Abstract
Multimateriality, the many ways in which living beings recruit, produce, select, displace, alter, combine and anticipate heterogeneous material forms, has an evolutionary history far older than tools. This chapter reconstructs that history as an accretive stratigraphy. Material dependence on found conditions and metabolic externalization form its preconditions, and above them lie six layers of relation to matter: recruited self-produced matter, recruited found matter, selected and displaced matter, transformed matter, combined matter and projectively fabricated matter. The criterion throughout is recursive uptake: a material difference becomes multimaterial mediation only when it enters an organism's further coordination. Found and made describe relations to aspects of material configurations rather than classes of objects, so that a nest, a cave or an atmosphere can be found in some respects and made in others. The first five layers are documented outside the human lineage, and which of them a lineage can elaborate depends on the medium in which it lives. The human transformation consists in the sustained recursive coupling of multimateriality with multisymbolization, which makes projection possible: a form can be anticipated, shared, corrected and transmitted before and across its material realization. Using a Roman bridge and the silicon chip as paradigm cases, the chapter argues for a principle of downward answerability: poietic elaboration enlarges the space of possible forms without abolishing the conditions of possible fit, and every later layer remains answerable to the older material and zoetic relations it transforms.
I. Introduction: The Missing Evolutionary History of Multimateriality
The vocabulary through which the life sciences and the human sciences describe living relations with matter is strangely fragmented. An ape probing a termite mound with a stripped stem is engaged in tool use. A weaver bird knotting grass into a suspended chamber displays construction behaviour. A spider's web may be described as an extended phenotype (Dawkins 1982), a termite mound as animal architecture or as part of an extended organism, and a beaver's pond as ecosystem engineering. When humans alter their surroundings, the process becomes niche construction; when they make pots, houses and bridges, the subject passes to material culture, technology or archaeology. Each of these concepts is productive within its own literature. Taken together, however, they cut a single evolutionary continuum at different and largely unrelated places, and the continuum itself drops out of view.
This chapter proposes that the underlying phenomenon is multimateriality: the many ways in which living beings recruit, produce, select, displace, alter, combine and eventually anticipate heterogeneous material forms in the course of living. Multimateriality is one of the five mediations through which a living being remains coordinated with a world, alongside multisensorial embodiment, being-with, multiversal dwelling and multisymbolization. Treated only synchronically, as one item in an inventory, it can look like a human speciality with a few animal precedents. Treated evolutionarily, it turns out to have a history as long as life's relation to matter, and a structure that no single existing category captures.
Two propositions organize the argument. The first is that multimateriality has a layered evolutionary history in which new relations to matter become possible without eliminating older ones. The relations accumulate rather than succeed one another, so that the most elaborately fabricated human world still contains, and depends on, relations to matter hundreds of millions of years older than any tool. A corollary follows at once. Found and made describe relations between living beings and particular aspects of material configurations, and they do not sort objects into two classes: a nest or a house is found in some respects and made in others, and many boundary disputes about what counts as an artefact dissolve once this is recognised. The second proposition is that the human transformation of multimateriality consists in the extraordinary recursive coupling of accumulated multimaterial forms with multisymbolization, rather than in the first appearance of making. Finding, collecting, secreting, transporting, reshaping and combining matter all have long nonhuman histories. What the human lineage added was a loop in which material forms and symbolic forms continually reorganize one another across individuals and generations, so that a form can be anticipated, compared and redesigned before and across its material realization.
The title marks the two ends of the story. Secretion names the oldest way in which living beings bring into existence matter that was not previously present in their surroundings, by producing it from their own metabolism. The silicon chip names the most projective matter living beings have yet made, a material whose working structure is specified at a scale of billionths of a metre before any wafer is cut, and which nevertheless works only for as long as its crystal lattice, its purity and the dissipation of its heat behave as the specification requires. Between the two lies an evolutionary sequence in which living beings acquired ever more elaborate ways of making their worlds, and a stone bridge stands in the middle of the argument as the case through which that sequence is tested.
That sequence ends in a paradox. Evolution progressively increased the capacity of living beings to make their worlds, but it never increased their capacity to make worlds that need not fit. The chapter closes on this principle of downward answerability, which it develops through the case of a Roman bridge and carries forward to the chip.
II. Beyond Tools: Why Existing Categories Begin Too Late
St Amant and Horton (2008) define tool use as the exertion of control over a freely manipulable external object, either to alter the physical properties of another object, substance, surface or medium through a dynamic mechanical interaction, or to mediate the flow of information between the user and its environment. The definition is admirably precise. It also places outside the category almost everything else living beings do with matter: webs, nests, burrows, dams, secretions, shells, occupied cavities and the whole domain of material formations that are attached, too large to wield, or produced by the organism itself.
Hansell and Ruxton (2008) supply the strongest critique from within behavioural ecology, arguing that tool use belongs within the wider field of animal construction and that there is no reason to presume tool behaviour cognitively more interesting than building. Breen (2021) extends the point to cultural transmission, arguing that research on animal material culture has concentrated disproportionately on tool-using primates and corvids, and that avian nest construction, which involves material selection and in some species social learning, should itself be studied as material culture.
Anthropology has its own version of the problem. Ellen's (2022) ethnography of minimal tools among the Nuaulu of Seram distinguishes found, minimally modified and repurposed objects, and shows that containers, scrapers, whetstones and strike-a-lights barely distinguishable from unworked matter remain pervasive and consequential within technologically complex repertoires. His findings quietly undermine any ladder running from primitive found objects to modern technology, since the found and the minimal persist beside the elaborate instead of giving way to it. Heersmink (2022) reaches a compatible conclusion from the philosophy of cognition, arguing that nonhuman animals already use a surprisingly diverse range of artefacts and that human distinctiveness lies in the flexibility, variety and complexity with which artefacts are integrated into activity.
The niche construction tradition approaches the same terrain from evolutionary biology. Odling-Smee, Laland and Feldman (2003) show that organisms modify the selection pressures acting on themselves and their descendants through metabolism, activity and choice; Laland, Odling-Smee and Feldman (2001) show how human cultural practices amplify this process; and ecologists have long described such organisms as ecosystem engineers (Jones, Lawton and Shachak 1994). Niche construction theory is broad enough to include almost every relation this chapter discusses, but its explanatory target is the feedback between organisms and selective environments, and the structure of living relations to matter as such lies outside its concerns.
