From Autopoiesis to Symbiotic Entanglement: Rethinking Enactivism Through Metabolism and Microbes

Abstract

Enactivism has recently faced criticism for either leaning too heavily on philosophical speculation without clear scientific grounding, or relying on quite old empirical work in cognitive science, especially concerning sensorimotor actions. While one can push back against such charges, in this paper we take a different approach. We will use metabolic and microbiome research as a case study to help make this problem vivid, and to outline a path forward. First, we contend that a closer look at metabolism and microbiota-host interactions places pressure upon some of the core ‘autopoietic’ concepts of enactivism, including self-production, autonomy, and operational closure. This research instead appears to emphasise heteronomy and symbiosis in cognitive, developmental, and evolutionary processes, posing in effect an ontological challenge. Second, it also raises some questions about enactivism’s traditional avoidance of reductionist explanations, suggesting that there is an epistemic need for a philosophy of science that clarifies how to integrate more reductive biological programs within holistic enactivist frameworks. To meet these challenges, we argue that enactivism needs to moderate its commitments to autopoietic theory.

Introduction

Enactivism has been an influential position in philosophy and the cognitive sciences for more than a generation, typically proclaiming greater biological robustness than its more functionalist and computionalist opponents, and arguing that dynamical systems models can capture both biological and phenomenological structures. It has inspired a lot of interdisciplinary work (Barrett & Stout, 2024), notably with regard to embodied cognition and sensorimotor action, and it has directly influenced some important empirical programs in neuro-phenomenology and micro-phenomenology, initially through the pioneering work of Francisco Varela.

Despite this, some recent literature has also criticised enactivism for either bifurcating between an ontology or philosophy of nature without obvious direct scientific credentials, or drawing on old ‘greatest hits’ research in cognitive science, predominantly concerning sensorimotor action (see, e.g. Crippen, 2025; Meyer & Brancazio, 2023). Such charges might be simply denied, of course, or said to be a naturalist blackmail. It might also be replied that the tension between these dual commitments is typical of the conjunction of philosophy and empirical work – if so, there is nothing for enactivism to worry about, in particular, since it is a general concern. Indeed, perhaps it is even a fertile tension between philosophy of nature and empirical work that enables ‘mutual enlightenment’ (Gallagher, 1997). In this paper, however, we begin from the view that there is something to these recent criticisms in relation to the life sciences. We will explain the contours of this problem, and offer an account of its genesis, with regard to what has come to be called ‘autopoietic’ enactivism, which is where the philosophy of nature dimensions of enactivism are clearest. ¹

Our way in to this topic will be through an inquiry into metabolism and microbiota. ² Although some of the main ideas of enactivism were developed in juxtaposition with niche-construction theory, evolutionary developmental biology (‘evo-devo’), and phenotypic plasticity (e.g. Varela et al., 1991; and especially Thompson, 2007), enactivists have not yet systematically integrated the details and implications of recent microbiome research, despite a long concern with metabolism and cellular life (Maturana & Varela, 1980). Much of that emerging work appears to challenge core enactivist ideas about ‘autonomy’ and ‘operational closure’ for the organisms in question, since they are dependent on other organisms (microbes), and in ways that appear deeper than our weak causal dependence on oxygen for survival, say. It also introduces heteronomy into processes of cognition and development, and ideas of symbiotic co-evolution, thus helping to better situate organismic agency within a broader physical and selective environment than merely an organism’s own directly experienced Umwelt. While enactivism has been suspicious of the reduction of complex wholes to their functional parts, instead favouring holist, dialectical and organismic trajectories, it needs to also address what we can learn from such efforts, reductive or otherwise, to comprehend the complexity of the ontological coimbrication in host-symbiont assemblages, as well as to generate new testable ideas, and assist with theory choice between more and less reductive and/or ecological programs in postgenomic science. Otherwise we might still prefer an enactivist philosophy of nature, but the grounds for this may be more philosophical and speculative, rather than empirical. There is no doubt a place for that, but it seems that enactivism should be more constrained by empirical matters than that, given the commitments to ‘mutual’ and ‘reciprocal’ constraints first announced by Varela (1996), and then endorsed by many others.

