recursive dependency into another. Recursive amplification occurs when the resulting classification re-enters living coordination and alters the processes it was introduced to describe. Someone comes to describe themselves as intolerant of uncertainty. They begin monitoring their reactions to unresolved situations. They interpret later apprehension through the classification, perhaps seek a treatment in which they are deliberately exposed to what they now classify as uncertain. At that point the concept has ceased to be merely a description and has become a multisymbolic counterpart within subsequent coordination, one whose presence can shape later questionnaire responses. Van Baar himself has suggested that young people’s greater familiarity with terms such as anxiety may make them more likely to report symptoms (Columbia Mailman School of Public Health 2026). The observation reaches further than a caveat about reporting. None of this proves that every use of the concept produces such feedback, but it identifies a mechanism that must be considered whenever psychological classifications are treated as stable observational categories across time.
The social-scientific concept of class offers an instructive parallel. A category such as class compresses vast ranges of embodied, relational, spatial, material and symbolic coordination into a portable classification. Once established, the category becomes consequential in its own right, organizing surveys, policies, self-descriptions and political mobilizations. It is not therefore fictitious, but the social reality it acquires through institutional circulation does not retrospectively prove that the original abstraction captured a discrete natural object. A psychological disposition can travel a comparable path, from heuristic description to clinical category to self-understanding, acquiring reality at each step without that reality vindicating the first step.
Ian Hacking’s account of the looping effects of human kinds is the obvious precedent. Hacking (1995) showed how classifications of people can change the people classified, who then change in ways that require the classifications to be revised. LVT extends this analysis in two directions. It supplies a more differentiated account of what the loop passes through, since a classification does not act on persons in general but enters particular mediations, engages particular types of recursive counterpart, and operates at particular recursivity levels. And it shows that recursivity is present at the origin as well as in the consequences. Hacking’s loop begins once a classification is applied, whereas the phenomena gathered under intolerance of uncertainty were recursively constituted before any classification arrived, so that the classification is recursive in its origin, in its application and in its effects.
The conceptual principle can now be formulated. Combining two recursively constituted abstractions does not yield a more determinate object; it may instead multiply the interpretive dependencies that must be resolved before the compound can be meaningfully applied to anyone. Recursivity should not be imagined as a quantity that literally multiplies, as though the compound contained twice as much of some substance. The claim concerns the proliferation of dependencies, each of which has to be settled before a number can mean what it is taken to mean.
IV. Reconstructing Uncertainty Through the Five Mediations, Five Levels and Three Counterparts
Critique alone would leave the field to the existing construct, so this section turns constructive. Its aim, which does not require classifying seventy-five kinds of uncertainty for the cells of a five-by-five-by-three grid, is to show that what psychology condenses into a single property can involve fundamentally different configurations of living coordination.
Consider first the five mediations. In multisensorial embodiment, someone awaits the results of a biopsy or notices an unfamiliar sensation in their chest. The vital stakes here are centrally bodily, but the bodily experience of apprehension must not be confused with the unresolved question to which it relates; a racing heart while waiting for results is part of the coordination rather than the content of the uncertainty. In multispecies being-with, someone wonders whether a partner will stay, whether a friend will reply, or whether a dog that has gone missing will find its way home. Such coordinations depend on the continuing responsiveness of others, and their future cannot be settled simply by obtaining more information, because the others concerned may revise what they intend. In multiversal dwelling, someone waits to see whether snow will close a mountain pass, whether the monsoon will arrive in time for planting, or whether a river will flood the lower fields. Here the counterpart is a non-human-made environment whose developments are indifferent to anyone’s anticipation of them and yet profoundly consequential for it. In multimateriality, someone does not know whether a train will arrive, whether an accommodation booking made online is valid, or whether a building with a cracked wall is safe. Technical infrastructure can transform dramatically what can be anticipated and what must be negotiated as events unfold. In multisymbolization, someone is uncertain about the meaning of a legal letter, an examination grade, a diagnostic label or an ambiguous text message. The uncertainty may concern the symbolic articulation itself rather than the material or relational processes it refers to: a person may know perfectly well what happened and still not know what the letter announcing it commits them to.