Godfrey-Smith (2024) has recently proposed reading the whole history of life from the side of what organisms do rather than what they are, as a history of organisms as causes of change in the Earth, running from the oxygenation of the atmosphere through the plant-shaped rivers of the Devonian to animal action and human culture. His account is the closest ally of the present one, and it supplies a distinction taken up below, between transformations of the environment that are byproducts of what organisms do and transformations that are the point of their activity. Its organising category, however, is action, and relations to formed matter enter it chiefly through the forms of action that produce them.
The difficulty with the tool concept therefore runs deeper than anthropocentrism. Tool use isolates one late and highly visible configuration, a detached object held and wielded by a mobile animal, within a much longer history of living relations with matter, while the broader concepts register that organisms change their surroundings without distinguishing the kinds of relation to matter through which they do so. The question accordingly shifts. Instead of asking when animals began using tools, it becomes possible to ask how living beings progressively acquired new ways of incorporating matter into the recursive continuation of life. Answering it requires beginning where the tool literature does not look at all: before making, in the relation between living processes and the material conditions from which they arose.
III. Before Making: Material Dependence, Dwelling and Metabolic Externalization
The first relation between life and matter is dependence. Living beings never encounter matter from nowhere: embodiment exists only within materially differentiated conditions of dwelling, gradients, surfaces, fluids, temperatures and chemical compositions that the organism did not make and cannot do without. One influential model of the origin of life makes this dependence unusually vivid. On the alkaline hydrothermal vent hypothesis, naturally occurring proton gradients across thin mineral walls, networks of micropores and catalytic mineral surfaces could have provided a structured geochemical setting in which early bioenergetic processes became possible (Lane and Martin 2012), and laboratory reactors have been built to test whether such conditions can drive the relevant chemistry (Herschy et al. 2014). The hypothesis remains contested, and nothing here depends on its being correct. What matters theoretically is that a leading model of life's beginnings starts with life exploiting a highly organised material environment it did not produce.
Material dependence therefore belongs to dwelling rather than to multimateriality, since it concerns conditions that are given, immovable and unmade by the organism that depends on them. Life begins by exploiting differentiated material conditions before it forms any matter of its own.
The second relation arises from the fact that no organism can live without changing its surroundings. Metabolism takes in, transforms and releases matter, and its products accumulate. Some of these accumulations have been geological in scale. The oxygen released by cyanobacterial photosynthesis, a metabolic waste product, gradually transformed the oceans and atmosphere of the early Earth, with consequences for every later form of life (Lyons, Reinhard and Planavsky 2014). The chalk cliffs of southern England consist largely of the calcite plates of coccolithophores. Darwin's (1881) last book showed that earthworms, simply by eating and excreting, turn over and remake the soil on which agriculture depends. Niche construction theory treats such metabolic change as one of the principal routes by which organisms modify the selective environments of their descendants (Odling-Smee, Laland and Feldman 2003), and recent work extends the same logic to organisms as Earth system engineers across deep time (Darroch et al. 2025).
Organisms alter their surroundings through growth and form as well as through chemistry. When land plants with extensive root systems spread in the Devonian, they held riverbanks in place and helped shift the characteristic form of rivers from broad sheets and braided channels toward confined, meandering courses (Gibling and Davies 2012; Godfrey-Smith 2024). No plant takes up a riverbank into its own coordination, yet the rivers of every later landscape run in part between banks that roots have held.
A distinction is needed at this point between passive consequence and recruited consequence. Oxygen dispersing into an atmosphere is metabolic externalization on an enormous scale, yet the cyanobacteria that released it never take the atmosphere up into their subsequent coordination as something they have formed. A metabolic product that merely disperses changes the world without yet entering a living relation to form. The crucial transition occurs when a material product of metabolism is recursively taken up into the organism's further coordination, so that the organism's own continuation comes to depend on the matter it has externalized.
Godfrey-Smith (2024) draws a related line between transformation and engineering. Early sponges filtering water may have gradually cleared the turbid seas, and early burrowers breaking through microbial mats aerated them, but in both cases the change was a byproduct of feeding; engineering, in his sense, is reserved for transformations that are the point, or part of the point, of the activity, as with a den or a nest. The two distinctions largely coincide, and where they diverge the divergence is instructive. Godfrey-Smith's criterion looks to production and asks whether the change was a goal. The criterion adopted here looks to uptake and asks whether the change enters the organism's further coordination. A material consequence produced for no purpose can later be recruited, as when a flake struck off in the course of some other activity is picked up and used. Recruitment is therefore the more basic relation for a history of multimateriality, since it tracks what matter comes to do within living coordination rather than what the organism was doing when the matter was formed.
The fate of unrecruited consequences also qualifies the sense in which dwelling is unmade. Godfrey-Smith (2024) stresses that an oxygen-rich atmosphere is chemically improbable for a planet like Earth, and that it was put in place by living activity along a specific historical path. What is found for every present breather was, in the deep past, the metabolic externalization of other lineages. Dwelling is unmade by the beings that depend on it, but much of it has been sedimented from the unrecruited consequences of earlier life. The relation between found and made therefore holds between particular beings and particular aspects of their world, and it does not divide the world into a natural region and a manufactured one.
Bacterial biofilms show how early and how basic the transition to recruitment is. Microbial cells secrete extracellular polymeric substances that form a matrix binding the community together, retaining water and nutrients, concentrating enzymes and buffering the cells against desiccation and antimicrobial agents (Flemming and Wingender 2010). The matrix is a product of metabolism and also the structure within which that metabolism continues: the cells live inside the material they have made. Shells, tubes, silk, mucus and secreted cements carry the same relation into animal life. An orb-weaving spider draws its web from its own silk glands, and many orb-weavers ingest the old web before spinning a new one, returning much of its protein to the metabolism from which it came (Peakall 1971). Here the circuit between living process and externalized matter closes in the most literal way.
Hansell's (2005) taxonomy of animal building materials already divides them into materials secreted by the builder and materials collected from the environment. The distinction is an important precedent, and it recurs throughout the stratigraphy below. Hansell classifies construction materials; the question here is whether the passage from secreted to collected matter, and from both to transformed and combined matter, marks successive transformations in the living relation to material form.
