Abstract. This article reverses the usual starting point of theories of prediction. Instead of beginning with an uncertain future and asking how organisms anticipate it, it begins with the extraordinary continuity of living coordination and asks how explicit prediction emerged as a specialized human practice within that continuity. Drawing on Living Value Theory, it argues that living beings sustain their worlds through continuous remediation, and that the resulting mesocosmic stability rests on two foundations: nonrecursively stabilized counterparts, both material and symbolic, and the characteristic regularity of recursive remediation itself. Recursivity, including interrecursivity, therefore reproduces stability far more often than it generates transformation. Prediction in the strict sense is defined narrowly as an L3-plus multisymbolic practice that articulates an assessable later occurrence, and is distinguished from continuous coordination, practical preparation, anticipation, planning and promise. Because every forecast isolates a selected feature from a massively mixed futurity, there is no predictability in general, only predictability specific to particular mediations, counterpart types, temporal horizons and outcomes. Mathematical models of processes with mixed recursivity capture their nonrecursive regularities, and their internal validity must be distinguished from their fit with living coordination. An evolutionary reconstruction, running from bacterial chemotaxis and circadian clocks to the Babylonian astronomical diaries, shows that the coupling of multimateriality and multisymbolization made futures recursively shareable while also generating new forms of future openness. The strongest of these is interrecursive indeterminacy, exemplified by partner formation, in which the future configuration is produced through encounters rather than disclosed by better information. Engaging Friston's free-energy principle, the article separates mathematical description, physical implementation and ontological constitution, and proposes criteria for telling substantive biological inference apart from redescription. Its governing claim is that the extraordinary predictability of living coordination makes prediction possible, while the recursive transformations of living coordination establish the limits of what can be predicted.
I. Introduction: The Problem of Prediction Has Been Inverted
A person walks down a familiar staircase in the morning. She does not predict that the next step will be there, that gravity will continue to act, that the staircase will hold, or that her legs will go on carrying her. She simply goes down. Her feet find each tread at the expected height, her weight shifts, a hand brushes the rail, and the descent unfolds as one continuous coordination of body, building and planet. Describing this as a sequence of predictions, each confirmed or disconfirmed by the next footfall, imposes an elaborate epistemological activity on something that has none of its features. Nothing is articulated, nothing is selected as the object of a forecast, and nothing is assessed. Life continues.
Later that day the same person tells a colleague that a stock market index will rise tomorrow. This is an explicit symbolic practice of an entirely different kind. It selects one future state, identifies one variable, and makes an assertion that can be checked the following afternoon. Yet almost everything that makes the forecast meaningful is taken for granted: that the exchange will open, that the securities will still exist, that currencies, settlement systems, reporting standards, communication networks and the law of contract will all continue much as they are. The forecast is possible because the surrounding world is overwhelmingly reliable. It singles out a thin, volatile edge of a vast and stable coordination, and treats that edge as the future.
The usual question in theories of anticipation, from biology to economics, is how living beings predict an uncertain future. This article asks a different question: how do living beings continuously sustain remarkably stable worlds, and how did explicit prediction emerge as a specialized practice concerned with selected future differences within them? The conventional starting point misleads in three ways. It mistakes a comparatively specialized symbolic activity for a general feature of living coordination. It exaggerates the ontological primacy of uncertainty while neglecting the extraordinary regularity of both the counterparts through which living beings coordinate and the processes through which they do so. And it fails to explain how evolution, above all the emergence of human poietic capacities, produced at once increasingly powerful means of forecasting and new forms of future openness.
The target of the argument is not prediction, which is often extraordinarily useful. The target is the assumption that predicting is the fundamental activity through which life relates to a separate future, an assumption shared, in different idioms, by theories of anticipatory systems (Rosen 1985), by the free-energy principle and predictive processing (Friston 2010), by much work on algorithmic forecasting, and by social theories that oppose a conservative, prediction-driven closure of the future to an open, creative anticipation (Adams, Murphy and Clarke 2009; Esposito 2011; Beckert 2016). Against all of them, the article develops a positive theory of predictability within Living Value Theory, whose governing claim can be stated in one sentence: the extraordinary predictability of living coordination makes prediction possible, while the recursive transformations of living coordination establish the limits of what can be predicted.
II. Continuous Remediation and the Extraordinary Stability of the Mesocosm
Many theories of anticipation rest on a temporal picture in which time is divided into a completed past, an instantaneous present and an unactualized future. The organism stands at the knife-edge of the present and must continually orient itself toward what lies beyond. The picture is intuitive, and it is the natural self-description of creatures who speak about time. Whether it adequately describes living at the first two levels of recursivity, absorbed coping and felt misalignment, is another matter. The mesocosm, the historically constituted world in which a living being’s coordination takes place, is not ordinarily encountered as a succession of isolated presents (Ecks 2026c). Breathing does not begin again at each instant. The body goes on, the staircase goes on, the household and the street go on, and the living being goes on with them. Temporal succession and biological irreversibility are entirely real, and nothing here denies temporal direction. What is questioned is the universal explanatory priority of a sharply divided past, present and future, which becomes a powerful distinction within symbolic articulation but should not be projected onto every living process (Ecks 2026d).