These are analytical emphases, and actual experiences cannot be sorted into five boxes, because every actual coordination is multimediated. The waiting patient is also in being-with clinicians and relatives, exposed to weather and season, dependent on the multimateriality of hospitals and transport, and caught in the multisymbolization of results and classifications. The analytical distinctions nevertheless suffice to remove any reason for assuming that the resulting experiences share one quantity called uncertainty. Waiting for a pass to open and waiting for a partner to decide are different kinds of waiting, rather than more and less of the same thing.
The three types of recursive counterpart sharpen the point. The mountain range is useful because its continued existence over a short horizon can be anticipated through nonrecursive processes that do not respond to anyone’s expectations about them. Burglary introduces something different, the independent actions of other people, whose intentions are formed in anticipation of how householders, neighbours and police will act and therefore cannot be treated as physical conditions awaiting observation. A romantic relationship provides an even clearer case of sustained interrecursivity, since each partner’s anticipation of the other’s response can change how both behave, so that the future of the relationship is jointly generated rather than discovered. Selfrecursivity introduces yet another structure. Someone may be uncertain how they will respond to a coming event such as an interview, and their anticipation of their own response can alter that response. These types should not be ranked as more or less predictable. A nonrecursive process can be chaotic or poorly understood, as weather notoriously is, and an interrecursive relationship can exhibit extraordinary stability over decades. The distinction concerns how anticipation enters, or does not enter, the unfolding process. Uncertainty about a nonrecursive counterpart can be reduced only by better knowledge of it, whereas uncertainty about an interrecursive counterpart may be reduced, increased or transformed by the very attempt to resolve it, as anyone who has asked a partner where the relationship is going will know.
The five recursivity levels address what happens when coordination becomes unsettled. At L1, coordination proceeds without any unresolved future needing to be formulated as a problem: a traveller follows a familiar route, a passenger keeps reading while the train is a minute or two late. At L2, something becomes misaligned. The train does not come, a message goes unanswered, a sensation becomes troubling, and something is felt to be wrong before anything has been said about it. At L3, the misalignment is articulated, as when the person tells themselves or someone else that they do not know what will happen, or that they are anxious. At L4, the experience is related to a classification already stabilized elsewhere, as when a person comes to believe they are generally intolerant of uncertainty, or a clinician records a score indicating that disposition. At L5, the classificatory framework itself becomes available for reflection, and the person, or the researcher, asks what uncertainty means and whether the classification fits.
The decisive question is how coordination moves between these levels. A felt misalignment at L2 need not prevent continuation at L1 across many other coordinations. Someone can remain apprehensive about a test result while cooking, teaching, caring for a child and sleeping reasonably well. Such continuation does not require suppressing awareness of the disturbance, and may instead involve accepting its unresolved status, rearranging practical priorities, or deliberately returning attention to something else. LVT calls the capacity to move between levels in this way recursive fluidity, and treats it, rather than permanent residence at L1, as the mark of vitality. What psychology calls tolerance of uncertainty might, in some circumstances, concern exactly this capacity to keep other coordinations going while a felt misalignment remains open. Even this possibility should not be promoted into another universal trait, since its manifestation and significance depend on the mediational configuration and the vital stakes involved. Remaining calm while awaiting the outcome of a job application and remaining calm while a child is missing are two very different coordinations, and a person who managed the first and not the second would reveal nothing about a deficit of tolerance.
The five-by-five-by-three framework is offered here as a discipline of inquiry, rather than as a new taxonomy for generating seventy-five subscale scores. Its value lies in allowing researchers to investigate what happens without prejudging the nature of the phenomenon, by asking which mediations carry the stakes, what type of counterpart is involved, and how coordination moves across levels over time. When the processes condensed under the name of uncertainty are reconstructed in this way, the apparent unity of the construct dissolves into configurations that differ in their stakes, their counterparts and their dynamics.