The essay will proceed with two main sections. The following section will introduce the role that metabolism and autopoiesis plays in enactivism, focusing on Maturana and Varela’s early work on cellular biology, but also subsequent enactivist texts like Varela et al.’s The Embodied Mind and Thompson’s Mind in Life, as well as work in this tradition that draws on Hans Jonas. We then discuss some of the challenges presented by the microbiome and by host-microbota relations for enactivist views of organismic autonomy, both ontologically and with regard to understanding and adjudicating disputes about the epistemic credentials of reductive and ecologically embedded microbiome research. Finally, we canvass some of the options available for future enactivist work, ultimately recommending caution about the commitment to autopoietic theory, which we think has contributed to the separation of philosophy of nature from empirical research, with regard to the life sciences.

Metabolism and Autopoiesis in Enactivism: From Maturana and Varela to the ‘Jonasian Turn’

In the philosophy of biology, metabolism is recognised as an important feature of living things, along with other crucial factors like reproduction, heredity, and evolution (Barandiaran & Moreno, 2008; Landecker, 2017). A grain of sand does not metabolise. Nor do non-living self-organising systems – that is, volcanoes, hurricanes, and flames – that involve open, far-from-equilibrium thermodynamic systems. Metabolic perspectives draw distinctions between living and non-living systems by pointing to a combination of special (self)organisational elements and the environmental dependency of the latter. Of these ostensibly core features of living organisms, it is metabolism that has been emphasised in the enactivist tradition, mainly by virtue of Maturana and Varela’s influence on this tradition through their idea of autopoiesis (which for them is also connected with cognition), and then the elaboration of the idea of autonomy in subsequent work.

Maturana and Varela on Metabolism and Autopoiesis

Metabolism raises a variety of intriguing philosophical issues, including ontological ones. On the one hand, it locates life in the physical world by emphasising the need to integrate and expel energy at the cellular and organismic levels (c.f. Boden, 1999). On the other hand, metabolism is also difficult to account for as ‘mere matter’ – that is, via any particular physical structure or configuration. It is hence unsurprising that metabolism has played a role within enactivist thought, which inherited Maurice Merleau-Ponty’s ambitions to resist binaries like materialism and idealism (c.f. Varela et al., 1991, p. xi). However, the autopoietic account of metabolism put forward by Humberto Maturana and Francisco Varela in the 1960s and 70s did not initially directly derive from phenomenologists like Merleau-Ponty or Jonas (Varela read both systematically after this period). Although it is not explicitly spelt out in their own key texts, which are presented both succinctly and abstractly, it builds on older early 20^th century traditions in holism (Ganzheitsbiologie), dialectical materialism, and organicism, which developed theories of self-organisation and systemic views of organismal growth and metabolism (Baedke, 2019, 2025; Baedke et al., 2024; Drack et al., 2007; Nicholson & Gawne, 2015), and by linking these approaches with mid-20^th century developments in cybernetics. ³

Maturana and Varela’s ‘Santiago approach’ focused on cellular metabolism and the creation of a surface membrane, and then the ongoing generative maintenance of boundaries between inside and outside through structural and functional change (1980). On their view, cells circularly constitute themselves and their component parts, while also being structurally coupled with their environment and responding (structurally, but not organisationally) to perturbations from the outside (1980, p. 81). The second half of their book connects their account of autopoiesis and the organisation of living systems to the biology of cognition, with cognition understood broadly as acting with relevance to the maintaince of itself. This makes life and cognition inter-dependent ideas, sometimes framed as the life-mind continuity thesis (LMCT). This was first formulated by Maturana, in strong terms: ‘A cognitive system is a system whose organisation defines a domain of interactions in which it can act with relevance to maintenance of itself, and the process of cognition is the actual acting or behaving in this domain. Living systems are cognitive systems and living as a process is a process of cognition. This statement is valid for all organisms, with or without a nervous system’ (1980, p. 13). In more recent times, Evan Thompson has explicated this view well, although in less commital terms, arguing that ‘life is mind-like’ and ‘mind is life-like’ (2007, p. 128). Just what is meant by mind, cognition or sense-making (in later enactivist construals) continues to be debated, and a point of some ambiguity, but the overlapping organisational principles held in common are said to concern autopoiesis, autonomy, and operational closure.