The first threshold can now be stated. Multimateriality proper begins when a material differentiation becomes recursively incorporated into an organism's subsequent coordination instead of remaining a mere metabolic consequence. Material dependence and metabolic externalization are accordingly its preconditions, the ground from which it arises, rather than its first layers. The criterion is recursive incorporation, and it does not require that the resulting structure lie clearly outside the body. The boundary between organism and externalized matter is in fact one of the most instructive places in the whole history: biologists can reasonably treat a mollusc's shell as part of the animal, Godfrey-Smith (2024) notes that it may have no definite answer whether a coral's secreted skeleton is its home or part of its body, and the biofilm matrix is at once the community's product and its medium. What matters is that the structure extends beyond metabolically active tissue as a relatively persistent material form that affects subsequent coordination, a nonrecursive counterpart the organism has itself brought into being.
IV. An Evolutionary Stratigraphy of Multimateriality
Above this ground, the relations between living beings and matter can be ordered as a stratigraphy of six layers. The ordering is partial rather than linear. Some layers do build on others, since selection and displacement presuppose a found material worth moving, and combination presupposes materials that can be joined. The stratigraphy is nevertheless no single chain of dependence through which every lineage passed in the same sequence: a hermit crab recruits found shells without any history of elaborate secretion, and a crow transforms twigs without secreting anything it builds with. Lineages have entered the stratigraphy at different points, elaborated some layers intensely and others hardly at all, and in several cases arrived at the same relation independently. The oldest macroscopic traces of life already join relations that the stratigraphy lists separately. Stromatolites grow as microbial mats secrete sticky films that trap and bind found sediment grains into domes and columns, so that self-produced and found matter were combined more than three billion years before any animal built anything (Godfrey-Smith 2024). What the stratigraphy describes is an expansion in the space of possible relations to matter available to life as a whole, and the way these relations stack within any single living configuration, where later relations incorporate earlier ones without superseding them.
The first layer is recruited self-produced materiality. The organism's own secretions or other products become functional parts of its further coordination: the biofilm matrix, the mollusc's shell, the spider's web, the tube of a polychaete worm, the wax comb of the honeybee. Hansell's (2005) category of secreted building material belongs here. This is the oldest form of making in the full sense, since it brings into being a material form that did not previously exist, though the maker's only material is its own metabolism.
The second layer is recruited found materiality. An organism incorporates an independently existing material structure into its life with little or no modification. A hermit crab occupies a gastropod shell made by another animal, and a single vacant shell can set off a chain of sequential shell exchanges that redistributes housing across many crabs (Lewis and Rotjan 2009; Rotjan, Chabot and Lewis 2010). A cavity-nesting bird occupies a hole it did not excavate. Found materiality may involve no alteration whatever, yet the found form is selected for what it affords and becomes part of how the organism lives.
The third layer is selected and displaced materiality. A found material is chosen, transported and repositioned because of what it can afford somewhere else. The veined octopus that carries coconut shell halves across open sand and later assembles them into a shelter is a clear case, since the shells are an encumbrance while carried and useful only at their destination (Finn, Tregenza and Norman 2009). Displacement introduces a new temporal structure into multimateriality: material action now occurs in anticipation of later material relevance. This anticipation should not be mistaken for symbolization. It is the embodied protention of an organism oriented toward what its situation is about to afford, and it operates as felt, practical orientation rather than as reference to something absent. Ellen's (2022) found and minimal tools belong partly here, because the selection of a suitable stone or shell can embody considerable practical knowledge even when the object is barely altered.
The fourth layer is transformative multimateriality. Found matter is modified, so that the usable form no longer simply exists in the encountered world. Materials are stripped, bitten, fractured, folded, woven, sharpened, crushed or heated. Chimpanzees in the Goualougo Triangle fray the ends of herb stems into brush-tipped probes and use them in sequence with puncturing sticks (Sanz, Call and Morgan 2009). New Caledonian crows cut and shape hooked tools from vegetation. The layer has an instructive threshold of its own. Wild bearded capuchins at Serra da Capivara strike quartzite cobbles against one another and in doing so detach sharp flakes resembling early hominin artefacts, but they do not use the flakes (Proffitt et al. 2016). The capuchins transform stone without recruiting the transformation, and their flakes stand to lithic technology as dispersing metabolic waste stands to the biofilm matrix. The case turns the threshold stated in Section III into a general criterion that holds at every level of the stratigraphy: material transformation becomes multimaterial mediation only when the transformation enters subsequent coordination. Producing a material difference and recursively recruiting that difference are distinct achievements, and the history of multimateriality is a history of the second. The Lomekwi 3 assemblage in West Turkana, dated to around 3.3 million years ago, shows hominins deliberately knapping stone before the appearance of the genus Homo (Harmand et al. 2015), and so crossing the threshold at which the capuchins stop.
The fifth layer is combinatorial multimateriality. Several independently sourced materials are assembled into configurations that have no found equivalent. Caddisfly larvae bind sand grains, plant fragments or tiny shells with their own silk into portable cases whose architecture is characteristic of each species. Weaver birds knot, loop and weave grasses into suspended chambers. New Caledonian crows in captivity have combined up to four elements, each too short to be useful on its own, into a compound tool long enough to reach food (von Bayern et al. 2018). Hominin combination eventually reached a scale and complexity that no found configuration could supply. At Kalambo Falls in Zambia, two logs were joined crosswise by an intentionally cut notch at least 476,000 years ago, the earliest known structural woodwork (Barham et al. 2023), and at Sibudu around 70,000 years ago stone points were hafted with compound adhesives of plant gum and ochre whose properties depended on controlled heating (Wadley, Hodgskiss and Grant 2009). Whether either case already involved projection in the strong sense is an open question the evidence cannot yet settle, but both show combination approaching the threshold of the next layer.
The sixth layer is projective multimateriality. A form is anticipated in symbolically and practically stabilised terms before it is materially realised, and it can be compared, corrected and redesigned in that anticipated state. Projection in this strong sense exceeds the embodied protention of the displacing octopus. It also exceeds a further capacity that the comparative record does document, private offline rehearsal. Rats can activate the internal maps of their hippocampal place cells to try out paths they have not taken, a pre-play of possible movement that Godfrey-Smith (2024) treats as an early form of plan-based rather than habit-based choice. Such rehearsal anticipates, but it remains within one animal and one episode. Anticipation becomes projective multimateriality when the anticipated form can be shared among makers, held across interruption, corrected in its anticipated state and transmitted to those who were not present. It depends on the coupling with symbolic operation discussed in Section VI. The first five layers all have substantial nonhuman instances, whereas the sixth is the layer for which the comparative record, so far as it has been assembled, offers no convincing nonhuman case, and it is here that the human poietic transformation becomes extraordinary.
