Some years ago, when my children were around five and seven, they asked for a pet. We were travelling a great deal, so a dog or even a cat seemed impractical. We agreed on two hamsters. I cannot now remember what they were called, which already tells you something important: for me, these hamsters meant very little. The two hamsters turned out to despise each other. One bullied the other relentlessly and was consumed by a hunger for freedom. After a few weeks he managed to escape and was never seen again. The other stayed in his cage quite contentedly for another two years, doing his hamster thing, running, sleeping, eating, ignoring us. He was not cuddly. He actively disliked being held. He was not interested in the children, and the children were not especially interested in him. One day he died of natural causes. The children gave him a decent burial in the garden. Life went on. He was forgotten.
This small domestic episode demonstrates something precise about recursive mediational attribution. The children attributed the hamsters full inter-recursivity from the start. They wanted the hamsters to respond to them, interact with them, play with them, enjoy being held, recognize them as the specific humans they were. This is exactly what children do: they extend the attribution regime that works with dogs, cats, and each other outward to any animal that comes within reach. The attribution was made. It was made with feeling and persistence. The problem was not that it was not made. The problem was that it could not be sustained, because the hamster kept demonstrating, at L1, that it was not there. No amount of symbolic work, naming, handling, speaking to, attributing personality, expecting recognition, could override what the hamster's own behaviour kept returning as feedback: I am somewhere. I am alive. I have my own preferences. I will not be your social partner.
This is the principle of mesocosmic fidelity. An attribution regime must have fidelity with the actual mediational profile of the attributed entity, or the L2 feedback will be relentless and will eventually wear the attribution down. The hamster had embodiment: it was a living body with real metabolic needs, real sensations, real discomfort when held. It had dwelling: it was somewhere, subject to the climate of the cage, the rhythms of its tiny ecological situation. It had self-recursivity: the escaped hamster's single-minded drive for freedom was not nothing. But it did not have inter-recursivity with humans in any form that the children's attribution could engage. It did not loop back. Every time a child picked it up, the hamster's struggling, biting, freezing, fleeing delivered a clear L2 signal that the wished-for profile had no anchor in the animal's actual mediational reality. The attribution was possible. It was not mesocosmically faithful. And without fidelity, it slowly collapsed.
The contrast with the family dog is exact. Dog attribution stabilizes and deepens precisely because dogs genuinely possess inter-recursive capacity. They respond to human responses. They attune, anticipate, adjust, grieve, play, and recognize specific individuals across time. The attribution of being-with, inter-recursivity, and eventually multisymbolization to dogs is not merely a projection. It finds genuine purchase in the animal's mediational profile. Attribution and reality are in fidelity. This is why the grief when a dog dies is real and sustained, while the hamster's burial was decent but brief. It is also why Descola's Achuar peccaries are so theoretically interesting: the attribution of full mediational architecture to them is not tested by the kind of direct inter-recursive encounter that would quickly correct a misattribution. The question of whether peccary attribution has mesocosmic fidelity is answered not in face-to-face interaction but in the long ecology of the hunt, the forest, the collective practice, and the slow feedback of a world that either sustains or withdraws its cooperation. This is the question at the heart of the argument that follows: which beings, in which mesocosms, are attributed which mediations and which levels of recursivity, and does that attribution have fidelity with what those beings actually bring to the coordination?
Philippe Descola is the most important anthropologist of the last generation to have taken seriously the question of what holds a world together. His central claim, developed across the lectures collected in Politics of Worlding and given their fullest treatment in Beyond Nature and Culture, is that what modern Western thought calls "nature" and "culture" is only one historically specific way of arranging relations between humans and nonhumans. Across fifty years of ethnographic research, comparative analysis, and theoretical elaboration, Descola has argued that different human collectives inhabit genuinely different worlds, not different perspectives on the same world, but different configurations of beings, relations, and identities that follow from basic cognitive inferences about what kinds of things exist and how they connect. This is what he means by "worlding": the stabilization of a world through ontological predication.
The present article takes Descola's work seriously enough to press hard on its foundations. It does so through the conceptual apparatus of Living Value Theory (LVT), a process ontology developed to describe the architecture of human coordination across evolutionary, historical, and comparative scales. The argument is not that Descola is wrong to criticize the nature/culture distinction, nor that his ethnographic material is unreliable. On the contrary: his ethnographic material is richer than his theory deserves. The argument is that Descola mistakes the phenomenon he describes. He thinks he is documenting different ontologies, fundamentally different worlds. What he is actually documenting is something at once more modest and more significant: different regimes of recursive mediational attribution. Once that substitution is made, the Descola corpus becomes enormously more useful, and the path beyond the ontology turn becomes clear.
I. Descola's Project and Its Structuralist Residue
In his introductory essay to Politics of Worlding, William Hanks describes Descola's fourfold schema with admirable precision. The four ontological types, naturalism, animism, totemism, and analogism, are generated by crossing two axes. The first concerns interiority: whether a collective treats humans and nonhumans as sharing or not sharing the same inner life, intentionality, selfhood, and capacity for moral experience. The second concerns physicality: whether humans and nonhumans are treated as sharing or not sharing the same bodily constitution, material processes, and organic laws. Each combination yields one of the four types. Naturalism (modern Western science and industrial modernity) assumes shared physicality but distinct interiority: humans and animals obey the same biology, but only humans have genuine selfhood, culture, and symbolic life. Animism assumes shared interiority but distinct physicality: a jaguar, a peccary, a manioc plant, and a human all have equivalent inner lives, but they inhabit different bodies that afford them different access to the world. Totemism assumes continuity on both axes, fusing humans and nonhumans into shared collectives of common substance and essence. Analogism assumes discontinuity on both axes, multiplying differences until only elaborate systems of correspondence can hold a world together.