V. What Questionnaires Actually Measure: From Recursive Experience to Nonrecursive Number
The items of the IUS-12 display the heterogeneity at the level of the instrument itself. Respondents are asked how characteristic of them it is to be greatly upset by unforeseen events, to be frustrated by not having all the information they need, to want always to know what the future holds, to feel they should be able to organize everything in advance, to be unable to function well when uncertain, and to be stopped from acting by the smallest doubt. One item reads: “When it’s time to act, uncertainty paralyses me” (Carleton, Norton and Asmundson 2007). These statements concern different coordinative processes. Some describe emotional responses, some concern preferences for information, some concern practices of planning, and others concern the interruption of action. The psychometric literature itself registers the division, distinguishing a prospective factor concerned with the desire for predictability and active information-seeking from an inhibitory factor concerned with difficulty acting when uncertainty is present. Birrell and colleagues’ review of the factor-analytic studies is particularly revealing, since it found that existing definitions of the construct were categorical and lacked specificity, and then sought a definition in the items that consistently loaded together across studies (Birrell et al. 2011). The procedure is telling: the definition was sought in the covariance of the questions rather than in the phenomena the questions were about.
The existence of two psychometric factors does not establish that they belong to one underlying natural capacity. Their practical implications may even be opposed. Active information-seeking can be a highly effective way of handling an unresolved situation, as with the traveller who telephones three hostels before arriving and copes well precisely by refusing to leave the night open. Inhibition of action, by contrast, may impede coordination. A single total score adds the person who copes by planning to the person who cannot act, and reports both as more intolerant.
It helps to follow the production of a score step by step. The researcher formulates generalized statements, which are already L4 stabilizations, since each gathers countless situations under a single sentence. The respondent encounters these statements in a particular setting, a lecture theatre or a phone screen between other tasks. They interpret what the statements mean, recall or imagine relevant coordinations, weigh them in some unstated way, and produce an articulated judgement at L3. That judgement is selected from a numbered response scale. The selections are summed, and the sum is treated as the indicator of a disposition. Every stage involves interpretation and recursion, and none involves direct observation of the L1 and L2 coordinations about which the respondent has been asked to generalize. Respondents need not be dishonest or inaccurate for this to matter, because the questionnaire elicits a particular kind of symbolic performance, a self-description produced on request, and the score records that performance.
The distinction between reliability and ontological validity follows. Internal consistency shows that responses to items covary. Test-retest reliability shows that scores are stable under specified conditions. Factor analysis identifies patterns of covariance among items. Predictive validity establishes relationships between scores and independently measured outcomes. These are real achievements. None of them, however, singly or automatically in combination, shows that the score represents an independently existing entity of the kind proposed. Borsboom, Mellenbergh and van Heerden (2004) made a related argument from within psychometrics, holding that a test is valid for an attribute only if the attribute exists and variation in it produces variation in test scores, which makes validity an ontological claim that correlational evidence cannot settle by itself. The LVT objection goes further, by asking whether the attribute has been individuated at all, that is, whether there is a single something whose variation could do the producing.
The parallel with self-efficacy is exact. Self-efficacy questionnaires capture a real signal from an entire mediational configuration, comprising the patient’s body, relationships, home, resources and symbolic frameworks, reduce it to a number, and then treat the number as the cause of the configuration it imperfectly reflects (Ecks 2026). The intolerance-of-uncertainty questionnaire may perform the same transformation on a different configuration. A person whose rent rises unpredictably, whose employment is casual and whose family lives on another continent may well agree that unforeseen events upset them greatly, and the score will register that agreement as a fact about them. The questionnaire does not discover an intrinsically numerical object; it establishes a procedure through which heterogeneous recursive coordinations become expressible as comparable numbers. Once expressed in that form, the coordinations acquire the appearance of a nonrecursive property, a quantity that sits still while it is measured, even though everything that produced it was recursive.