These layers overlap in any real configuration. A nest may occupy a found cavity, incorporate transported twigs, include fibres the bird has stripped and softened, bind them with saliva or mud, and follow a construction procedure partly learned from others. It is misleading to ask whether the nest itself is found or made, because different aspects of the same configuration instantiate different multimaterial relations at once. The caddisfly makes the point in miniature, since its case is at once secreted, found, selected, displaced and combined.
Turner's (2000) The Extended Organism shows that structures built by animals, most famously the mounds of fungus-growing termites, perform continuing physiological work by regulating flows of air, water, heat and gases, blurring the boundary between an organism and its constructed surroundings. The question posed here is how durable nonrecursive material forms enter and reshape future recursive coordination while remaining counterparts, whether or not the mound literally becomes part of the organism. The mound regulates the colony's breathing, but it can be breached, repaired, abandoned and reoccupied, and each of those possibilities depends on its remaining matter the colony is related to rather than tissue the colony is made of.
The stratigraphy is also answerable to the medium in which it unfolds. Godfrey-Smith (2024) observes that engineering flourished on land to a degree it never reached in the sea. In water, objects drift, force is hard to exert quickly and shaped material rarely stays where it is put, so that most marine building is done by animals that secrete their structures from their own bodies, as corals do, while bilaterian animals build comparatively little. On land, shaped soil and saliva harden and persist, and termite towers can rise several metres above the colonies they ventilate. The octopus, the great manipulator of the underwater world, tends and fortifies its dens but builds nothing comparable to a weaver bird's nest or a termite mound. Which layers a lineage can elaborate therefore depends on the dwelling within which it lives, and the upper layers of the stratigraphy are in large part a terrestrial achievement. Downward answerability, developed in Section VII for a single bridge, operates here at the scale of evolution itself, since the medium shapes which relations to matter can accumulate at all.
The stratigraphy yields a principle of accretion rather than succession: evolution adds new relations to matter without deleting the older ones. The octopus that carries shells still depends on the chemistry of seawater, still metabolizes, still secretes mucus and ink, and still shelters in found crevices. Each new relation is added to those a lineage already has, and any living being possessing the upper layers continues to live through the lower ones and through the preconditions beneath them.
V. Layering Without Supersession: Cave, Nest, Womb and Food
The stratigraphy becomes phenomenologically obvious once it is followed through a few ordinary configurations. Consider first a bird nesting in a tree hollow. The hollow is found, a cavity the bird did not make. The twigs are found, selected and displaced from where they fell. Some fibres may be stripped or softened, and some material secreted or mixed with saliva. The finished configuration alters temperature, humidity, exposure to predators, the timing of reproduction and the being-with of parents and nestlings. No single term, whether tool, artefact, nest or niche, captures all of these relations, because the configuration is a stack of relations rather than one.
The human cave shows the same stacking at the hinge between dwelling and multimateriality. A cave is terrain, a feature of a landscape no human made, and in that respect it belongs entirely to dwelling. Occupying it is a relation of found dependence, shelter provided by rock that answers to geology and to no maker. Once people return to it, multimaterial layers begin to accrete on this found ground: a hearth, bedding of grass or hide, stones dragged into windbreaks, stored food, pigment on the walls, a narrowed entrance. Each addition belongs to a different layer, and none of them turns the cave into something made. Made forms are laid upon found conditions without replacing them, so that the most elaborately furnished cave remains answerable to the rock, the water table and the draughts that were there before anyone arrived.
The embryo supplies a reminder that the mediations cannot be mapped onto separate spatial regions. The mother is relevant to the developing embryo at once as the material and metabolic condition of its embodiment, as the place in which it lives, and as another living being with whom its coordination is interrecursive. The interrecursivity takes the form of reciprocal physiological responsiveness: fetal activity and signalling alter maternal processes, while maternal nutrition, activity and physiology alter fetal development. The interface through which this exchange runs is itself partly self-produced, since the placenta forms jointly from the embryo's trophoblast and the maternal decidua. The embryo thus lives from found conditions through a structure it has in part made. Super and Harkness's (1986) concept of the developmental niche, which treats the physical and social settings of childhood as mutually adjusted components of a single system, offers an established precedent for thinking such entanglement after birth. The mediations are analytically irreducible and existentially entangled, and the same concrete being can figure in several of them at once.
Food offers the strongest parallel case of layering, because it runs through every layer. Heterotrophs must recruit matter they did not themselves produce, and every later food practice elaborates that recruitment. Human foodways add selection, transport, storage, grinding, soaking, fermenting, heating, cultivation, breeding and industrial transformation. Sterelny (2010) describes the result as scaffolded digestion: cooking, pounding, chopping and fermenting shift a substantial part of digestive work outside the gut, a change Wrangham (2009) connects with the reduced teeth and guts of later hominins. Wollstonecroft (2011) treats food processing explicitly as niche construction, with reciprocal consequences for culture, biology and environment. Yet none of these layers supersedes metabolism. However much work is moved outside the body, the result must still be digestible, and the body still does the final work.
The history of maize shows what happens when an upper layer is transferred without the relation beneath it. In Mesoamerica and across much of the Americas, maize was traditionally cooked with lime or wood ash in the process known as nixtamalization, which releases bound niacin and makes it available to digestion. Katz, Hediger and Valleroy (1974) showed that societies relying heavily on maize almost always practised some form of alkali treatment, and that pellagra, a disease of niacin deficiency, was not endemic among them. Maize crossed the Atlantic without the technique. Pellagra was first described in Spain in the 1730s and became endemic among poor peasants in Italy, southern France, Romania and elsewhere, later also in Egypt, South Africa and the southern United States, wherever poverty confined diets to maize. The crop travelled as cultivated matter and as a named foodstuff. The processing that fitted it to human metabolism stayed behind, and metabolism exacted the difference.
Part of the reason the technique stayed behind lies in how such techniques are held. Henrich (2016) argues that food-processing traditions, such as the elaborate detoxification of bitter manioc, frequently work without their practitioners being able to state what they accomplish, so that their function is carried in practice rather than in any articulated account, a point Godfrey-Smith (2024) takes up in asking how far useful and seemingly arbitrary elements of a tradition can be told apart. A named crop can travel as a symbol and as a commodity. A function lodged in unarticulated practice travels only with the people who practise it.