Maturana and Varela’s accounts of metabolism and autopoiesis are quite abstract and formal in character. Although metabolism is indirectly implicated within their account of the cell (and multi-cellular organisms) maintaining integrity despite ongoing material exchange with an environment (ingestion, excretion, at the organismic level), their emphasis was on the ‘operational closure’ concerning the cell and its putative outside. Operational closure indicates, as Thompson puts it, that ‘every process or activity in the system enables and is enabled by another process in the system’ (2025, p. 7), in contra-distinction to what is not within the system. And, for Maturana and Varela, autopoietic organisation is not dependent on particular material substrates that are part of the structure of the system. Rather, the crucial issue is ‘whether the system’s proper components (whatever they are) realise the autopoietic organisation’ (Thompson, 2007, p. 105). For enactivists in this tradition, then, what mattered was the form/organisation of such systems, albeit with metabolism itself understood as ‘the biochemical instantiation of the autopoietic organization’ (Thompson, 2004, p. 389). Of course, there are many very different biochemical instantiations of metabolic processes, across species and between individuals of a given species, so the relationship between autopoietic organisation and actual biochemical processes is not yet clarified, other than through pointing to the competing demands of the ‘internal’ organisation and its resistance to ‘external’ factors, like evolutionary pressures, entropic forces, etc. As Maturana and Varela put the point: ‘although indeed energetic and thermodynamic considerations would necessarily enter in the analysis of how the components are physically constituted, […] these considerations do not enter in the characterization of the autopoietic organization’ (Maturana & Varela, 1980, p. 89).

Some questions have been raised about this stark distinction between organisation and structure. For example, is it a new version of hylomorphism (DiFrisco, 2014)? It might also be wondered if the view is truly dialectical, if organisation and structure are separated in this asymmetrical way. Of course, it is true that all cognising systems are ‘structurally coupled’ with their environment for Maturana and Varela. Every interaction with an environment brings about changes to the system itself, which responds to those ‘perturbations’ with changes to the system’s structure, but not to its autopoietic form. The autonomy they grant to autopoietic organisation is ostensibly warranted by the idea that this recursive self-organsation is found in different physical systems, as well as in the idea that life (understood via metabolism and autopoietic organisation) might be thought to be all or nothing, survival or disintegration, even if cognition/sense-making is not reducible to this binary (Di Paolo, 2005). ⁴ For Maturana and Varela, however, the autopoietic aspect constitutes what they call the ‘invariant configuration of relations’ (1980, p. xxi) of living systems. But just what do they mean by invariant at such an abstract level, as opposed to – say – processual flux? Can we distinguish between those two notionally opposed hypotheses? It is not clear how to give content to these relations and processes strictly formally, nor to individuate them with regard to existing taxonomies of organisms, something that any empirical sciences of metabolic processes will involve.