Continuity must not be confused with immobility. Three phenomena need to be distinguished. Persistence is the continued presence of something over an interval, as when a building stands or a path remains usable. Recurrence is the regular repetition of a pattern, as in breathing, sleeping, eating, the seasons and institutional routines. Continuous transformation is the change of a process that nonetheless reproduces a recognizable organization, as when a body grows older while remaining the same living being. These are different forms of stability, and breathing shows how they combine. Its continuity consists of regular differences between inhalation and exhalation, and the organism persists by changing in recurrent ways rather than by avoiding change. Mesocosmic stability, in short, is the reliable continuity of coordination through change, and never the absence of change.
Each of the five mediations contributes its own forms of continuity. Multisensorial embodiment contributes metabolic rhythms, bodily capacities, practised movement and stable sensory discrimination. Multispecies being-with contributes repeated coordination among living beings, from the mutual accommodation of a household to the seasonal return of the birds in a garden and the microbial populations that accompany every body. Multiversal dwelling contributes the continuity of conditions that no one made: gravity, the alternation of day and night, the seasons, terrain and climate. Multimateriality contributes the persistence and maintenance of everything made, from tools and staircases to buildings and infrastructures. Multisymbolization contributes the stability of names, classifications, conventions, records and shared forms of intelligibility. These mediations cannot be reduced to one another, and their temporalities differ enormously (Ecks 2026f). Their ongoing coordination nonetheless makes the mesocosm reliable enough for life to continue, and this reliability is an outcome of coordination rather than a goal towards which it is directed.
This yields the first two propositions of the article, which differ in status. The first is an ontological commitment, the continuity principle: living coordination proceeds through continuous remediation rather than through serial encounters with a separately constituted future. The second is an empirical hypothesis, the mesocosmic stability hypothesis: many mesocosmic coordinations exhibit extraordinary persistence and recurrence at the temporal scales relevant to their ongoing reproduction. As a hypothesis it must be assessed domain by domain, with temporal scale, aspect and level of organization specified, since a coordination can remain highly stable in its organization while its constituent processes change considerably. Rather than claiming that the world changes slowly, it claims that the reliability of mesocosmic coordination is among the most important facts a process ontology must explain, and that predictability begins there, with the continuity of living coordination, and not with symbolic attempts to forecast its future.
III. Two Foundations of Predictability: Stabilized Counterparts and Recurrent Remediation
The first foundation of predictability is the prevalence of nonrecursively stabilized counterparts. A counterpart is nonrecursive when it participates in coordination without altering in relation to its own ongoing state or to the responses of those engaged with it. Nonrecursivity must not be identified with inert matter. A material arrangement can function nonrecursively, but so can an established symbolic form. A familiar word does not have to be renegotiated each time it is spoken, a road sign does not reconsider its own designation, and an institutional role or procedure offers a ready-made coordination that can be taken up without being reinvented. Such deposits, material and symbolic, are far from passive backgrounds. They make enormous quantities of coordination immediately available, and much of what feels like the predictability of daily life is the reliability with which they are there when needed. Nonrecursivity is nonetheless no guarantee of stability. A storm, a failing generator and a contaminated batch of medicine are all nonrecursive, and all can be highly unpredictable. What nonrecursive counterparts do not do is respond to the expectations held about them, and this is one reason why their behaviour, when regular, can be forecast with confidence.
The second foundation is less familiar, and it is the principal theoretical advance of this article. Recursive remediation is itself frequently highly regular. Breathing, digestion, the regulation of posture, habitual movement, the turn-taking of conversation and the routines of institutional work, far from merely taking place against stable backgrounds, exhibit characteristic patterns, cycles and regularities of their own. No one breathes at exactly three-second intervals, and respiration responds continually to exertion, emotion and altitude. The important fact is that it is highly recurrent and responsive within physiologically viable ranges. The stability of living coordination is therefore generated in large part by the regularity of the very processes that continually remediate it.
Increasing recursivity accordingly does not mean increasing unpredictability. A walker who stumbles adjusts her footing and goes on walking. Two people walking together adjust to each other and keep a shared pace. A long-married couple responds to small changes in each other’s conduct while reproducing a relationship whose patterns their friends could describe in detail. A hospital adapts to each day’s admissions while remaining a recognizably continuous organization. In each case responsiveness preserves an ongoing coordination. Three outcomes of recursive activity can be distinguished: recursion that maintains an existing coordination, recursion that modifies it within recognizable limits, and recursion that reopens or transforms a previously stable coordination. These are possible outcomes of recursive activity and not additional levels of recursivity. Recursive opening, in which a settled coordination becomes questionable and is changed, is real and consequential, but it is the exception against a background in which recursion overwhelmingly maintains.