A general principle follows. Later layers reorganize earlier conditions, but they cannot abolish the conditions on which their own viability depends. Agriculture reorganizes plant growth and reproduction without suspending plant physiology. Cooking reorganizes digestion without replacing it. Every elaboration of human food remains addressed, in the end, to a gut.
VI. The Human Transformation: Recursive Coupling of Matter and Symbols
An account of the human case has to resist two temptations at once: the exceptionalism that locates a categorical break at the first tool, and the continuity that dissolves every difference into degrees. The comparative record makes the first untenable. Finding useful objects is not uniquely human, and neither are selecting and transporting matter, building, altering materials or, as the crows' compound tools show, combining them. Socially learned material practices are not confined to humans either, since chimpanzee tool traditions vary across populations and nest construction in birds can involve social learning (Breen 2021). Mesoudi and Thornton (2018) distinguish core criteria of cumulative cultural evolution, which several nonhuman species may meet, from extended criteria, among them open-ended repertoires and functional dependence among traits, which have so far been documented only in humans.
The difference that remains lies in the relation between two mediations. In the human lineage, multimateriality entered a sustained recursive coupling with multisymbolization, and every mode of symbolic operation takes part in it. Symbols in coordination mode organize the joint work of makers as it happens; demonstrations, models and drawings re-present possible forms to perception; and in reference-to-absence mode a symbol can stand for a technique not currently being performed, a tool not currently in hand, a maker not currently present or a use not yet arisen. Reference to absence has a special role, because it is what makes strong projective fabrication possible, but it works alongside the other modes rather than in place of them. A made form can then be named, distinguished from its raw material, counted, demonstrated step by step, remembered, taught to someone who was not present at its making, compared with earlier versions, evaluated and criticized. Those symbolic differentiations alter subsequent making, and the altered material configuration becomes the ground for further symbolic distinction. The loop can be written as a sequence: make, discriminate, symbolize, demonstrate, reproduce, alter, redescribe, remake. Each pass deposits something durable into the material world and something durable into the symbolic repertoire, and each deposit changes what the next pass can do.
Recursive coupling in this sense has four features. It is bidirectional, since material change reorganizes symbolic possibilities and symbolic change reorganizes material possibilities. It persists across generations. It transfers across domains, so that innovations arising in one practical field migrate into others. And it is open-ended, continuing to widen the range of possible transformations instead of converging on a stable repertoire. The caddisfly's case, the weaver bird's nest and the chimpanzee's termite probe display extraordinary multimaterial sophistication, and their transmission may involve social learning and signalling. What no nonhuman lineage has so far been shown to sustain is a coupling of material and symbolic form with the open-ended, cross-domain, cumulative and intergenerational scope characteristic of humans.
The idea that material and symbolic life coevolved has distinguished predecessors, and the claim to originality here is correspondingly specific. Leroi-Gourhan's (1993 [1964]) Gesture and Speech treated technical action and language as twin products of a single evolutionary process, progressively exteriorized into tools, writing and machines (Copple 2003). Malafouris's (2013) Material Engagement Theory argues that engagement with matter is constitutive of cognition, so that the knapper thinks through the stone. Sinha (2015) comes closest of all, treating language as itself a kind of artefact and describing human evolution as the co-development of material culture, the technosphere, and symbolic culture, the semiosphere, within a biocultural niche. Rocha (2026) has recently synthesized niche construction theory, Sterelny's account of scaffolding and the archaeology of material innovation into an account of material innovations as evolutionary scaffolds, running from stone tools and fire to medicine and neural interfaces.
Each of these accounts begins with already recognizable human or hominin achievements, stone tools, gesture, fire, agriculture, and asks how material and symbolic capacities developed together from there. The argument here places that coupling at the top of a much deeper stratigraphy and claims that the human poietic mesocosm emerges when durable material deposits and durable symbolic deposits begin recursively generating, constraining and reorganizing one another across individuals and generations. Rather than creating multimateriality, the coupling seized a multimateriality that already possessed five layers and added a sixth, projective multimateriality, which no other lineage has been shown to enter.
Godfrey-Smith (2024), drawing on Wright (1976), offers a way to state this threshold sharply. Something can come about because of its effects in two broad ways. In the first, feedback from past cases does the work: earlier hearts kept animals alive, earlier nests raised young, earlier techniques succeeded and were copied, and later forms exist because of what earlier ones did. In the second, a future outcome is represented and pursued, even where nothing has ever produced it before. The first five layers of multimateriality can be sustained entirely by feedback from past cases, through natural selection, individual learning by trial and error, and the imitation of successful others. Projective multimateriality adds the second route, and with it the possibility of a made form without precedent.
Projection in this sense need not involve anything like a modern blueprint. Geometry, measurement, templates, verbally stabilised procedures, demonstrations, rules of thumb, recipes, partial plans and the experienced anticipation of practised makers can all stabilise a form before and across its realization, and can do so across makers who never meet and across intervals longer than any one life. The carved limestone pillars of Göbekli Tepe, raised between roughly 9500 and 8000 BCE, before or at the very start of farming in the region, show material and symbolic deposition coupled at monumental scale thousands of years before writing (Godfrey-Smith 2024). The asymmetry between the two poietic mediations matters here. Symbolic operation articulates, abstracts and compares, while material form is where the results of that articulation are deposited back into working, unreflective use. A procedure climbs to explicit articulation, and the finished object returns it to the level at which it can simply be lived with, until some failure sends part of it back to be articulated again.
Writing is the point at which the coupling of the two poietic mediations becomes literal. A written mark is a small piece of material engineering and at the same time a message addressed to a reader, who may be the writer at a later date, so that a single deposit belongs fully to both mediations at once (Godfrey-Smith 2024). What cognitive science calls offloading is accordingly more transformative than the word suggests, since an idea set down in marks takes on a form that can be worked on in ways unavailable to memory. Mathematical notation shows this most clearly, because to know a notation is to know how to produce new formulas from old ones. Writing did not begin the coupling, which is far older, but it gave the coupling a medium in which material and symbolic deposition coincide.
A Roman bridge can therefore exist as symbolically and practically stabilised anticipated form, and as a sequence of operations known to its builders, before any of its stones has been cut. That capacity returns the argument to the question with which the stratigraphy began. If the bridge is the most fully fabricated kind of object the stratigraphy allows, how free is it from the layers beneath?