They specify that ‘organisation’ is to be understood as the ‘relations between processes of production of components; a domain of processes, not of concatenation of processes’ (Maturana & Varela, 1980, 79fn). This invocation of ‘relations’ and ‘processes’ has seemed suggestive of a process philosophy to some (c.f. Miencke, 2019), ⁵ as well as to a dialectical position (Di Paolo, 2022). But both claims might be reasonably disputed, because of the fundamental asymmetry between organisation and structure, and the closely related distinction between autopoietic and allopoietic levels or systems. Autopoietic systems are distinguished from allopoietic systems that generate a product that thereafter is independent of its origins (for example, a factory may produce a car but, after leaving the assembly line, the car doesn’t then recursively continue to impact the ongoing operations of the factory) ⁶ . For Maturana and Varela, whatever allopoietic processes of interaction occur, they do not directly impact the organisational level of autopoiesis, at least while the organism is alive (although they might cause the death of cells, organs, or the organism itself). ⁷ Such a view does not seem to be ‘dialectical all the way up and all the way down’, as advocated by Gilbert and Tauber (2016). Rather, at most there seem to be two dialectical processes – autopoietic and structural/allopoietic – that remain distinct. This is more akin to a dualism or hylomorphism than the intermingling and inter-penetration one might expect from a dialectical position.

Classical Enactivism

Many contemporary enactivist thinkers have drawn heavily on this autopoietic theory, albeit adjusting some elements of the picture. Without being exhaustive, notable changes include affirming autonomy as the key idea, but with autopoiesis at the cellular level rendered a subset of that (Di Paolo, 2005; Thompson, 2007). In addition, they seek to ameliorate an implication of Maturana and Varela’s early work regarding temporality: in brief, Maturana and Varela seem to counterpose an atemporal autopoietic organisational structure with external temporality (c.f. Williams, 2024), the latter of which might include structural and physical elements (entropy), as well as more obviously temporal definitions of living things, like reproduction, inheritance and beyond. The original autopoietic view thus arguably sidelines diachronic evolution, but later enactivist work is more obviously historical and temporal (Di Paolo, 2022; Thompson, 2007). It also emphasises a normative dimension of adaptivity and flourishing (Di Paolo, 2005), including with reference to organismic viability conditions, albeit with adaptation not understood in a neo-Darwinian manner as deriving from selection effects within a competitive world.

These changes are important, but we think it is arguable that some tensions persist, which derive from ongoing commitments to some core aspects of the earlier autopoietic theory. First, there is still a strong distinction drawn between organisation and structure (cf. Di Paolo et al., 2022). ⁸ Second, there is still an emphasis upon autonomy, with ideas like self-constitution still crucial, something which still confronts ontological questions with regard to the microbiome, as we will see. Third, there is also a holism about the organism and its milieu, and resulting reservations about approaches that are reductive in orientation. There hence remains a question about how one is to understand the many mechanistic and reductionist programs within biological science, since their limitations and insufficiencies appear to directly derive from core enactivist concepts, rather than from engagement with the pros and cons of any empirical programs in their specificity. The crucial concepts are quite disjunct from the empirical research, and they seem to entrench a divide between theoretical (and formal) reflections on life and the actual scientific work on metabolism (and the microbiome, as we will see), notwithstanding that both Maturana and Varela themselves conducted detailed and important scientific work across a number of diverse fields.

If we consider the classic text of enactivism, The Embodied Mind (Varela et al., 1991), it has only a few fleeting references to metabolism, but it extends Maturana and Varela’s organisational account of cellular processes to multi-cellular organisms, and to the sense-making activities of those organisms. It retains the holist and organismic aspects of Maturana and Varela’s Santiago period, and it continues to argue for the need to see reproduction, inheritance and development as co-evolved: that is, as interconnected and non-separable processes. This is surely an important reminder, in contrast to the ‘fractionation’ of such processes that became a cornerstone for the modern evolutionary synthesis in the second half of the 20th century (Walsh, 2015). Although this is supported by some empirical work connected with the so-called Extended Evolutionary Synthesis, in order to facilitate new empirical research, and grasp and assess competing views of the evolutionary role of the complex whole organism, the epistemic role of micro inquiries (even those that are reductive) needs greater attention, and in ways that cut across the organisation/structure binary. Otherwise, the grounds for preferring autopoietic enactivism against other views rests heavily on an under-specified philosophy of nature. Whether this line of criticism holds for enactivism in regard to cognitive science would require separate consideration, but it is also notable that that trajectory of enactivist work does not seem to depend as crucially on autopoietic theory, which is not mentioned in some of the major essays, like those on neuro-phenomenology (c.f Varela, 1996).