The two foundations are analytically independent. A hospital building may remain stable while its internal activities vary from hour to hour; a clinical workflow may be regular while the equipment used in it is replaced; a bank may keep its name and legal identity while its personnel and practices are transformed. The stability of a counterpart and the regularity of its remediation must therefore be investigated separately, and the distinction becomes essential whenever institutional or technological change is at issue.
The third and fourth propositions follow. The double foundation of predictability, which is an ontological claim with empirical consequences, holds that mesocosmic reliability depends both on stabilized counterparts and on the characteristic regularity of remediation itself. The recursive stabilization principle holds that recursivity, including interrecursivity, can reproduce stability as well as generate transformation. This corrects a widespread association, running through social theory from accounts of reflexive modernity to theories of interaction and relationality, between reflexivity and indeterminacy. Recurrent recursive adjustment is among the principal means through which the stability of the mesocosm is achieved.
IV. What Prediction Is, and What It Is Not
A theory of predictability needs a deliberately narrow definition of prediction. In what follows, prediction is a multisymbolic practice in which a specified state, event or difference at a later time is articulated as a possible or expected occurrence, in a form that can in principle be assessed against what subsequently happens. The definition excludes much that is conventionally called prediction. A physiological rhythm is not by itself a prediction, nor is a sensorimotor adjustment, a learned habit or an organism’s response to its environment. None of these articulates a later state, and none can be right or wrong about it in the way a forecast can.
Several adjacent phenomena must be kept apart. Continuous coordination is living activity that proceeds without representing any later condition, as on the staircase. Practical preparation arranges the present for conditions that may arise, as when a squirrel caches nuts or a household stacks firewood, possibly without any formulated forecast. Anticipation is the broader orientation toward possible developments, which may be embodied, felt or symbolic: the tensing of a body before a blow, the dread before a difficult conversation. Prediction in the strict sense is an explicitly articulated proposition or model about what will occur. Planning coordinates present activities toward an intended outcome. Goal-setting establishes a future-oriented stake. A promise or commitment is a present symbolic act that alters subsequent coordination by creating obligations, and it shapes the future without forecasting it. A forecast is not automatically a goal, a goal is not automatically a forecast, and a promise, which binds rather than foretells, is neither.
Prediction in the strict sense requires at least L3 symbolic articulation. At L3 a person can formulate or question a forecast: will it rain tomorrow? At L4, predictions are stabilized through recognized classifications, procedures, models, professional standards and institutions, as in a national weather service or a central bank’s inflation projection. At L5 the predictive frameworks themselves become objects of reconsideration, as when economists ask whether their models could ever have anticipated a crisis. A short forecast is not thereby an L3 forecast, since the level depends on how the forecast is recursively organized and stabilized and not on its linguistic simplicity. L1 and L2 remain indispensable to the practices through which predictions are produced and used, from the meteorologist’s practised reading of a chart to the trader’s felt unease, but they must not themselves be relabelled as explicit prediction.
A further clarification concerns the future itself. Recursive engagement takes place through actual living mediation. A person can engage recursively with an actual symbolic representation of a possible future, such as a forecast, a plan or a fear put into words. The unactualized future occurrence is not yet a coordination whose recursivity can be investigated. This does not make future possibilities unreal as present stakes. Their present symbolic and material mediations, from a weather warning to a mortgage, can exert enormous influence on what people do now.
Finally, prediction and predictability are different concepts. Prediction is a practice. Predictability is the reliability with which particular features of a process can be projected over a specified interval, given available information, methods and conditions. That one feature can be predicted establishes nothing about the predictability of the mesocosm as a whole. Two further propositions are definitional in status. The symbolic specificity of prediction holds that prediction in the strict sense is an L3-plus multisymbolic practice articulating an assessable later occurrence or condition. The selection principle holds that every prediction isolates particular features of future coordination while presupposing extensive background continuities. Explicit prediction is therefore a narrow symbolic practice nested within much broader forms of temporal coordination.
V. Why There Is No Predictability in General: Massively Mixed Futurity
Consider four predictions: that a stock market index will rise tomorrow; that the same index will be higher in ten years; that demand for a consumer product will grow over the next year; and that a particular film will find an audience. Forecasting practice tends to treat these as variations in the amount of one homogeneous uncertainty, to be expressed as probabilities and confidence intervals. They are in fact predictions about fundamentally different arrangements of mediations, recursivity levels, counterparts and temporal scales.
Predictability varies across the mediations. In a financial market, accounting conventions, exchange procedures and legal forms are highly stable multisymbolic and multimaterial deposits, while particular prices are volatile. In consumer demand, the embodied needs that a product answers may recur with great regularity, while fashions, household arrangements, incomes, material availability and symbolic distinctions shift around them. A film’s multimaterial production is finished and fixed long before its reception, which will be generated through encounters among audiences, critics, distributors and platforms that have not yet taken place. A single probabilistic forecast compresses these heterogeneous processes into one quantified outcome. The compression may be useful, but it is not an ontologically adequate description of what produces the outcome.