VI. When Uncertainty Correlates with Anxiety: The Problem of Recursive Double-Counting
The most consequential use of intolerance-of-uncertainty scores is relational. Gentes and Ruscio (2011) examined the cross-sectional association of the construct with symptoms of three disorders. Using the definition developed in research on generalized anxiety, they reported mean correlations of 0.57 with symptoms of generalized anxiety disorder, 0.53 with symptoms of major depression and 0.50 with obsessive-compulsive symptoms. McEvoy and colleagues (2019) later extracted 181 studies with 52,402 participants and 335 independent effect sizes, finding an overall correlation of 0.51 between intolerance of uncertainty and symptoms across seven groups of disorders, with substantial heterogeneity across studies. They concluded that the construct’s transdiagnostic character had been robustly established. These are real statistical results, obtained with care across large literatures. The LVT question is what kind of relationship they establish.
Anxiety must be subjected to the same scrutiny as uncertainty, rather than left standing as the unquestioned outcome against which a questionable predictor is validated. The word can designate bodily apprehension, the anticipation of harm, fear of evaluation, persistent worry, disturbed sleep, avoidance of action, or distress about one’s own distress. These processes are profoundly multimediated. They are also characteristically selfrecursive. Someone who fears becoming anxious may become anxious about that possibility; their awareness of bodily arousal alters the arousal; their anticipation of future distress changes how they coordinate in the present. In interrecursive situations the responses of others enter the same loop, as when a partner’s reassurance is itself read as a sign that something must be wrong. Anxiety cannot simply be assumed to be a discrete state existing independently of the living coordination in which it arises. The clinical usefulness of identifying anxiety symptoms is not in question. The objection concerns the promotion of these descriptions into unified explanatory entities, which can then stand on either side of a correlation, without sufficient analysis of what they individuate.
Consider a person awaiting a diagnosis after a scan has found a shadow on a lung. Over the following fortnight they experience bodily apprehension, search repeatedly for information, rehearse future consequences, find it difficult to concentrate on work, sleep badly, and tell friends that they are worried sick. Suppose this person is asked to complete both an intolerance-of-uncertainty scale and an anxiety inventory. The first asks whether unforeseen events upset them, whether lacking information frustrates them, and whether uncertainty stops them functioning. The second asks whether they feel nervous, whether they cannot stop worrying, and whether they have trouble relaxing. The same ongoing coordination will generate high answers on both, and the two scores will correlate because the instruments attend to overlapping aspects of one evolving process. The correlation may then be interpreted as evidence that a property called intolerance of uncertainty predisposes the person to a state called anxiety. Yet the existence of two instruments and two scores does not by itself establish the existence of two phenomena.
This possibility can be named recursive double-counting: the classification of overlapping features of one recursively unfolding coordination as two different properties, the measurement of each with its own instrument, and the interpretation of the resulting association as a relationship between two things. It claims neither that all such correlations are tautologies nor that intolerance of uncertainty and anxiety are empirically identical, but something narrower and harder to evade: statistical associations between classifications derived from overlapping coordinations cannot independently establish the ontological separateness of those classifications. Recursivity makes the problem acute rather than incidental. Because selfrecursive processes alter in relation to their own ongoing state, the person’s distress about not knowing and their distress about being distressed feed one another, so that any division of the process into a predisposing factor and a resulting symptom draws a line through a loop.
Psychology has begun to recognize part of this problem. The jangle fallacy, named by Truman Kelley (1927), is the assumption that two measures with different names must measure different things. Fitzgerald and colleagues (2026) asked directly whether anxiety sensitivity, intolerance of uncertainty and experiential avoidance, three processes widely treated as transdiagnostic foundations of anxiety pathology, are adequately distinct from one another. Using data from 300 treatment-seeking individuals, they were able to differentiate the constructs, particularly at the level of subfactors, while also finding problems of extrinsic convergent validity, that is, redundancy in how the constructs related to external criteria. Earlier meta-analytic work on six cognitive vulnerabilities to depression and anxiety, intolerance of uncertainty among them, had found that their intercorrelations were best represented by a single common factor (Hong and Cheung 2015). Psychometric differentiation is meaningful. If measures have distinguishable structures and predict different outcomes, they are not simply interchangeable descriptions, and an argument that ignored this would be easily dismissed.