VII. The Roman Bridge Principle: Poietic Freedom Without Escape
Take the bridge over the Tagus at Alcántara in western Spain as a paradigm of strongly fabricated matter. Built under Trajan in the early second century CE, and credited in its dedicatory inscription to the architect Gaius Julius Lacer, it carries a road across a deep river gorge on arches of granite. Unlike a cave or a conveniently forked branch, the bridge did not pre-exist in anything like its eventual configuration. Its stone was selected, quarried, transported, dressed, positioned and combined according to an anticipated relation stabilised in geometry, measurement, established procedure and the practised judgement of its builders. It belongs to the uppermost layer of the stratigraphy.
Yet being human-made does not make the bridge autonomous. It must fit gravity. Its granite must withstand compression, weathering and the impact of floodwater. Its foundations must fit the riverbed and the rock of the gorge. Its arches must fit their spans and loads, and its piers must fit the volume and speed of the river in flood. Its roadway must fit the gait of animals and people and the wheels of carts. Its location must fit patterns of dwelling and travel that existed before it and that it would reorganize. Its construction had to fit the coordination of quarrymen, carters, masons and overseers. Its measurements and procedures had to fit what stone actually does.
The building of a Roman arch makes this answerability visible. A stone arch cannot stand until its keystone is placed; until then its wedge-shaped voussoirs must rest on centering, a temporary timber framework built to the curve of the arch and struck once the arch can carry itself. The most projective feature of the bridge thus depends on a second, provisional multimaterial form whose only purpose is to hold the first in place while it becomes able to answer to gravity on its own. Vitruvius, writing more than a century before Trajan, advised that building stone be quarried and left exposed for two years before use, so that pieces damaged by the weather could be identified and relegated to foundations, while those that survived had passed what he called the test of nature (De architectura II.7). A symbolically stabilised rule of building here defers, explicitly, to a period of material testing that no rule could replace.
Ingold's (2010) critique of hylomorphism already rejects the idea that making imposes a preconceived form on passive matter. For Ingold, making unfolds as a correspondence with the forces and flows of materials, a textility in which the maker follows as much as directs. The present argument retains that insight and widens it, because the bridge must correspond with far more than its material. Here cosmic fit, mesocosmic fit and intersymbolic fit intersect without being reducible to one another. Cosmic fit is the bridge's answerability to gravity and to the strength of granite, conditions indifferent to every living being. Mesocosmic fit is its answerability to the river and the gorge, to the bodies that cross it and to the people who build, use and repair it. Intersymbolic fit is the consistency of its measurements, procedures and rules with one another. Success at one level cannot purchase success at another. Symbolic correctness cannot compensate for unsuitable material. Excellent material cannot compensate for foundations that fail. Structural integrity cannot compensate for a bridge that no living being can use. Social usefulness cannot make gravity disappear. A set of specifications can be perfectly consistent at the intersymbolic level and still describe a bridge that falls.
This gives the strongest formulation of the argument: poietic elaboration increases the space of possible forms without abolishing the conditions of possible fit. It also yields a principle of downward answerability. Evolutionary layering is asymmetrical, since later layers can reorganize older conditions while their continued viability remains answerable to those conditions. The word downward is genealogical rather than hierarchical. It names evolutionary and mediational dependence, and it claims no causal priority for matter in every event. Multisymbolization and multimateriality recursively transform embodiment, being-with and dwelling, as the bridge itself transforms the gorge it spans. Transformation, however, differs from escape. The bridge reorganizes the relation between two banks without suspending the river.
Downward answerability has a temporal dimension that the Alcántara bridge displays across nineteen centuries. Its temporal reach vastly exceeds the metabolic duration of any of its makers, which is one of the defining achievements of multimaterial deposits. That reach, however, has been sustained only through continued fit and continued repair. The bridge has been partly demolished in several wars and repeatedly restored, its broken arches rebuilt to a form that the surviving stone still embodied. The damage is recorded in historical accounts; the maintenance that kept the bridge standing between those episodes has left almost no trace at all, because successful maintenance prevents the very disruption that would have called attention to it. Misfit leaves an archive, while fit leaves a bridge.
Food makes the same point from the other end of the stratigraphy. Industrial food may be poietically elaborated almost beyond recognition, designed, branded, regulated and distributed through global infrastructures, yet if it ceases to be metabolically usable its symbolic sophistication is irrelevant. At both ends, in stone and in bread, the upper layers answer downward. Humans progressively acquired the freedom to make worlds, but never the freedom to make worlds that do not fit.
VIII. Multimateriality, Alignment and the Persistence of Older Layers
Older multimaterial relations remain continuously active inside the newest ones. A smartphone is maximally poietic in one sense and radically dependent in another. Its fabrication involves specification at almost every scale from the crystal lattice to the global supply chain. Yet it works only if its minerals retain their electrical properties, if human fingers and eyes remain the size and sensitivity for which its surface was designed, if energy systems, networks and repair infrastructures continue to function, and if the symbolic conventions it displays remain intelligible to those who hold it. The more layers accumulate within a configuration, the more a functioning object requires simultaneous fit across heterogeneous relations.
The silicon chip at the heart of such a device is the most projective matter the stratigraphy contains. Its circuits are specified symbolically at a scale no hand or eye can reach, and realised through sequences of deposition, patterning, etching and doping that answer at every step to the specification. Yet the chip works only because a crystal lattice conducts as the physics of semiconductors requires, because its silicon has been purified to a degree no found material approaches, and because the heat its operation generates can be carried away. Godfrey-Smith (2024) extends a version of this answerability to minds themselves. He wagers that felt experience depends on physical peculiarities of living nervous systems, their large-scale rhythms of electrical activity as well as their cell-to-cell connections, so that no program running on a present-day computer would thereby bring a feeling being into existence. Whether or not the wager proves right, it states the principle of this chapter in its most demanding form: what a made thing can become answers to what it is made of.
This runs close to the reverse of the familiar progressive narrative in which technology liberates humans from nature. Technology frequently shifts constraints, buffers them, redirects them, distributes them across other people and places, and temporarily hides them. Clothing buffered human bodies against cold without abolishing the need to regulate body temperature, and it introduced new dependencies on hides, fibres, needles and the skills to use them. Rocha (2026) notes that material scaffolds relax some selection pressures while introducing others, and that constructed niches often feed back against their constructors through resistant pathogens, new disease reservoirs and destabilized ecosystems. Under technological elaboration, material and zoetic answerability becomes more widely distributed and harder to see.