Predictability also varies across counterpart types. Where the counterparts are nonrecursive, a forecast does not alter what it forecasts: an eclipse is indifferent to the almanac. Where the counterpart is selfrecursive, a living being’s engagement with its own condition alters its subsequent conduct, so that a patient told her prognosis may change her life in ways that change her prognosis. Where counterparts are interrecursive, participants respond to one another’s expectations, declarations, forecasts and actions. A market forecast enters the market it describes, which is why financial models can function, in Donald MacKenzie’s phrase, as an engine and not a camera (MacKenzie 2006), and why Robert Merton’s self-fulfilling prophecy, a false definition of a situation that evokes behaviour making it come true, was first illustrated by a run on a bank (Merton 1948). A film predicted to fail may be given a smaller release and fail for that reason. Interrecursivity does not, however, entail instability. The same interrecursive processes reproduce durable conventions, prices that hold within narrow bands for years, and markets that clear every day.
The temporal horizon can change the very object of prediction. Forecasting the market ten years ahead is not a longer version of forecasting it tomorrow. Over a decade regulations, technologies, institutions and the very kinds of thing classified as financial assets may change. The difficulty is not only that errors accumulate. The mesocosmic configuration that makes the forecast intelligible may itself be transformed, so that the question that was asked no longer has the same referent when the answer arrives. This places principled limits on treating long-horizon prediction as extrapolation from a stable process.
These observations can be gathered under the concept of massively mixed futurity. Every human mesocosm contains innumerable simultaneously recurring coordinations, each with its own rhythm, degree of stability and possibilities of transformation, and their pasts and futures are mixed in the sense that what is settled in one coordination is open in another (Ecks 2026d). These coordinations continually intersect, and their intersections can be consequential even when each remains regular on its own terms. Predictability therefore depends not on the properties of separate processes alone but on their changing coordination. The five mediations, five recursivity levels and three counterpart types furnish an ontological means of investigating this heterogeneity. They are not a classification table from which forecasts could be generated mechanically; they identify the different kinds of coordination whose properties require empirical investigation in each case.
Two consequences follow for mathematical forecasting. The first concerns what a model captures. Complex living processes have mixed recursivity: some of their aspects are nonrecursive, some recursive. Any mathematical model of such a process latches onto its nonrecursive aspects, the regularities that persist whatever anyone expects of them, and cannot model the recursive aspects as recursive (Ecks 2026e). The regularity of recursive remediation described in the previous section is captured by models precisely as a nonrecursive pattern, a rhythm or a distribution, while the recursion that produces it remains outside the model. The second concerns the difference between knowing a distribution and knowing a generative process. Population statistics can represent with great accuracy the distribution of outcomes produced by millions of recursive trajectories without recovering the generative history of any one of them. Assortative mating correlations above .5 for political values, religiosity and education are robust, yet each is compatible with several different mechanisms of partner formation, and none tells how one particular stranger became, to one particular person, irreplaceable (Ecks 2026a). This takes nothing away from statistics. It clarifies what statistical regularity establishes and what it does not.
The heterogeneity principle, the seventh proposition and an empirical claim open to comparative test, follows: predictability is specific to particular mediations, recursive configurations, temporal horizons and selected outcomes. Every prediction isolates a selected feature from a massively mixed future, and its validity cannot be generalized to the coordination as a whole.
VI. The Evolutionary Emergence of Prediction and Unpredictable Possibilities
The evolutionary starting point is continuous zoetic remediation, not forecasting. The emergence of life presupposes regularities in chemistry and physics, and metabolism depends on repeatable reactions sustained through recurrent exchanges with an environment. Living systems maintain their organization by regulation and responsiveness, and none of this requires any symbolic representation of future states. This is not to deny early organisms anticipatory organization. Many bacteria swim up chemical gradients by comparing present concentrations with those encountered moments before, extending runs while conditions improve and tumbling when they worsen. Cyanobacteria possess circadian clocks whose cycle runs in step with the alternation of day and night, so that photosynthetic machinery is prepared before dawn. These are remarkable achievements, and they are routinely described as anticipation. Robert Rosen’s definition of an anticipatory system as one containing a predictive model of itself or its environment that allows it to change state now in accordance with the model’s predictions about a later instant (Rosen 1985) was formulated precisely to capture them. Yet what they show is something more basic than a model. The bacterium’s comparison and the cyanobacterium’s clock are recurrent remediations coupled to the regularities of chemistry and of multiversal dwelling. The clock is regular because the planet’s rotation is regular, and the coupling between them was produced by selection over countless generations. Such organization should not automatically be equated with multisymbolic prediction, and calling it prediction obscures what it demonstrates, namely how deeply the regularity of dwelling is built into the regularity of life.
Evolution then expanded the range of variable encounters. The contrast between rooted plants and mobile animals is useful as a heuristic, provided it is not turned into an equation of autotrophy with stability or of heterotrophy with mobility; plants respond intricately to light, touch, herbivores and neighbours, and many animals are sessile. Mobility nonetheless enlarges the importance of navigation, search, pursuit, escape and chance encounter. Predators and prey continually respond to each other, and social animals coordinate through long histories of mutual responsiveness. These developments greatly increase the complexity of particular forms of being-with and of engagement with dwelling, but none requires symbolic forecasting. A hawk that intercepts a pigeon coordinates its flight with the pigeon’s through sensorimotor processes of great precision without formulating any proposition about where the pigeon will be in three seconds.