Statistical distinction is nevertheless not equivalent to ontological independence, since two measurements can register different features of one complex, evolving coordination without revealing two separate causes, much as a thermometer and a barometer register different features of one weather system without either reading being a cause of the other. The conceptual problem is therefore deeper than the jangle fallacy. The question is not only whether different names designate the same construct, but whether the constructs themselves are legitimate individuations of living coordination at all. Factor analysis can show that items cluster, but it cannot show that the clusters correspond to the joints of the process from which the answers were drawn.
The progression from correlation to explanation makes the issue consequential. The literature moves readily from the finding that intolerance of uncertainty correlates with anxiety to the suggestion that it contributes causally to anxiety, and from there to the proposal that interventions targeting it will reduce anxiety. If two scales partly capture one unfolding process, their association may be substantial without either measuring a separate cause of the other. Longitudinal prediction does not resolve the issue automatically. An earlier articulation of distress about unresolved developments may predict later anxiety symptoms simply because the same coordination continues or deepens, and the person with the shadow on their lung who scores high in the first week is likely to score high on anxiety in the third. That prediction may be clinically useful, but it does not establish that the earlier measure isolated an independent mechanism. A causal claim would require evidence of distinguishable processes, of appropriate temporal ordering between them, and of interventions or observations capable of discriminating between the hypothesis of two interacting mechanisms and the hypothesis of one coordination described twice.
What emerges is a new form of unearned confidence. Psychological research has been shown to misidentify distributed coordination as a property of individuals (Ecks 2026). The present analysis shows how such properties are then related statistically to one another, and how the resulting correlations come to appear as corroboration of the constructs’ independent existence, even where the relationship partly reflects the conceptual procedures through which the constructs were produced. Each new correlation adds to the impression that the construct is real, because real things are what enter into relationships. Psychology may have discovered, rather than one psychological entity causing another, that two ways of describing an unfolding coordination tend to occur together. What appears to be an explanatory relationship may partly be the statistical residue of conceptual division.
VII. The Historical Transformation of Predictability: Why 2012 Matters, but Not in the Proposed Way
The historical finding can now be revisited, and here LVT does more than dissolve a conclusion, since it generates historically substantive questions that the construct obscures. Consider the contrast between travelling through Europe on an Interrail pass in the 1980s and making the same journey today. An earlier traveller could arrive in an unfamiliar city at nightfall without knowing where they would sleep. Finding a bed meant walking from the station, asking at hostels, negotiating with landladies, watching the sky for rain, and adjusting plans as the evening went on. A contemporary traveller can book a room weeks ahead, inspect photographs, compare prices, read reviews and follow a blue dot on a screen to the door. The organization of coordination has changed profoundly, and many situations that once required active coordination amid open possibilities can now be organized around reliable expectations. This is a change in multimateriality (booking platforms, phones, payment systems), in multisymbolization (reviews, confirmations, maps) and in multispecies being-with (hosts are now strangers met through ratings rather than across a doorstep), while dwelling, the night, the weather and the unfamiliar terrain, remains what it was but is now encountered through a thicker multimaterial buffer. None of this is a change in individual psychology.
The shift in expectations may matter most. When reliable information becomes an ordinary condition of L1 coordination, its absence becomes a more conspicuous L2 disruption. The traveller of 1985 who arrived without a bed was in an ordinary situation; the traveller of 2025 whose booking has been cancelled on arrival is in an emergency. The same objective openness is lived very differently depending on what the surrounding coordination has made normal. A questionnaire item about being upset by unforeseen events could thus receive higher endorsement in a world with fewer unforeseen events, because each one now stands out against a background of reliability.
Technological gains in predictability are, moreover, uneven. In some domains information and planning have improved dramatically, while in others long-term arrangements have become less secure. A young adult can know to the minute when the next train will arrive while being unable to know whether they will be able to afford housing in five years, whether their degree will lead to employment, or whether the region they live in will remain climatically hospitable. These are not two quantities that can be summed to produce an overall level of uncertainty, since they differ in mediation, horizon and counterpart. Information technologies may also generate new expectations of availability from other people. A message marked as delivered and read, with no reply, produces a distinctive kind of open question, in which the nonrecursive transmission is certain while the interrecursive response remains unknown and now visibly withheld. Such technologies can increase predictability in nonrecursive processes while intensifying tension in interrecursive ones.