Jonas on Metabolism and Mind

Another additional issue with the enactivist efforts to integrate the empirical and the philosophical, with regard to the life sciences, is the role of Jonas’ work. Although it is not a consensus position for the field as such, Varela and Weber influentially argued for a ‘Jonasian turn’ in enactivism, beginning with a paper Varela completed towards the end of his life with Andreas Weber (Weber & Varela, 2002; for survey, see also Hverven & Netland, 2023). Some have argued against Jonas’ more teleological interpretation (see Benevides et al., 2023), but it is sufficiently influential to warrant a brief discussion here.

Jonas’ arguments about metabolism were part of his broader phenomenology of life (Jonas, 1966, chap 3), and they were expressly around the disanalogies with machinic (and mathematical) processes, and the problem that ensue with understanding living things as efficient causes or via mechanistic explanations. Again, however, Jonas appears to pre-emptively limit just what we might learn from the consideration of mechanistic and micrological processes, but which surely cannot be fully grasped from metabolic considerations based on the organismic level alone. Any contemporary enactivist framework also needs to be able to address and incorporate molecular biology, say, but without succumbing to reductionist ‘smallism’ (Wilson, 2004) or the ‘tyranny of the micro-physical’ (Dupré, 2024), and without losing sight of organismal phenomena by focusing solely on supra-organismal assemblages (populations) and sub-organismal entities (genes), as suggested by the modern evolutionary synthesis (Rosslenbroich et al., 2024; Walsh, 2015).

Jonas’ phenomenology of life has a quasi-existentialist interpretation of the organism’s active striving for the future, which borrows from Heidegger’s focus on Sorge (concern), but extended to animalia and beyond. For Jonas, unlike Heidegger, the striving is tied to concrete embodied existence via metabolic processes, and to what Jonas calls ‘needful freedom’ (1966, p. 80, c.f. also 4). As he puts it, ‘animal metabolism makes mediate action possible; but also makes it necessary. The animal, feeding on existing life, continually destroys its mortal supply and has to seek elsewhere for more’, and the food is itself motile, especially for carnivores (Jonas, 1966, p. 105). Metazoa need energy for their survival, ⁹ and later versions of enactivism emphasise this precarity and vulnerability too, even to the extent that it is a condition for identity and autonomy (c.f. Di Paolo et al., 2018). It is typically held that flourishing is not reducible to survival (which would be a reductive view of adaptation), and enactivists like Di Paolo and others retain the earlier argument (from Maturana and Varela) that normativity derives from the organism itself, rather than any bio-statistical account of flourishing with regard to standard deviations from an organismic mean.

Jonas also makes claims about an inwardness and interiority – perhaps something akin to sentience – with regard to feelings of strife within an ecological context, and his work is hence sometimes drawn on in debates about the life-mind continuity thesis (LMCT) within autopoietic enactivism. Sometimes his claims about interiority are focused on animalia, but he also makes remarks about cells and life more generally (Hverven & Netland, 2023). We cannot seek to settle Jonas interpretation here, but agree with those who argue his apparent endorsement of LMCT is an example of an overly speculative philosophy of nature (i.e. Kee, 2020; Benevides et al., 2023), and not readily amenable to empirical work (Fábregas-Tejeda & Sims, 2025), and also difficult to justify philosophically. ¹⁰ Once we introduce criteria – ‘hallmarks’ of cognition or ‘marks of the mental’ – we see a wide variety of capacities, rather than any necessary and sufficient conditions. With respect to sentience, there are ongoing debates about these (see Crump et al., 2022), but they can be separated into three broad categories: behavior (which is functional but also tacitly phenomenological, apprehended by us as sentient creatures, and with some empathic biases on account of this); evolutionary considerations (a plausible story about genetic similarity or homology); and physiology (mainly the ability for (neural) information transmission and/or the presence of nocioceptors to signal pain). Do the lowest forms of the organic have all hallmarks? No, although we can certainly recognise proto-versions in many organisms, using these criteria. Should we call the proto-versions ‘minded’ or ‘mind-like’? We can do so stipulatively, of course, but it may bias our inquiries as we seek to discover human-like neural patterns, and it is doubtful that such moves are well-aligned with research programs in the life sciences and epistemically fruitful.