It would be a mistake to tell this history as a story in which early life inhabited a predictable world and humans an unpredictable one. Evolution produced successive changes in the configuration of predictability. Some innovations increased responsiveness to environmental variation, others created new recurrent coordinations, others enlarged the range of possible interactions, and still others generated new capacities to transform environments. Every one of them produced new regularities and new contingencies at once. A nervous system makes behaviour more flexible and also makes possible habits more rigid than any reflex; social life multiplies the responses an animal must reckon with and also creates the stable hierarchies and routines that let it reckon with them.
The distinctively consequential development is the recursively coupled expansion of the two poietic mediations, multimateriality and multisymbolization, in human beings (Ecks 2026b). Humans preserve symbolic designations in material form, talk about absent occurrences, accumulate accounts of past events and establish shared representations of possible futures. A forecast can then survive the person who first made it. It can be compared, taught, rejected, corrected, standardized, institutionalized and built into material coordination. The Babylonian astronomical diaries, kept for some six centuries from at least the mid-seventh century BCE, recorded nightly observations of the sky alongside the weather, the level of the Euphrates and the prices of barley, dates and wool, and the scribes who kept them learned to predict eclipses and planetary phenomena from the cycles their records revealed (Sachs and Hunger 1988). Here the future has become a recursively shareable symbolic object. This is different from the future event becoming available for engagement. What exists is the present symbolic and material coordination around a possible later occurrence, a clay tablet in a temple archive, and its power lies in the extraordinary stability of the deposits that carry it.
The same poietic capacities also expanded enormously the human capacity to transform living worlds. People build cities, found markets, create professions, invent technologies, reorganize landscapes and institute forms of life that did not exist before. Each of these creates new possible futures. A child born today may spend her working life in an occupation, live in a settlement, or belong to an institution that does not yet exist. Such possibilities are not hidden in the present waiting to be forecast; they arise through the historical development of mesocosmic coordination itself.
Two histories must therefore be distinguished. The evolution of prediction is the emergence and elaboration of explicit symbolic practices for identifying later events and conditions. The evolution of future openness is the emergence of living capacities and configurations through which further kinds of transformation become possible. Neither reduces to the other, and the development of increasingly powerful predictive systems has run alongside the creation of increasingly novel possibilities, often through the same institutions. The evolutionary differentiation principle, the eighth proposition and one that requires substantial evolutionary reconstruction rather than direct test, holds that the evolution of biological coordination, the evolution of symbolically articulated prediction and the evolution of future-making capacities are distinct but interconnected histories. Its central consequence can be put plainly. Rather than simply evolving more powerful means of predicting an independently given future, humans evolved means of transforming the mesocosm that continually generate new possible futures, together with symbolic and material practices for attempting to forecast them, so that the evolution of predictive capacity must always be studied together with the evolution of the world-making capacities that alter what there is to predict.
VII. Interrecursive Indeterminacy: When the Future Must Be Produced
The account of sexual partner formation developed in “Die Liebe fällt, wohin sie will” distinguished three senses in which a future can be undetermined (Ecks 2026a). Ontogenetic openness is the fact that a future configuration remains unspecified at an earlier stage of development: at birth, no one can name the person a child will marry, any more than the city in which she will die. Selection underdetermination is the fact that presently available information does not determine which outcome will occur, even when a great deal is known; stated preferences, institutional filters and compatibility algorithms narrow the field without settling it. Both are familiar, and the second is close to what Frank Knight distinguished as uncertainty from measurable risk (Knight 1921). The third sense is stronger. Interrecursive indeterminacy obtains when a future coordination is partly generated through reciprocal recursive engagements that have not yet taken place. It is not another way of saying that forecasting is difficult.
Partner formation is the exemplary case. Two people meet. Each evaluates the other, each becomes aware, to varying degrees, of being evaluated, and each awareness enters the next evaluation. A person can prefer mango to apple without the mango revising its qualities in response; another human being, aware of being considered, does respond, and the response alters the consideration. The relationship that results is generated through this process. Its particular character cannot be read off the participants’ previously established attributes, because the object being predicted does not yet exist independently of the interaction that will produce it.
The distinction between ontological production and incomplete knowledge is the most important in this section. Some future occurrences are hard to predict because their determinants are unknown or measured poorly; better data could, in principle, close the gap. Others are hard to specify because the relevant configuration will be constituted through events and engagements that have not yet occurred. This does not license the claim that such processes are unforecastable in every respect. The probability that two people who meet will form a couple, the population patterns of partnership, and some features of relationships once formed can all be estimated meaningfully. What cannot be assumed is that a complete prior description of two people already contains the particular relationship they will go on to generate.