Digitally mediated intimacy illustrates the hypothesis. Dating applications expand the information available about potential partners, and messaging makes communication immediate and its absence visible. More information does not, however, remove the independent recursivity of another person, whose intentions remain open to revision by the very interaction that is supposed to disclose them. The hypothesis, then, is that digital coordination raises expectations of immediate information in precisely those situations where the underlying interrecursive relationship remains open by its nature, and that reports of distress about delayed replies, inconsistent communication or unclear commitment partly reflect these changed expectations and communicative arrangements. This is a hypothesis requiring investigation, not an established generational fact.
The rise in scores might therefore reflect changing expectations of predictability, changed experiences of L2 disruption, changing self-descriptions shaped by the circulation of psychological vocabulary, altered material conditions such as housing costs and casual employment, changing institutional demands in universities, or real changes in people’s capacities to continue coordinating amid unresolved difficulties. These possibilities can coexist, and they are empirically distinguishable. The acceleration after 2012 is suggestive, but the study does not establish that smartphones caused it, and its authors do not claim that it does. What LVT contributes is a specification of what historical research would need to investigate. The central claim is that greater reported intolerance is compatible with greater objective predictability across many everyday coordinations, and may even be partly produced by it.
VIII. Towards a Science of Living Coordination
Conceptual scepticism would be a poor place to end. Once generalized traits are set aside, inquiry can begin from actual situations instead. A research programme on a person awaiting an important medical result would establish what is at stake, what remains unresolved, what the person is doing, what resources and relationships are available, and how coordination changes over the weeks of waiting. It would identify which mediations carry the stakes, whether the decisive counterparts are nonrecursive, selfrecursive or interrecursive, and how the person moves between L1 and L5: when the matter is held at L2, when it is articulated, when it is classified, and when the classification itself is questioned. The purpose is to establish which details are consequential for the coordination under investigation, rather than to record every detail that could be recorded.
A different key question would replace presumed tolerance: whether, and how, a person continues other important coordinations while an unresolved difficulty remains present. This can be studied through observation, interviews, repeated situation-specific reports over time and clearly specified behavioural outcomes, and quantitative measurement has a place wherever the observable relationships are well enough defined. Researchers could compare what happens to travellers whose accommodation is cancelled at short notice, varying their access to money, transport, alternative beds and local support. A high level of distress among those without money or support would then indicate a severe practical disruption rather than deficient tolerance. Equally, researchers could examine whether people who describe themselves as poor at handling uncertainty nevertheless adapt effectively when unfamiliar situations actually arise. Such findings would distinguish generalized self-description from demonstrated coordinative capacity, a distinction the existing instruments cannot draw.
From this follows a protocol for examining any proposed psychological construct. For each candidate entity one would ask what living coordinations it describes, which mediations and types of recursive counterpart it involves, how it is differentiated from neighbouring concepts in terms of the processes it picks out as well as statistically, which recursivity levels the evidence for it actually accesses, whether measuring it alters the process measured, and what would count as evidence against its proposed unity. The governing requirement is that a concept retain a recoverable relationship to the coordinations from which it was abstracted, so that a researcher can always say of which situations, under which conditions, the number is a number.
LVT must apply the same requirement to itself. The five mediations, the five recursivity levels and the three types of recursive counterpart are themselves L4 scientific stabilizations. Their justification cannot rest on internal coherence or elegance, and must rest instead on what they allow researchers to distinguish and investigate that would otherwise remain invisible. The test is the one applied above to intolerance of uncertainty, namely whether the concepts keep open the route back to what is actually happening. That reflexive commitment is what separates a stronger science from a rival set of privileged abstractions.