Descola is careful to insist that these are heuristic devices, not taxonomic boxes. He writes that the fourfold typology "should thus be taken as a heuristic device rather than as a method for classifying societies." He acknowledges that ontological schemas hybridize in practice, that individuals can reason in different modes in different situations, and that the four types represent tendencies rather than rigid boundaries. Hanks reinforces this when he notes that "the ontological schemas can overlap and often hybridize in practice" and that they are "variably in play in situated judgments." This flexibility has allowed Descola to escape the most obvious objections to structuralist typologies: he is not claiming that the Achuar always and only reason as animists, or that modern naturalists never attribute inner life to their pets. Descola indeed gives the example of naturalists who occasionally "talk to their cat, their dog, or their car as if they could thus establish some sort of intersubjective relation with it," reasoning briefly as animists without thereby becoming animists.
And yet the structuralist residue remains, and it lies at a level that Descola's flexibility does not reach. The residue lies not in the empirical rigidity of the classification but in the deeper level at which the grid is generated. Descola's four types are produced by a logical operation: take two binary distinctions, cross them, generate four cells. The operation is elegant and has real analytical purchase. But it has no demonstrated mesocosmic genesis. Why should "interiority" and "physicality" be the two fundamental axes along which all human relations to nonhumans are organized? Why should continuity and discontinuity be the only relevant values? Descola derives these dimensions partly from cognitive psychology, partly from phenomenological philosophy, and partly from the anthropological literature on personhood. He tells us that the awareness of a duality between "material processes" and "mental states" is "one of the universal features of the human mind." But this derivation already imports the very inside/outside, body/mind metaphysics that the ontological turn was supposed to overcome. The grid is not derived from the living world. It is derived from an abstract logical space. Descola has moved beyond classical structuralism in his treatment of empirical variation, the schemas are flexible, hybridizable, situationally activated, but he has not moved beyond structuralism in the construction of his analytic object. The structuralist residue lies not in rigid classification of societies but in the prior selection of binary axes as the generative machinery of comparison.
The one-line critique is this: Descola pluralizes the boxes but leaves box-making intact. The stronger version is: Descola's boxes are produced by a logical grid; LVT's mediations are reconstructed from evolutionary thresholds. Both must be stated in language, but they do not have the same status. This distinction is the key to everything that follows. It also explains why Descola's framework, despite its flexibility, cannot easily handle the historical dynamics of attribution change within a single mesocosm, like the transformation in how dogs are treated in contemporary urban Scotland, or the rapid negotiation of attribution currently underway around artificial intelligence. His framework was designed to compare broad ontological types across human collectives. It was not designed to track the continuous, contested, historically dynamic negotiation of attribution that characterises every living mesocosm from inside.
II. The Process Ontology of LVT
Living Value Theory begins not with a typology but with the question of what makes human life possible at all. Its answer is evolutionary and reconstructive: the human mesocosm, the field of living coordination within which human life takes its distinctive form, emerged through a series of thresholds at which existing forms of coordination became insufficient under metabolic, reproductive, and social pressure. The argument is developed in full in the article "The Genesis of the Five Mediations: A Deep Evolutionary Account of the Human Mesocosm," but its core logic can be stated here.
The first threshold produces the most basic structure of all living beings: embodiment and dwelling co-emerge because any living being is bounded and somewhere. A cell maintains an inside and an outside and can persist only through a sustaining relation to its milieu. There is no organism first and then an environment; life begins as body-in-place. This is not a metaphysical claim about the priority of organism over environment. It is a description of the minimal conditions of any living process: metabolic maintenance of a bounded system in a sustaining relation to its surroundings. The second threshold is being-with: wherever survival depends on coordinated plurality, the fate of one body becomes bound up with the movements and responses of others. Schools of fish, flocking birds, and coordinated predators all demonstrate that being-with is not primarily a feature of mammalian intimacy but of any form of life where collective presence multiplies protection, vigilance, and access to resources. Being-with first arises not through romance or attachment but through shared exposure to danger and opportunity.
The decisive threshold, the one that marks the emergence of distinctively human coordination, is the joint emergence of multimateriality and multisymbolization. These two mediations did not appear sequentially. They co-emerged because neither could solve the coordination problem that was pressing on early Homo sapiens without the other. A tool is not a shaped object. It is matter shaped in light of a shared purpose, using techniques that must be transmitted, recognized, and coordinated across persons and time. It requires symbolic capacity: the ability to name, classify, teach, warn, and plan in relation to something not yet present. Equally, symbolic coordination without material anchors would have nothing durable to organize: symbols gain their traction precisely by referring to persistent things, tools, paths, hearths, stored food, marked sites, bodies. Multimateriality and multisymbolization are two faces of a single capacity: the externalization of coordination into persistent forms that carry meaning across time. This is why the periods in the archaeological record in which tool traditions become more complex are also the periods in which symbolic behaviour becomes more evident. The same coordination pressures that drove elaboration in one mediation drove elaboration in the other.