Openness can nonetheless become stability. A relationship that began as a wholly unspecified possibility can become one of the most enduring coordinations in either person’s life, reorganizing where they live, how their money is spent, which relatives matter, what their days look like and who is present at the end. Marriage is a late and deliberate attempt to stabilize a relationship that has already been generated, and it stabilizes status, obligation and recognition while the interrecursive process continues beneath them (Ecks 2026a). Interrecursive indeterminacy can thus produce some of the most stable futures any human being will inhabit. The openness of the process and the durability of its outcome are not in contradiction, and this is perhaps the clearest illustration of the governing sentence of this article: the stability of the outcome makes much of a married life highly predictable, while the recursive process that produced it set the limits of what anyone could have predicted beforehand.
The partner-choice argument proposed a general hypothesis that can now be placed within a theory of predictability. Where vital stakes are exceptionally high and outcomes remain incompletely determined, people construct elaborate practices for stabilizing the surrounding coordination. Partnership, serious illness, death, political succession and financial markets are the obvious candidates. These practices include predictions, from prognoses to economic forecasts, but also contracts, rituals, conventions, legal procedures, insurance and material preparations. Their importance lies largely in making it possible to live and act without knowing exactly what will happen. Culture gathers, with unusual density, at the places where life will not stay still.
The ninth proposition, the interrecursive indeterminacy principle, holds that some future configurations are generated through recursive encounters rather than merely disclosed through improved information. It is an ontological claim, but it has testable consequences: the recursive-depth experiment proposed for partner formation, which holds a person’s stable attributes constant while manipulating what each party is told about what the other knows, would distinguish it from ordinary updating on new information (Ecks 2026a). Future indeterminacy is accordingly no uniform property of the future but emerges in specific recursive configurations, some of which generate outcomes whose particular form does not exist independently of the process producing them.
VIII. The Emergence of Mathematics and Its Mesocosmic Fit
Modern prediction is overwhelmingly mathematical, and the most ambitious biological theory of prediction describes organisms in the language of probability. Before asking whether organisms calculate, it is worth asking what calculation is. Mathematical practices emerged through historically developed ways of distinguishing, comparing, enumerating, grouping, marking, recording and manipulating quantities. They are not pre-existing mental operations that were later written down. Material practices participated in their emergence from the start.
The history of counting technologies shows this clearly. Fingers, notched bones and sticks, clay tokens, numerical inscriptions, counting boards, abacuses and eventually mechanical and electronic calculators all provide materially stabilized means of organizing numerical operations. In the ancient Near East, small clay tokens used to keep track of goods were sealed in clay envelopes, then impressed on the envelopes’ surfaces, and these impressions developed into the numerical notations from which writing emerged (Schmandt-Besserat 1992). Karenleigh Overmann has reconstructed how numbers themselves emerge and are elaborated through such materials, from fingers and tallies to notations, so that numerical concepts develop with and through the devices that make quantity tangible and manipulable (Overmann 2023), and Lambros Malafouris’s theory of material engagement treats such things as constitutive of thinking rather than as its external aids (Malafouris 2013). These are substantial precedents, and the contribution here is to place them, rather than to rediscover them, within a process ontology in which mathematics is a multimaterially stabilized multisymbolization, made reproducible by the coupling of the two poietic mediations, whose operations can therefore be accumulated, taught, compared and elaborated across generations.
This makes a central distinction possible: the internal validity of a mathematical operation differs from its external fit with a living coordination. A calculation can be perfectly reproducible within an established symbolic system, and whether it captures a particular living process is a separate question. The reliability of arithmetic establishes nothing about the adequacy of its application to embodiment, being-with, dwelling, multimateriality or multisymbolization in any given case. A statistical model may forecast a selected output accurately while leaving most of the process generating that output unexplained. The problem is not that mathematics is too precise to describe life; mathematics models noise, variability, stochasticity and nonlinearity with great sophistication. The problem is that every formalization requires selections, assumptions, definitions and criteria of fit, and that, as argued above, what it captures of a process with mixed recursivity it captures as nonrecursive regularity (Ecks 2026e).
There is a stronger observation still. Mathematics is itself an unusually stable human mesocosm. A calculation performed in Babylon can be checked in Edinburgh because recursively maintained symbolic and material practices preserve particular operations with a reliability that few other human coordinations achieve. Humans then use these exceptionally stable practices to investigate living processes whose own regularities may be of quite different kinds. Mathematical exactness is an achievement of a particular human symbolic and material coordination, and its fit with any other coordination must be established independently. The tenth proposition, the mathematical fit principle, states this as an ontological and methodological claim: mathematical formalization is a multimaterially stabilized multisymbolic achievement whose mesocosmic fit must be established in each application.