Godfrey-Smith's (2024) critique of Uexküll's image of each animal enclosed in its own perceptual bubble helps explain why. The effects of any organism's actions ramify through networks far beyond anything it perceives or acts upon, as the burrowing earthworm alters the soil for trees it will never encounter. Made worlds extend this ramification in the other direction, toward dependence rather than effect. The person holding a phone coordinates with its screen, while the mines, refineries, fabrication plants, power stations and repair workshops on which its working depends lie wholly outside anything that person coordinates with. The alignments a made thing requires increasingly run through regions its users never reach.
Multimaterial fit is never solely material fit. A dwelling, a food system, a tool, a road or a bridge is embedded simultaneously in embodied, interrecursive, dwelling, material and symbolic relations, and its viability depends on the alignment of all of them. The structure of this alignment differs from redundancy in the familiar perceptual sense. Bahrick, Lickliter and Flom (2004) show that information available to several senses at once, such as the rhythm of a bouncing ball that is both seen and heard, recruits infant attention and guides perceptual learning more effectively than information available to one sense alone. In intersensory redundancy, several channels carry the same amodal property. Multimediational alignment differs in kind, because the five mediations carry no common property through different channels. The strength of a granite arch, the gait of a mule, the depth of a river, the trust between a mason and an apprentice and the numbers in a builder's procedure share no content. Their convergence consists in irreducibly different relations jointly sustaining one viable situation, and misalignment can begin in any one of them.
This is why the upper layers of multimateriality are at once the most powerful and the most exposed. Highly poietic constructions generate dependencies, and with them sites of failure, that did not exist before them: a bridge can fail at its foundations, through the lapse of its maintenance, or through the collapse of the institution responsible for its repair. The more layers a configuration recruits, the more kinds of consequential misalignment become possible, even where redundancy and repair can buffer individual failures. Accretion enlarges what living beings can make, and it multiplies the kinds of fit on which the made thing continues to depend.
IX. Conclusion: An Accretive Evolution of Made Worlds
Tools are real, but they represent one local configuration within a far longer evolutionary history of multimateriality. That history rests on two preconditions, material dependence and metabolic externalization, and unfolds through six layers: recruited self-produced matter, recruited found matter, selected and displaced matter, transformed matter, combined matter and projectively fabricated matter. The layers form a partially ordered, accretive architecture rather than a ladder with abandoned lower rungs. Human beings still find, collect, occupy, metabolize, select and minimally modify things, as the whetstones and strike-a-lights of the Nuaulu show, while also building nuclear reactors and etching circuits onto silicon.
The three zoetic mediations, embodiment, being-with and dwelling, are always already co-present in living coordination. The precursors of material forming reach deep into nonhuman evolution, to biofilms, stromatolites, shells and webs, and each of the first five layers of multimateriality is documented outside the human lineage, most richly on land, where shaped matter stays where it is put. What becomes extraordinary in the human mesocosm is the open-ended expansion and recursive coupling of the two poietic mediations, multimateriality and multisymbolization, which allowed material forms to be specified, compared and redesigned through symbolic forms before and across their material realization.
Neighbouring accounts have supplied indispensable parts of this picture: the secreted and the collected, organism-environment reciprocity, the history of organisms as causes of planetary change, the distinction between byproduct and engineered transformation, the physiology of built structures, the persistence of minimal tools, the correspondence of making with materials, the coevolution of gesture and speech, the co-development of technosphere and semiosphere, technology as evolutionary scaffold and the ratchet of cumulative culture. The claim advanced here is their synthesis as a layered evolutionary history of multimateriality, organised by the criterion of recursive uptake, together with the principle that later poietic layers remain dependent on the continuing fit of the older relations from which they emerged.
That principle has consequences beyond evolutionary theory. The dominant modern story about technology, in which each advance loosens humanity's dependence on what it did not make, misdescribes what advances do. Every elaboration of made worlds adds to the number of alignments that must hold, and every one of those alignments reaches downward, eventually, to metabolism, to the bodies that use what is made, to the others with whom it is made and used, and to conditions of dwelling that no maker produced, even where those conditions were themselves laid down long ago by life. The maize that travelled without its processing, the bridge that stands only as long as it is repaired and the chip that works only while its lattice conducts, its heat escapes and its networks and users remain aligned are instances of a single asymmetry.
Life did not move from finding worlds to making them. Finding never stopped. Metabolism never stopped. Dwelling never stopped. Material dependence never stopped. Evolution added new ways of recruiting, moving, altering, combining and eventually anticipating matter, layering made worlds upon found ones. Human poiesis enormously expanded what could be made, but it never suspended the conditions under which making succeeds. The history of multimateriality is therefore the cumulative expansion of the relations through which living beings can recruit matter into further coordination, each later relation remaining available alongside the earlier ones and answerable to what living beings are and to the worlds in which they must continue to live, rather than a history of emancipation from the given.
Selected References
Bahrick, Lorraine E., Robert Lickliter, and Ross Flom. 2004. "Intersensory Redundancy Guides the Development of Selective Attention, Perception, and Cognition in Infancy." Current Directions in Psychological Science 13 (3): 99–102.
Barham, Lawrence, Geoff A. T. Duller, Ian Candy, et al. 2023. "Evidence for the Earliest Structural Use of Wood at Least 476,000 Years Ago." Nature 622: 107–111.
Breen, Alexis J. 2021. "Animal Culture Research Should Include Avian Nest Construction." Biology Letters 17: 20210327.
Copple, Mary M. 2003. "Gesture and Speech: André Leroi-Gourhan's Theory of the Co-evolution of Manual and Intellectual Activities." Gesture 3 (1): 47–94.
Darroch, Simon A. F., M. M. Casey, A. T. Cribb, A. E. Bates, M. E. Clapham, et al. 2025. "Earth System Engineers and the Cumulative Impact of Organisms in Deep Time." Trends in Ecology & Evolution 40 (11): 1066–1076.
Darwin, Charles. 1881. The Formation of Vegetable Mould through the Action of Worms, with Observations on Their Habits. London: John Murray.
Dawkins, Richard. 1982. The Extended Phenotype: The Gene as the Unit of Selection. Oxford: W. H. Freeman.
Ellen, Roy. 2022. "Describing and Conceptualizing Minimal Tools in an Ethnographic Setting: Implications for Understanding Technological Systems Holistically." Journal of Anthropological Research 78 (4).