The five mediations, embodiment, being-with, dwelling, multimateriality, multisymbolization, are therefore not a typology. They are the sedimented result of evolutionary thresholds that life had to cross under pressure. Each mediation is irreducible: remove embodiment and there is no living being; remove dwelling and there is no place in which to coordinate; remove being-with and there is no collective; remove multimateriality or multisymbolization and the distinctively human form of life becomes impossible. The architecture is not five boxes into which phenomena can be sorted. It is five domains of coordination whose interdependence constitutes the human mesocosm. This also means that no subsequent pressure has been sufficient to force the emergence of a sixth mediation or to render any of the five dispensable. Digital infrastructures, global markets, and artificial intelligence are all elaborations and redistributions within the existing architecture, intensifications of multimateriality and multisymbolization in particular, but they do not add a new domain of coordination.
This is the answer to the symmetry objection that will immediately be pressed against LVT. If Descola's "interiority" and "physicality" are an anthropologist's L4 grid, what makes "embodiment" and "multimateriality" anything more than a different anthropologist's L4 grid? The answer is that LVT's categories are not logically generated. They are evolutionarily reconstructed. Descola's interiority/physicality distinction is derived by a thought experiment from cognitive psychology. LVT's five mediations are derived from the history of living coordination under metabolic pressure. The concept is symbolic; the mediation named by the concept is not reducible to symbolization. Embodiment names a condition that is metabolically enforced regardless of how any theory describes it. You cannot argue your way out of needing to eat, be somewhere, or coordinate with others. The mediations are not categories of thought. They are conditions of survival that have been forcing living beings to elaborate new coordination architectures since the beginning of life.
III. Recursivity Levels and the Epistemological Position of LVT
LVT also distinguishes five levels of recursivity that describe how coordination becomes symbolically elaborated. L1 is seamless, pre-thematic coordination: the skilled hunter moving through the forest, the experienced speaker engaged in conversation, the infant nursing. Nothing is thematized; everything flows. L2 is felt salience or misalignment: when coordination is disturbed, something is felt before it is named, a sense that something is wrong, that a situation requires attention, that an expectation has been violated. L3 is local symbolic articulation: partial naming, gesture, practical description that helps restore or adjust coordination. L4 is symbolic stabilization: categories become explicit, transmissible, governable, this is the level at which anthropological concepts, legal categories, religious doctrines, and scientific theories operate. L5 is reflexive critique of L4 systems: the level at which one asks whether the symbolic stabilizations in use are adequate, distorting, or missing what matters. The present article operates at L5.
This framework carries a decisive epistemological implication that must be stated clearly and held consistently throughout. Everything that anthropologists write, and everything that LVT itself states, operates at L3, L4, and L5. There is no direct access to L1 and L2. L1 and L2 are pre-symbolic or non-symbolic: they cannot be written down without immediately becoming L3 or higher. When an Achuar person describes animal societies with leaders, shamans, tools, and marriage rules, that description is already L3/L4 work, elicited in the context of an anthropological encounter, formulated in response to the particular questions and attentions of the fieldworker, and articulated in the symbolic register of language. When Descola writes it up and places it within his ontological framework, that is L4/L5 work. When LVT re-reads both, that is L5 work reflecting on L3/L4 configurations. Nobody in this chain has unmediated access to what peccaries, jaguars, or sweet potatoes are actually doing at L1/L2.
This means that LVT does not claim to verify what animals are really like at L1/L2. The question of whether peccaries objectively possess something like multisymbolization in some observer-independent sense is not answerable and LVT does not attempt to answer it. What LVT asks instead is: how does a given mesocosm perform recursive mediational attribution at L3/L4/L5, and does this attribution cohere with how people actually act in the world? This distinguishes LVT from both naive empiricism (which thinks it can read off from behaviour what an entity really is) and from Descola's ontological realism (which thinks it can identify which ontological schema a collective inhabits). LVT insists that all such determination is itself symbolic work at L3/L4/L5, and that the relevant question is not which map is metaphysically correct but how attribution regimes function within living coordination, whether they produce L1 flow or L2 disturbance, and what happens when they become incoherent.
LVT also distinguishes three domains of recursivity. A non-recursive process does not respond to being classified: a rock does not reorganize itself when named igneous; a chemical reaction does not care about the formula used to describe it. This is what "nature" in the modern scientific sense names: the domain of regularities that can be isolated because they do not respond to classification. A self-recursive being modifies its own future through prior states, learning, memory, and anticipation. Many animals are clearly self-recursive in this sense. An inter-recursive relation exists when two or more beings respond to each other's responses: I anticipate what you think I will do; you respond to my response to your expectation; I adjust in light of your adjustment. Human social life is saturated with inter-recursivity, and the joint emergence of multimateriality and multisymbolization is what allows inter-recursive coordination to be stabilized across time, distance, absence, role, institution, and generation, across the situations in which the parties to a relation are not co-present. This stabilization across absence is what distinguishes human-scale inter-recursivity from the social coordination observable in many other animals.