IX. Does Life Predict? Friston, Predictive Processing and Biological Inference
Karl Friston’s free-energy principle is the most ambitious attempt to make prediction the foundation of life, and it deserves to be engaged in its strongest form. Its starting point is the same problem with which this article began: how living systems maintain their characteristic organization in the face of environmental and physiological variation. Any system that persists, Friston argues, must keep its states within a bounded range, and this can be described as minimizing a quantity called variational free energy, which places an upper bound on surprise, the improbability of the system’s sensory states given what kind of system it is (Friston 2010). Systems that persist can be partitioned, by a Markov blanket of sensory and active states, into internal and external states, such that the internal states can be read as parameterizing a probability distribution over the external ones; they then appear to infer the causes of their sensations, and to act so as to bring their sensations into line with their predictions, which is active inference (Friston 2013; Kirchhoff et al. 2018). The framework is not a theory of conscious forecasting, and it does not require organisms to manipulate explicit numerical probabilities. In its stronger interpretations, however, the physical dynamics of living systems are understood as implementing inference, so that to live is to predict.
The shared problem, the persistence and self-maintenance of living organization, should be acknowledged. The disagreement lies in how three explanatory claims are related. The first is mathematical description: a system’s dynamics can be accurately represented by probabilistic equations. The second is physical implementation: the system’s actual mechanism realizes the computational or inferential relations those equations characterize. The third is ontological constitution: inference is what fundamentally explains the system’s living organization. The three require different kinds of justification. A mathematical correspondence does not by itself establish an implemented algorithm, and a successful implementation model does not by itself establish that inference is the fundamental ontology of life. Critics of the framework have pressed precisely this point. Jelle Bruineberg and colleagues distinguish the Markov blanket as a formal tool for representing conditional independence in a statistical model from its use as a metaphysical boundary of organisms, and argue that the framework repeatedly slides from the first to the second (Bruineberg et al. 2022). Mel Andrews argues that the principle is best understood as a formal modelling framework, a map rather than a description of the territory (Andrews 2021), and Matteo Colombo and Cory Wright examine the organicist commitments and the limited mechanistic content that its universal scope entails (Colombo and Wright 2021).
The difficulty can be seen in simple cases. A pendulum conforms to mathematical regularities without anyone supposing that it calculates its trajectory. Respiration can be modelled through physiological equations without establishing that the body calculates its next breath. A bacterium’s run-and-tumble movement admits stochastic and even Bayesian modelling without demonstrating that it performs probability calculations. The weak version of this objection, that simple systems cannot implement sophisticated mathematics, should be rejected. Analogue physical systems can realize complex functions without anything like mathematical understanding; the centrifugal governor of a steam engine regulates speed continuously and exactly without computing anything in the sense of manipulating representations, which is why Tim van Gelder used it to ask what cognition might be if not computation (van Gelder 1995). The real question is what makes the inferential interpretation explanatorily and empirically warranted in a given case, over and above the existence of a formal correspondence.
Living Value Theory begins elsewhere, with the actual organization of living coordination. Breathing is metabolic remediation. Walking is recurrent embodied coordination with a relatively stable built and planetary environment. Living processes respond to changing circumstances while maintaining viable organization, and their regularity is not by itself evidence that prediction, explicit or implicit, is their constitutive activity. The free-energy formalism may capture important features of these processes, and on the mixed-recursivity hypothesis it will capture precisely their nonrecursive regularities: the bounded ranges, the recurrent patterns, the statistical structure of their coupling to an environment. Its explanatory status for the recursive processes that produce those regularities has to be established rather than assumed.
A further question concerns optimization. The free-energy principle explains living organization through the minimization of a single quantity. Living Value Theory investigates heterogeneous vital stakes across irreducible mediations, whose relative weight changes with circumstances, so that a living being may face several compatible forms of viable coordination without any uniquely optimal one. This does not refute optimization frameworks, which can represent multiple constraints and tolerances, but asks whether the quantity they select explains the full living coordination or only particular aspects of it, and whether the irreducibility of the mediations can survive translation into one currency.
The constructive conclusion is a set of criteria for distinguishing substantive biological inference from mathematical redescription. A claim of biological inference should identify the mechanisms proposed to implement it; specify what evidence would count against that implementation; demonstrate explanatory or predictive gains over alternative physiological and dynamical accounts; and refrain from treating the formal properties of a representation as automatically establishing corresponding properties of the organism. Some predictive processing research, particularly in specific perceptual and motor systems, may meet these criteria. The universal principle, as formulated, does not yet do so. The eleventh proposition, the ontological non-equivalence principle, states the conclusion: the successful mathematical modelling of a living process as inference does not, without further justification, establish inference as its constitutive biological mechanism. The capacity to model continuous remediation as inference does not show that continuous remediation consists of inference.
X. Implications for Social Theory and Research on Prediction
Social theories of the future often oppose prediction, as a conservative technique for closing the future and extending the present, to anticipation, imagination or speculation, as the creative opening of other possible futures (Adams, Murphy and Clarke 2009). Elena Esposito’s analysis of how financial and algorithmic forecasts shape the futures they forecast (Esposito 2011), and Jens Beckert’s account of the fictional expectations through which actors coordinate under uncertainty (Beckert 2016), are among the most sophisticated versions of the wider argument that futures are made rather than found. The account developed here agrees that some futures are produced, and specifies where and why. It rejects the opposition between conservative prediction and creative openness. Prediction does not impose stability on an otherwise open future, because it ordinarily presupposes an existing stability without which it could not be formulated. Political transformation, in turn, depends on extensive continuities of communication, institutional coordination, infrastructure, bodily capacity and shared intelligibility, without which no alternative future could be organized. The question is never whether prediction is intrinsically stabilizing or politically conservative, but how particular predictive practices enter living coordination and change its possibilities.