Conclusion: The Problem Is Not the Numbers
The researchers behind the historical meta-analysis assembled tens of thousands of observations, applied established scoring procedures and identified a substantial historical change. The results are statistically meaningful. The proposed interpretation, however, rests on an ontological transformation that statistical procedures cannot independently justify. Uncertainty, intolerance and anxiety, far from being self-evidently separable entities, are conceptual abstractions from living coordination, often involving recursive processes that change through their own anticipation, interpretation, articulation and classification. The compound expression “intolerance of uncertainty” lends apparent unity to two already heterogeneous abstractions. The questionnaire renders the compound numerically comparable. The historical meta-analysis gives the resulting score a temporal trajectory. The correlations with anxiety place it in a network of relationships with other scores, which may derive from overlapping aspects of the same ongoing coordination. Together these steps produce a remarkably persuasive scientific object, one that may owe much of its apparent unity to the procedures through which it has been classified and measured.
None of this denies the reality of apprehension, suffering, paralysis or disrupted action, or the practical difficulties people face when important matters remain unresolved. These experiences deserve more careful scientific attention than a generalized trait allows. The task of a science of living coordination is not to deny what psychological measurements record, but to discover what is actually happening before deciding what, if anything, their numbers measure.
Selected References
Birrell, Jane, Kevin Meares, Andrew Wilkinson, and Mark Freeston. 2011. “Toward a Definition of Intolerance of Uncertainty: A Review of Factor Analytical Studies of the Intolerance of Uncertainty Scale.” Clinical Psychology Review 31 (7): 1198–1208.
Borsboom, Denny, Gideon J. Mellenbergh, and Jaap van Heerden. 2004. “The Concept of Validity.” Psychological Review 111 (4): 1061–1071.
Carleton, R. Nicholas, M. A. Peter J. Norton, and Gordon J. G. Asmundson. 2007. “Fearing the Unknown: A Short Version of the Intolerance of Uncertainty Scale.” Journal of Anxiety Disorders 21 (1): 105–117.
Columbia Mailman School of Public Health. 2026. “Sharp Rise in Key Psychological Risk Factor May Help Explain Growing Anxiety Among Young Adults.” News release, 18 November.
Ecks, Stefan. 2026. “Unearned Confidence: Why Psychology Mass-produces Compelling But Misleading Claims.” Living Value Theory, livingvaluetheory.org.
Fitzgerald, Hayley E., et al. 2026. “Associations among Anxiety Sensitivity, Intolerance of Uncertainty, and Experiential Avoidance: Are These Key Processes Underlying Anxiety-Related Pathology Adequately Distinct from Each Other?” Journal of Anxiety Disorders 123: 103215.
Freeston, Mark H., Josée Rhéaume, Hélène Letarte, Michel J. Dugas, and Robert Ladouceur. 1994. “Why Do People Worry?” Personality and Individual Differences 17 (6): 791–802.
Gentes, Emily L., and Ayelet Meron Ruscio. 2011. “A Meta-Analysis of the Relation of Intolerance of Uncertainty to Symptoms of Generalized Anxiety Disorder, Major Depressive Disorder, and Obsessive-Compulsive Disorder.” Clinical Psychology Review 31 (6): 923–933.
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.
Hong, Ryan Y., and Mike W.-L. Cheung. 2015. “The Structure of Cognitive Vulnerabilities to Depression and Anxiety: Evidence for a Common Core Etiologic Process Based on a Meta-Analytic Review.” Clinical Psychological Science 3 (6): 892–912.
Kelley, Truman L. 1927. Interpretation of Educational Measurements. Yonkers-on-Hudson, NY: World Book.
McEvoy, Peter M., Matthew P. Hyett, Sarah Shihata, Jordan E. Price, and Laura Strachan. 2019. “The Impact of Methodological and Measurement Factors on Transdiagnostic Associations with Intolerance of Uncertainty: A Meta-Analysis.” Clinical Psychology Review 73: 101778.
van Baar, Jeroen M., et al. 2026. “Intolerance of Uncertainty among US and Canadian Young Adults Attending College, 1992–2023: A Meta-Analysis.” JAMA Psychiatry. https://doi.org/10.1001/jamapsychiatry.2026.3843.