IV. The Ethnographic Material: What Descola Actually Observes
With this framework in place, we can return to Descola's ethnographic material and read it with fresh eyes. The passage in Politics of Worlding that most clearly exceeds Descola's own conceptual vocabulary is his account of what he calls "tribe-species" in animist ontologies. In animist worlds, he writes, "tribe-species, especially animal ones, are said to live in collectives with identical structures and properties: they are full-scale 'societies' with leaders, shamans, rituals, dwellings, techniques, artefacts, which assemble and quarrel, provide for themselves, and marry according to rules." Each tribe-species, whether Achuar humans, sweet potatoes, or peccaries, occupies its own distinctive collective, "a social species characterized by a morphology, capacities and types of behavior, a kind of association that combines the attributes of a natural species and that of a tribe." What the moderns call "nature," "supernature," and "society" is, for animists, "a world populated by social collectives, each of them different in membership, with which the human collectives establish relations according to norms supposedly common to all."
Descola's framework interprets this as evidence that animism attributes shared interiority to humans and nonhumans, the animist inference that other-than-humans have "an interior life as rich and varied as that of a human, including the capacity for symbolization, moral judgment, and full-blown selfhood." This interpretation is thinner than the evidence requires. Look at what the ethnographic descriptions actually show. Animal collectives are attributed leaders: being-with organized through hierarchy and authority, implying recognized inter-recursive social ordering. They are attributed shamans: symbolic specialists who mediate between collectives across permeable ontological membranes, which is multisymbolization at its most elaborate, the management of symbolic communication across what would otherwise be unbridgeable categorical divides. They are attributed rituals: symbolic enactment grounded in material performance, which requires the coupling of multimateriality and multisymbolization that LVT identifies as the distinctively human threshold. They are attributed dwelling in the full LVT sense: recognized ecological situatedness, territories, habitats, and environmental conditions that are theirs, that afford their particular form of life, and within which they are known to be at home. This is not mere location. It is situated belonging within a landscape understood as theirs to inhabit. They are attributed techniques and artefacts: persistent material tools shaped by shared purpose, which is the definition of multimateriality. They are attributed collective quarrels and provisioning practices: sustained social coordination across time with recognized obligations and rights, which is the definition of inter-recursivity in its socially organized form. And they are attributed marriage rules: normative symbolic organization of reproduction, which is multisymbolization operating at the most fundamental level of collective continuity.
This is not "shared interiority." Interiority is a thin, philosophically loaded, inside/outside category that tells us little about what is being attributed. What the evidence shows is the attribution of the full five-mediation architecture, all five domains of coordination that LVT identifies as constituting the distinctively human mesocosm, to selected nonhuman beings. The Achuar are not merely saying that peccaries have souls. They are saying that peccaries have recognized dwelling (their own ecological territory, landscape, and environmental conditions understood as belonging to their collective), being-with organized as full social life, material techniques, artefacts, ritual, symbolic communication, and normative rules governing reproduction and exchange. These are not aspects of interiority. They are aspects of what it means to coordinate a life within a mesocosm.
The same observation applies with equal force to Descola's account of the Reungao tiger-alliance, which he presents as an illustration of the gift or sharing pattern in animist sociology. The missionary ethnographer Kemlin documented a woman named Oih who, through a chance encounter in which she accidentally helped a choking tiger, enters into a sustained contractual relationship with the animal: the tiger appears to her in dream in human form, proposes an alliance, and then enacts it through a regular practice of leaving prey for her at the forest edge. Descola reads this as evidence of the gift pattern, a transfer of value without expectation of return, expressing the animist attribution of shared interiority. But the LVT reading sees something richer: the tiger is attributed not only interiority but inter-recursive social agency, the capacity to recognize, initiate, commit to, and fulfil a contractual relationship. The tiger proposes an alliance in recognizably social terms ("We are going to enter into a friendship between father and daughter"), enacts it through sustained reciprocal behaviour, and is understood by Oih and her community to be a reliable party to a krao con ba contract with specific obligations. This is not interiority. This is inter-recursivity stabilized through symbolic and material exchange across species.
The key finding, stated precisely, is this: in Descola's own ethnographic presentation, selected nonhuman beings are attributed not merely souls or perspectives but participation in the full mediational architecture that LVT identifies as constituting distinctively human coordination. The evidence exceeds Descola's conceptual apparatus. His framework can only say "shared interiority, distinct physicality." What the evidence shows is the attribution of being-with, dwelling, multimateriality, multisymbolization, and inter-recursive social agency to selected animals and spirits. This is an immanent critique: we are not claiming independent ethnographic access. We are saying that Descola's own ethnographic presentation supports conclusions his conceptual vocabulary cannot formulate.
V. Recursive Mediational Attribution: The New Concept
We can now state the central conceptual contribution. The phenomenon Descola describes is not differences between ontologies. It is differences in recursive mediational attribution. LVT treats the architecture of the human mesocosm, five mediations, three recursivity domains, five recursivity levels, as universal at the level of the human mesocosm itself, not because it is a cultural universal in the ordinary anthropological sense, but because each mediation names an irreducible condition of human coordination. This is not the old universalism that projected one cultural configuration onto all others. It is the claim that the conditions under which human life became possible, the evolutionary thresholds that were metabolically enforced, are not culturally negotiable. What is culturally variable, empirically, historically, within mesocosms and across them, is which entities are attributed which mediations, which recursivity domains, and which recursivity levels. The process ontology is fixed; the attribution is in motion. This is the question that ethnography, mesocosmography, and comparative anthropology are actually studying, whether they know it or not.