This suggests a research agenda for any forecasting practice, framed as questions rather than as a procedure for generating forecasts. What exactly is being predicted, over what interval and by what criterion? Which mesocosmic continuities are taken for granted? Which mediations and counterpart types are involved in producing the outcome, and how stable are the counterparts and their ongoing remediations? Does the forecast itself alter the coordination it describes? Does the outcome depend on a configuration that will only be generated through recursive engagements yet to occur? And what does predictive accuracy establish, and what does it leave unexplained about the generative process? Asked of a weather forecast, a credit score, a sales projection, an epidemiological model or a dating algorithm, these questions yield very different answers, which is precisely their purpose. They allow forecasting practices to be compared without assuming a homogeneous object, making the background stability of living coordination analytically visible while permitting precise investigation of particular forms of indeterminacy.
XI. Conclusion: Prediction Is a Derivative Achievement of Living Worlds
The person walking downstairs is not solving a general problem of predicting the future. She is continuing an established living coordination, in which the staircase, her body, her practised movements and the planet’s gravity are regular enough for it to go on. Explicit prediction emerges only when selected later conditions become objects of symbolic articulation, and its possibility depends on the vast continuities of living coordination it takes for granted. Evolution differentiated those continuities, introduced new forms of responsiveness and interrecursivity, and eventually produced the human poietic capacities through which possible futures became recursively shareable. The same capacities enable the generation of possibilities that did not exist before, which is the twelfth proposition, the human futurity paradox: human poietic mediations increase both the capacity to symbolize possible futures and the capacity to generate novel coordinations that alter what can be predicted. The deepest puzzle is therefore not how living beings predict an uncertain future, but how extraordinarily stable living worlds continuously reproduce themselves while generating the possibilities of their own transformation. Prediction is one historically specialized activity within that larger process.
Selected References
Adams, V., Murphy, M. and Clarke, A.E. 2009. ‘Anticipation: Technoscience, Life, Affect, Temporality.’ Subjectivity 28(1): 246–265.
Andrews, M. 2021. ‘The Math Is Not the Territory: Navigating the Free Energy Principle.’ Biology & Philosophy 36(3): 30.
Beckert, J. 2016. Imagined Futures: Fictional Expectations and Capitalist Dynamics. Cambridge, MA: Harvard University Press.
Bruineberg, J., Do??ga, K., Dewhurst, J. and Baltieri, M. 2022. ‘The Emperor’s New Markov Blankets.’ Behavioral and Brain Sciences 45: e183.
Colombo, M. and Wright, C. 2021. ‘First Principles in the Life Sciences: The Free-Energy Principle, Organicism, and Mechanism.’ Synthese 198(Suppl. 14): 3463–3488.
Esposito, E. 2011. The Future of Futures: The Time of Money in Financing and Society. Cheltenham: Edward Elgar.
Friston, K. 2010. ‘The Free-Energy Principle: A Unified Brain Theory?’ Nature Reviews Neuroscience 11(2): 127–138.
Friston, K. 2013. ‘Life as We Know It.’ Journal of the Royal Society Interface 10(86): 20130475.
Kirchhoff, M., Parr, T., Palacios, E., Friston, K. and Kiverstein, J. 2018. ‘The Markov Blankets of Life: Autonomy, Active Inference and the Free Energy Principle.’ Journal of the Royal Society Interface 15(138): 20170792.
Knight, F.H. 1921. Risk, Uncertainty and Profit. Boston: Houghton Mifflin.
MacKenzie, D. 2006. An Engine, Not a Camera: How Financial Models Shape Markets. Cambridge, MA: MIT Press.
Malafouris, L. 2013. How Things Shape the Mind: A Theory of Material Engagement. Cambridge, MA: MIT Press.
Merton, R.K. 1948. ‘The Self-Fulfilling Prophecy.’ Antioch Review 8(2): 193–210.
Overmann, K.A. 2023. The Materiality of Numbers: Emergence and Elaboration from Prehistory to Present. Cambridge: Cambridge University Press.
Rosen, R. 1985. Anticipatory Systems: Philosophical, Mathematical and Methodological Foundations. Oxford: Pergamon Press.
Sachs, A.J. and Hunger, H. 1988. Astronomical Diaries and Related Texts from Babylonia, Volume I: Diaries from 652 B.C. to 262 B.C. Vienna: Österreichische Akademie der Wissenschaften.
Schmandt-Besserat, D. 1992. Before Writing, Volume I: From Counting to Cuneiform. Austin: University of Texas Press.
van Gelder, T. 1995. ‘What Might Cognition Be, If Not Computation?’ Journal of Philosophy 92(7): 345–381.