Host-Microbiota Systems and Autopoietic Enactivism

We have argued that there are some difficulties faced by autopoietic enactivism, concerning both the epistemic value of reductive programs and a more heteronomous relational picture with the outside. In this section, we will extend the case by engaging with some recent microbiome research that more starkly shows the dilemma for enactivism with regard to empirical work and better reveals the ontological presuppositions enshrined in core aspects of autopoietic theory, as well as more recent revisions wherein autonomy is the key concept.

The Microbiome

The impact of microbes on health in humans and other mammals has long been known, commencing from the 17^th century and the discoveries of Antonie van Leeuwenhoek. But we now know a lot more about the vast numbers and influence of microbiota within the human body (and the bodies of many other mammals and organisms; Gilbert et al., 2012). They coat the body’s surfaces, including the surfaces of organs. Although a microbiome is acquired at birth, in humans, it interacts with food and the broader environment and is better described as a site of porosity or symbiotic entanglement, rather than operational closure, but it is also seemingly necessary for the functionality and health of the host. Often microbiota also seem to ‘act’ together as something like a colony within the body of the organism itself, which we perceive as a totality (a single body or so-called holobiont; Baedke et al., 2020). While in humans they do not act as one individual when it comes to reproduction (e.g. microbes are transmitted independently from human genes) they form tightly integrated functional wholes when it comes to physiological and metabolic processes, thus acting in some coordinated and joint goal-directed ways to maintain these processes.

Due to these tight connections, in humans many of the ca. 38 trillion microbes are crucial for the health of the host, where research increasingly uncovers their causal relevance for autoimmune diseases (Wen et al., 2008), metabolic disorders, and mental health (Valles-Colomer et al., 2019). Much of the scientific (and popular) attention has focused on the gut microbiome, and the connections between the enteric system and the central nervous system. Building mostly on animal models and comparisons with humans, in this field some are referring to this connection as the ‘gut microbiota-brain axis’ (Loh et al., 2024; Morais et al., 2021), due to the ca. 100 million nerve cells lining the gastrointestinal tract and the bidirectional communication between gut microbiota and the central nervous system. In fact, gut microbiota do not just passively receive signals from the central nervous system. They also convey information to it, and can thereby influence the brain in various ways, for instance, with regard to brain development, mental health, neurological disorders and neurodegenerative diseases. Just how crucial this enteric system is to cognitive processes is critically discussed across various research fields, but several connections have been established between neuroactive substances produced by microbiota and, for example, anxiety, cognitive impairment, or depression, and Parkinson’s/Alzheimer’s (for survey, see Boem et al., 2021; Arneth, 2025). Microbes are also needed in order to generate and develop healthy organs, implicating both the gut and the brain amongst other anatomical sites with regard to development, as well as in relation to functionality.