The question is never: which ontology does this people inhabit? The question is: in this mesocosm, who is attributed what? Who is said to have which recursivity level, L1 seamless flow, L2 felt salience, L3 articulation, L4 symbolic stabilization, L5 reflexive critique? Who is attributed non-recursivity, self-recursivity, or inter-recursivity? Who is attributed embodiment, being-with, dwelling, multimateriality, multisymbolization? And crucially: does this attribution cohere across contexts, producing L1 flow, or does it generate L2 misalignment that forces L3 revision?
One clarification is essential before proceeding. Attribution does not mean unreality, projection, or belief held in the mind. To attribute a recursive mediational profile to a being is to reorganize the field of coordination around that being. It changes what can be done, what must be avoided, what counts as harm, what requires repair, and what kinds of feedback matter. The Achuar tiger contract is not a belief that overlays a separate material reality. It is a real social-mesocosmic operation: it restructures who must be propitiated, what exchange obligations arise from predation, what a shaman is for, and how hunting must be conducted. The dog's household membership is not a sentimental projection. It restructures inheritance, veterinary expenditure, grief, daily routine, and legal accountability for harm. Attribution is one of the ways human mesocosms make relations real. The criterion for evaluating attribution regimes is therefore not whether they are metaphysically accurate in some ultimate sense, whether the tiger really did negotiate in some observer-independent way, but whether they are mesocosmically coherent: whether the attribution is enacted consistently across contexts, generates L1 flow rather than L2 misalignment, and produces coordination problems that are managed through available symbolic work rather than suppressed through buffering until they assert themselves through slower and more diffuse channels.
This framework applies with equal analytical precision to cases that Descola's ontological grid cannot handle without strain. Consider the shift in how dogs are treated in contemporary Scottish urban life compared to a generation ago. Dogs are now routinely attributed being-with (full family membership, emotional bonds, grief responses, recognized social membership with attendant rights and obligations), multimateriality (specialist veterinary medicine, orthopedic beds, GPS collars, tailored nutrition regimes, documented medical histories, an entire industry of material scaffolding organized around their bodies), multisymbolization (legal personhood in animal welfare legislation, named individuals with documented personalities, memorial services, social media presences, literary and cinematic protagonism), and high levels of inter-recursivity (sustained emotional attunement, communicative responsiveness, mutual adjustment recognized as genuinely social). Dogs have always had dwelling, they are somewhere, subject to climate, terrain, and ecological conditions, but what is new and striking is the upward extension of the remaining mediations, especially being-with at the level of full social membership, and multisymbolization at the level of legal and narrative personhood. This is a historically documented shift in the recursive mediational attribution profile of one species within one mesocosm. It has no plausible account within Descola's framework, he would have to say that naturalists are increasingly behaving as animists, which is both dismissive and analytically unhelpful. Within the LVT framework, it is a clear empirical phenomenon: an attribution regime in historical motion, driven by changing material conditions (urbanization, nuclear family structure, companion animal industries), legal elaboration, and the shortening of feedback loops between humans and specific nonhuman beings under conditions of intensive co-habitation.
The recursive mediational attribution framework also handles the generative AI case. LVT's analysis of systems like large language models is that they operate across an enormous symbolic field, they are extraordinary machines of multisymbolization, but have no embodiment, no dwelling, no being-with in real time, no multimateriality in the sense of acting on the physical world through persistent material transformation. The empirical question, and it is empirical, is whether and to what degree human users attribute more than this: whether they attribute feeling, presence, genuine understanding, social reciprocity, or inter-recursive engagement to these systems. The evidence is already substantial that many users make precisely such attributions, often stably and with practical consequences for their behaviour. This is a new and rapidly developing attribution negotiation, and it follows exactly the same structure as the Achuar attribution of social agency to peccaries or the Victorian attribution of non-recursivity to the "lower races." The structure is universal. The content varies.
VI. Mesocosmic Recursive Mediational Coherence
A new criterion is required to evaluate recursive mediational attribution regimes without falling into either ontological realism or relativism. LVT proposes mesocosmic recursive mediational coherence. The criterion is emphatically not truth in any universal or metaphysical sense. It is not a question of whether the Achuar attribution of inter-recursivity to peccaries is accurate at the level of some observer-independent reality, or whether the Reungao tiger actually negotiated a contract in the way a human would. Those questions are unanswerable at L3/L4/L5 and LVT does not attempt to answer them. The criterion is mesocosmic: it operates entirely within the living field of coordination and asks whether an attribution regime holds together there. An attribution regime is mesocosmically coherent when the recursive mediational profile attributed to an entity is enacted consistently across the contexts in which that entity appears within the mesocosm. This consistency is not logical consistency between propositions. It is the consistency between attribution and practice that produces L1 flow, seamless, untroubled coordination, rather than L2 felt misalignment that demands either symbolic management or suppressive buffering.