Finn, Julian K., Tom Tregenza, and Mark D. Norman. 2009. "Defensive Tool Use in a Coconut-Carrying Octopus." Current Biology 19 (23): R1069–R1070.
Flemming, Hans-Curt, and Jost Wingender. 2010. "The Biofilm Matrix." Nature Reviews Microbiology 8: 623–633.
Gibling, Martin R., and Neil S. Davies. 2012. "Palaeozoic Landscapes Shaped by Plant Evolution." Nature Geoscience 5: 99–105.
Godfrey-Smith, Peter. 2024. Living on Earth: Life, Consciousness and the Making of the Natural World. London: William Collins.
Hansell, Mike. 2005. Animal Architecture. Oxford: Oxford University Press.
Hansell, Mike, and Graeme D. Ruxton. 2008. "Setting Tool Use within the Context of Animal Construction Behaviour." Trends in Ecology & Evolution 23 (2): 73–78.
Harmand, Sonia, Jason E. Lewis, Craig S. Feibel, et al. 2015. "3.3-Million-Year-Old Stone Tools from Lomekwi 3, West Turkana, Kenya." Nature 521: 310–315.
Heersmink, Richard. 2022. "Human Uniqueness in Using Tools and Artifacts: Flexibility, Variety, Complexity." Synthese 200: 442.
Henrich, Joseph. 2016. The Secret of Our Success: How Culture Is Driving Human Evolution, Domesticating Our Species, and Making Us Smarter. Princeton: Princeton University Press.
Herschy, Barry, Alexandra Whicher, Eloi Camprubi, et al. 2014. "An Origin-of-Life Reactor to Simulate Alkaline Hydrothermal Vents." Journal of Molecular Evolution 79: 213–227.
Ingold, Tim. 2010. "The Textility of Making." Cambridge Journal of Economics 34 (1): 91–102.
Jones, Clive G., John H. Lawton, and Moshe Shachak. 1994. "Organisms as Ecosystem Engineers." Oikos 69 (3): 373–386.
Katz, Solomon H., Mary L. Hediger, and Linda A. Valleroy. 1974. "Traditional Maize Processing Techniques in the New World." Science 184 (4138): 765–773.
Laland, Kevin N., John Odling-Smee, and Marcus W. Feldman. 2001. "Cultural Niche Construction and Human Evolution." Journal of Evolutionary Biology 14 (1): 22–33.
Lane, Nick, and William F. Martin. 2012. "The Origin of Membrane Bioenergetics." Cell 151 (7): 1406–1416.
Leroi-Gourhan, André. 1993 [1964]. Gesture and Speech. Translated by Anna Bostock Berger. Cambridge, MA: MIT Press.
Lewis, Sara M., and Randi D. Rotjan. 2009. "Vacancy Chains Provide Aggregate Benefits to Coenobita clypeatus Hermit Crabs." Proceedings of the Royal Society B 276.
Lyons, Timothy W., Christopher T. Reinhard, and Noah J. Planavsky. 2014. "The Rise of Oxygen in Earth's Early Ocean and Atmosphere." Nature 506: 307–315.
Malafouris, Lambros. 2013. How Things Shape the Mind: A Theory of Material Engagement. Cambridge, MA: MIT Press.
Mesoudi, Alex, and Alex Thornton. 2018. "What Is Cumulative Cultural Evolution?" Proceedings of the Royal Society B 285: 20180712.
Odling-Smee, F. John, Kevin N. Laland, and Marcus W. Feldman. 2003. Niche Construction: The Neglected Process in Evolution. Princeton: Princeton University Press.
Peakall, David B. 1971. "Conservation of Web Proteins in the Spider, Araneus diadematus." Journal of Experimental Zoology 176 (3): 257–264.
Proffitt, Tomos, Lydia V. Luncz, Tiago Falótico, et al. 2016. "Wild Monkeys Flake Stone Tools." Nature 539: 85–88.
Rocha, João. 2026. "Material Innovations as Evolutionary Scaffolds: A Niche Construction Perspective on Human Evolution." Evolutionary Biology, published online 11 September 2026. https://doi.org/10.1007/s11692-026-09677-y.
Rotjan, Randi D., Jeffrey R. Chabot, and Sara M. Lewis. 2010. "Social Context of Shell Acquisition in Coenobita clypeatus Hermit Crabs." Behavioral Ecology 21 (3): 639–646.
Sanz, Crickette, Josep Call, and David Morgan. 2009. "Design Complexity in Termite-Fishing Tools of Chimpanzees (Pan troglodytes)." Biology Letters 5 (3): 293–296.
Sinha, Chris. 2015. "Language and Other Artifacts: Socio-cultural Dynamics of Niche Construction." Frontiers in Psychology 6: 1601.
St Amant, Robert, and Thomas E. Horton. 2008. "Revisiting the Definition of Animal Tool Use." Animal Behaviour 75 (4): 1199–1208.
Sterelny, Kim. 2010. "Minds: Extended or Scaffolded?" Phenomenology and the Cognitive Sciences 9 (4): 465–481.
Super, Charles M., and Sara Harkness. 1986. "The Developmental Niche: A Conceptualization at the Interface of Child and Culture." International Journal of Behavioral Development 9 (4): 545–569.
Turner, J. Scott. 2000. The Extended Organism: The Physiology of Animal-Built Structures. Cambridge, MA: Harvard University Press.
Vitruvius. 1999. Ten Books on Architecture. Translated by Ingrid D. Rowland, commentary by Thomas Noble Howe. Cambridge: Cambridge University Press.
von Bayern, Auguste M. P., Sarah Danel, Alice M. I. Auersperg, Berenika Mioduszewska, and Alex Kacelnik. 2018. "Compound Tool Construction by New Caledonian Crows." Scientific Reports 8: 15676.
Wadley, Lyn, Tamaryn Hodgskiss, and Michael Grant. 2009. "Implications for Complex Cognition from the Hafting of Tools with Compound Adhesives in the Middle Stone Age, South Africa." Proceedings of the National Academy of Sciences 106 (24): 9590–9594.
Wollstonecroft, Michèle M. 2011. "Investigating the Role of Food Processing in Human Evolution: A Niche Construction Approach." Archaeological and Anthropological Sciences 3: 141–150.
Wrangham, Richard. 2009. Catching Fire: How Cooking Made Us Human. New York: Basic Books.
Wright, Larry. 1976. Teleological Explanations: An Etiological Analysis of Goals and Functions. Berkeley: University of California Press.