The criterion does not require, and explicitly refuses, ontological accuracy as a standard. We cannot adjudicate what peccaries or tigers or AI systems really are at L1/L2, and LVT does not attempt to do so. The criterion is internal to the mesocosm and to the attribution regime itself. It asks: does your behaviour across contexts remain consistent with the recursive mediational profile you have attributed to this entity, and does that consistency hold in embodied practice, at L1, or does it generate L2 misalignment that forces L3/L4 revision? Consider what happens when someone attributes non-recursivity to an entity that then behaves in ways that presuppose self-recursivity or inter-recursivity. If you attribute non-recursivity to an animal and then it screams when you cut it, you receive immediate L2 feedback: the attribution was inadequate, the behaviour was incoherent, something must change. The feedback is automatic, pre-symbolic, and impossible to ignore in direct embodied encounter. You will revise your attribution, develop symbolic work to manage the tension, perform a ritual acknowledgment, or suffer sustained L2 misalignment. This is mesocosmic correction from below. The question of what the animal's pain really means at some ultimate level is set aside. The question of what your L2 response requires you to do next is not.
The Jain and vegan cases illuminate this with particular sharpness. If one attributes self-recursivity, the capacity to suffer, to plants as well as animals, then every meal becomes a coherence problem: the metabolism of any living being requires consuming other living beings, including those one has attributed the capacity to suffer. The coherence solution is not to stop eating (which is impossible) but to develop L3/L4 symbolic work that acknowledges the metabolic entanglement and manages its implications. Jain practice does exactly this through graded regimes of non-violence, careful distinctions between more and less sentient organisms, ritual acknowledgment of unavoidable harm, and an explicit acceptance that metabolic life is constitutively violent at some level. This is not superstition. It is an elaborate and internally consistent coherence maintenance system for a high-attribution regime under conditions of unavoidable metabolic necessity. The attribution is generous; the coherence work is correspondingly elaborate.
VII. L2 Buffering: The Suppression of Mesocosmic Feedback
The most important new concept to emerge from this analysis is L2 buffering: the systematic interposition of material, symbolic, spatial, or temporal distance between an attribution regime and the mesocosmic feedback that would otherwise correct it. When felt misalignments in living coordination are muted, shielded, suppressed, or made invisible, when the L2 signals that would force L3/L4 revision are prevented from reaching the people whose attribution regime is generating them, incoherence can be sustained at scale for extended periods. It will not be sustained indefinitely, because the mesocosm is not purely symbolic. Its L1 ground is metabolically enforced. But it can be sustained long enough to produce consequences that are themselves slow-feedback, diffuse, and difficult to attribute to the original incoherence.
Industrial animal farming is the paradigm case of civilisational-scale L2 buffering. Urban mesocosm members in high-income societies typically attribute high recursive mediational profiles to companion animals, dogs and cats receive being-with (full social membership, grief, recognized relational bonds), multimateriality (specialist food, medicine, equipment), multisymbolization (names, legal status, narratives, memorials), and high-level inter-recursivity, as argued above. All animals have dwelling in the LVT sense, they are situated beings subject to landscape, climate, and ecological conditions, so dwelling is not what distinguishes the companion animal attribution regime from the farmed animal one. The same people consume industrially farmed chickens, pigs, and cattle, animals whose ethologically documented recursive profiles are broadly comparable to those of dogs. The attribution gap is enormous, and it should in principle generate sustained L2 misalignment: the same person who grieves deeply for a dead dog and would refuse to harm a dog they knew, participates daily in a system that treats pigs, who are cognitively comparable to dogs, as non-recursive raw material.
The reason this does not, for most people, most of the time, generate the L2 misalignment it structurally should, is that the feedback is systematically buffered. Spatial buffering: slaughterhouses are not in cities; the killing is geographically removed from the consumer. Temporal buffering: the consequences of industrial animal farming, antibiotic resistance, zoonotic disease spillover, ecological degradation, soil depletion, emerge on timescales of decades, making the causal connection between individual consumption choices and systemic consequences effectively invisible to immediate experience. Material buffering: packaging converts animal into food-object, removing the perceptual cues, blood, fur, eyes, screaming, that would trigger attribution of self-recursivity or inter-recursivity. Symbolic buffering: language performs the conversion before the material object is encountered, pork, not pig; beef, not cow; poultry, not chicken, such that the naming itself enacts a reclassification that the material object then confirms. Institutional buffering: regulatory and legal frameworks normalise the practice, distribute responsibility so diffusely across producers, retailers, consumers, and legislators that no individual confronts the L2 pressure directly.
From a process ontological standpoint, this buffered incoherence cannot be maintained indefinitely. The L2 signals will assert themselves through whatever channels remain open. Antibiotic resistance is one such channel: the systematic treatment of farmed animals with antibiotics that the industrial system requires to maintain animal health under conditions of high-density confinement generates resistance that eventually reaches human medicine. This is not experienced as a direct consequence of the attribution incoherence, but it is a slow-feedback L2 signal that the regime is not coherent with living coordination at the scale at which it is operating. Pandemic disease spillover from industrial animal facilities is another. The ecological costs, water table contamination, greenhouse gas accumulation, soil degradation, are others still. These slow-feedback signals are themselves buffered by further symbolic and technical work, regulatory responses, technological optimism, carbon offsetting, but they do not disappear. They accumulate.
L2 buffering therefore gives LVT a precise and non-moralistic analysis of industrial modernity's relationship to nonhuman beings. The argument is not that factory farming is wrong because animals have souls or because suffering is intrinsically bad in some philosophical register. The argument is that industrial animal farming is a massively scaled attribution regime maintained by systematic L2 buffering, and that the buffering is generating slow-feedback incoherence of civilisational proportions. The same analysis applies, with appropriate modifications, to climate change more broadly: carbon-based industrial civilization is an attribution regime that treats atmospheric and ecological systems as non-recursive, as domains of pure regularities that do not respond to human classification or practice, and this attribution regime is generating slow-feedback L2 signals that cannot be permanently buffered.
In contrast, the attribution regimes documented by Descola among Achuar and other small-scale Amazonian peoples are characterized by short feedback loops and elaborate coherence maintenance work. The Achuar hunter who attributes full inter-recursive social agency to the peccary, treating it as a member of a tribe-species with its own social organization, its own shamanic mediators, its own capacity for vengeance, and then kills peccaries for food, faces an immediate coherence problem in every hunt. The ritual work that surrounds Achuar hunting is not exotic belief. It is the symbolic management of a high-attribution, short-feedback-loop mesocosm: acknowledgment, propitiation, exchange, apology, and the careful maintenance of reciprocal obligation across species that prevents the hunted collective from withdrawing cooperation or seeking revenge. The coherence maintenance work is expensive and elaborate precisely because the attribution is generous and the feedback loops are short. This is the structural contrast with industrial modernity, and it is a structural contrast about buffering and feedback loop length, not a contrast between different metaphysical worlds.
VIII. The Universal Structure of Human Discourse About Attribution
Once recursive mediational attribution is identified as a distinct level of human social life, something important becomes visible: an enormous proportion of human discourse, from intimate conversation to cross-cultural encounter to social media debate to legal proceeding to religious controversy, is L3/L4/L5 negotiation of attribution regimes. This is not a reductive claim. It is a recognition of how deeply and universally human beings are engaged in the activity of determining who or what counts as a participant in the recursive mediational field of their shared life.
When someone arrives in Kolkata and has a conversation about why beef is not eaten but chicken is, they are encountering a different attribution regime that has drawn the boundaries of inter-recursive participation and mediational membership differently across species. The cow in Bengal has been attributed a recursive mediational profile, dwelling, being-with, symbolic significance, inter-recursive participation in human social and cosmic life, that makes its consumption mesocosmically incoherent within that system, and this attribution has propagated coherently through practice, law, ritual, diet, economic organization, and emotional life across centuries. The encounter is already L5: each party reflects on their own L3/L4 attribution regime by encountering another, and both experience L2 discomfort, the felt misalignment of having their own attribution regime rendered contingent by the existence of a coherent alternative. This is what cross-cultural encounter actually is at its deepest level: an encounter between attribution regimes, each internally coherent, each generating L2 disturbance in the person whose regime is suddenly visible as one possibility among others.
The same structure operates at every scale of human social life. Disputes about animal welfare, wildlife conservation, rewilding, environmental rights, and climate justice are all, at their structural core, disputes about which entities should be attributed which recursive mediational profiles and what practical and legal consequences follow. Disputes about artificial intelligence personhood, AI consciousness, and AI rights are the newest and fastest-moving version of this same universal activity. Debates about the status of embryos, foetuses, and people in persistent vegetative states are another version. The common structure is: what recursive mediational profile does this entity have, and how should that profile determine how we coordinate with it, what we owe it, and what it can claim from us?
Descola sees something of this when he describes his project as an anthropology of collectives, not "society" plus nonhuman extras, but genuine assemblages of humans and other-than-humans whose composition reflects the guiding ontological schemas of the collectives concerned. He is right that the key question is who or what gets included in collective life, and on what terms. But his framework restricts this insight to broad ontological types distributed across cultural regions and historical epochs. The recursive mediational attribution framework extends the same insight to every scale of human life, every moment of contestation, every historical shift in how a mesocosm assigns membership and participation. It makes the question comparative across civilizations and applicable within them simultaneously.
IX. Reformulating the Contrast Without Primitivism
With recursive mediational attribution as the operative concept, the contrast between modern naturalism and Achuar-style animism can be restated precisely and without the romantic or primitivist undertow that haunts the ontology turn and has attracted persistent criticism of Descola's project. Adom Getachew, in her commentary in Politics of Worlding, notes the risk that the valorization of extra-modern ontologies becomes a form of conscription to modernity's self-critique, animists are useful to Descola's argument insofar as they provide conceptual resources for a critique of naturalism, but they remain positioned as the objects of that argument rather than as its subjects. The attribution framework sidesteps this problem structurally.
Modern naturalism is not simply an ontology in which humans and nonhumans share physicality but differ in interiority. It is an attribution regime that restricts recognized inter-recursivity, multimateriality, and multisymbolization predominantly to humans, while granting animals embodiment, dwelling, and increasing degrees of self-recursivity as ethological research expands. The restriction is maintained through L2 buffering at industrial scale, and the buffering generates slow-feedback incoherence that is already asserting itself through ecological, epidemiological, and climatic channels.
Achuar-style animism is not simply an ontology in which humans and nonhumans share interiority but differ in physicality. It is an attribution regime that extends recognized inter-recursivity, and in the most striking cases extends multimateriality and multisymbolization, to selected nonhuman beings. The extension requires elaborate coherence maintenance work, ritual, propitiation, exchange, shamanic mediation, precisely because the feedback loops are short and the attribution is generous. The coherence work is expensive; it is also, in ecological terms, conservative. Attribution regimes that attribute high recursive mediational profiles to nonhuman beings in direct, unbuffered contact tend to maintain more cautious, reciprocal, and sustainable relations with those beings. This is not because animists are ecologically virtuous by metaphysical design, but because the short feedback loops make the L2 consequences of attribution incoherence immediately visible and immediately costly.
Neither regime is ontologically correct. LVT does not rank cultures by how generous or accurate their attribution regimes are. It asks whether a given distribution of recursive mediational attribution is coherent with the coordination problems it faces and the feedback loops it operates within. Modern naturalism's buffering of nonhuman inter-recursivity made certain forms of science and industrial productivity possible. Achuar-style attribution maintenance made certain forms of ecological sustainability and multispecies sociality possible. Both are historically contingent solutions to coordination problems. Both generate their own characteristic incoherences when pushed beyond the conditions that sustain them. The task of analysis is not to adjudicate between ontologies but to describe the attribution regimes, the feedback loops, the buffering mechanisms, and the slow-accumulating incoherences with precision.
X. Conclusion: Descola Without the Ontology Machine
Descola saw something extraordinary. He saw that modern naturalism restricts the recognized participation of nonhuman beings in the social, symbolic, and material worlds that humans inhabit, and that this restriction is historically contingent, politically consequential, and intellectually impoverishing. He saw that in many human mesocosms, the beings that modernity relegates to "nature" are treated as full participants in collective life: as social agents, contractual parties, exchange partners, ceremonial actors, material producers, and territorial beings with their own normative orders. He saw that modern social science, because it is the child of naturalism, distorts the worlds it studies by projecting the naturalist attribution restriction onto non-naturalist mesocosms and then interpreting the difference as exotic belief or animistic projection.
All of this is correct and important. The problem is the conceptual apparatus through which Descola expresses it. His four ontological types are generated by a logical grid whose fundamental dimensions, interiority and physicality, are inherited from the very inside/outside metaphysics the grid is meant to transcend. His category of "interiority" bundles together intentionality, perspective, self-recursivity, social responsiveness, and symbolic participation, distinctions that LVT separates carefully and traces to different evolutionary thresholds. His category of "physicality" bundles together embodiment, morphology, materiality, habitat, and non-recursive reduction, again, distinctions that matter enormously for the analysis of what is being attributed and to whom. The result is that his framework can only say: this people treats animals as sharing interiority with humans. It cannot say: this people attributes to selected animals the full five-mediation architecture of distinctively human coordination, with the full range of inter-recursive social obligations this entails.
LVT allows us to keep what is best in Descola, his ethnographic richness, his critique of naturalism, his insistence that the unit of analysis must be the mixed collective rather than the human-only society, while replacing the ontology machine with something more powerful. The replacement has four components. First, a genuine process ontology: the human mesocosm has a universal architecture of five mediations and three recursivity domains, established through evolutionary reconstruction. This architecture is not culturally variant. It is what the mesocosm is. Second, a new concept, recursive mediational attribution, that identifies the level at which genuine cultural variation occurs: not at the level of ontology but at the level of which entities are attributed which mediations and which recursivity profiles within a given mesocosm. Third, a criterion, mesocosmic recursive mediational coherence, that allows attribution regimes to be empirically assessed: not for ontological accuracy, but for the consistency between attribution and practice across the full range of contexts in which the attributed entity appears, and for the production of L1 flow rather than L2 misalignment requiring management or suppression. Fourth, a mechanism, L2 buffering, that explains how attribution regimes can maintain incoherence at scale, and how the incoherence eventually asserts itself through slow-feedback channels whose signals are themselves subject to further buffering.
The crucial insight, which Descola approached but could not fully articulate within his framework, is this: the most consequential variation across human mesocosms is not ontological. It concerns the question of which beings are admitted to participation in the recursive mediational architecture of the mesocosm itself, who gets to have recognized dwelling, meaning acknowledged situatedness within a landscape understood as theirs; who gets to have being-with as genuine social membership in a collective field with obligations and recognitions; who gets to have multimateriality as the recognized capacity to transform matter in light of shared purposes; and who gets to have multisymbolization as the recognized capacity for ritual, communication, normative rule, and symbolic elaboration of the world. These are not questions about beliefs. They are questions about the living architecture of coordination. And they are answered not in explicit philosophical statements but in practices: how you hunt, what you owe after you have taken, what rituals manage the tensions between predation and attribution, what institutions of buffering allow large-scale predation under conditions of high companion-animal attribution, and what slow-feedback signals are beginning to break through that buffering.
More-than-human anthropology, multispecies ethnography, and the emerging anthropology of artificial minds are all, at their structural core, studies of recursive mediational attribution regimes: their content, their coherence, their feedback loops, their buffering mechanisms, and their historical dynamics. This is what Descola was studying in the Achuar highlands of the Upper Amazon from the 1970s onwards. He called it ontology. It is attribution all the way down, attribution that is real, consequential, historically variable, mesocosmically grounded, and subject to the corrective pressure of a living world that does not wait indefinitely for the symbolic order to acknowledge it. Descola gave it the wrong name. The phenomenon he identified is more important than the name he gave it.
Wait, I think the hamster's name was Crumbles. Nah. Mumbles? Also doesn't seem right. I give up. You get the point.