Microbes and Cognition

There are a series of possible theoretical positions that one might adopt regarding the nature of these host-microbiota connections and microbial constraints. At the most fundamental level, that is whether they are mere causal support for neuronal (informational) mechanisms that are the basis for cognition proper, or whether they are an essential part or aspect of cognition itself, as is intimated by popular slogans like ‘gut thinking’ (Schmidt, 2015). This is not resolved and, in fact, this view is critically received by many scientists, especially with regard to the extent of the interaction with the CNS. ¹¹ But despite the still mostly lacking mechanistic support for a microbial view of cognition, current first findings on bacteria’s neuroactive substances pose some intriguing questions. If this suggested causal influence could be strengthened, it could mean that we have to extend cognition beyond the putative biological individual, depending on how that individual is defined. If microbes contribute to human cognition ¹² , for example, just how are forms of basal cognition and human cognition coordinated to create a unified and purposeful acting holobiont? ¹³ This dependency of metabolism upon micro-organisms of multiple different species complicates biological individuality and any ‘selfhood’ in question (Gilbert et al., 2012; Triviño & Suárez, 2020), including the kind of boundaries and distinctions posited by autopoietic enactivism. Some treat the microbiome as a ‘commensal’ (within a holobiont assemblage), as ‘multi-species community’, or as another organ (Baquero & Nombela, 2012), but it seems different from other organs in various significant ways (Riccio & Rossano, 2020).

Might it be protested that this is an example of the causal-constitution fallacy (Adams & Aizawa, 2008), akin to claiming that the pencil is part of the mathematical cognition, or the heart is a constitutive part of cognition rather than just being causally necessary for it? We cannot think without oxygen and a functional heart, admittedly, but is it really part of the cognition? By extension, that kind of response might be given regarding the microbiome too, but it seems unlikely for an enactivist to give it, given the strong appeal to the whole embodied organism against the more neurocentric tendencies of functionalist or computationalist views. In addition, in the case of the microbiome we have seen that there appears to be a reciprocal and physiologically deep integrated link between us and other species, and which also could be linked to the mechanistic neural foundations that we deem relevant for cognition. As Jablonka and Ginsburg note, ‘not everything that causally affects the systems functioning can be considered to be part of it, to constitute it. In order to count as part of some whole (a system or an individual), the system must be shown to have coherent, stable properties, to which the parts contribute’ (2019, p. 173). It is arguable that this is what microbiome research is beginning to show. At the very least, there seems a genuine issue to address here, which might include philosophical and conceptual clarification to differentiate the various uses of the term ‘cognitive’ across the different sciences and philosophy. Committed to the LMCT, however, enactivism tends to consider them all as examples of sense-making, and hence may not parse this as precisely as is needed.

Challenges of Autopoietic Enactivism

Microbiome research also raises a direct puzzle for autopoietic enactivism, and some of its central concepts. As we have seen, autopoietic enactivism has tended to emphasise the way in which metabolism contributes to demarcating a cell/organism from what is not the cell/organism, by maintaining boundaries and unity, and with a form of individualism about cognition, agency and control (Dengsø & Kirchhoff, 2023). But it now seems apparent that this ostensible unity depends on alterity too, not just in the sense of the need for external resources to be absorbed (and then cease to be), nor merely by causing structural but not organisational perturbations to a given unity. Rather, the microbiome is akin to an other (foreign) body within the body. There is hence a paradoxical intertwining of self and other(s) that the classical enactivist concepts of self-production and autonomy don’t adequately recognise. Consider again some of the well-known statements of the view cited above (in Section 2), but also the following: ‘A system is autonomous if it can specify its own laws, or its own norms’ (Maturana & Varela, 1987, p. 48). This seems to be a restating of their distinction between organisation and structure. Perhaps, on more recent enactivist construals, it may not directly preclude external conditioning, but might be understood as simply requiring some autonomous normativity for the organism/system in question. But if so, we then come back to the question of the empirical relation between the internal organisational processes and external structural factors. In that regard, the current microbiome research seems to suggest that the laws or norms of even these ostensibly autonomous systems come partly from without, not just from within. It does not seem right to maintain that an autopoietic system ‘attains coherence through its own operation, and not through the intervention of contingencies from the environment’ (Varela, 1987, p. 55), nor that it ‘erects itself by its own means, and constitutes itself as distinct from the environment by way of its own dynamics’ (Maturana & Varela, 1987, p. 30). The indispensable metabolic and possible cognitive co-producing role of the microbiome mitigates against these views, including with regard to the sorts of features focused on by autopoietic and autonomy theory, like self-repair and intrinsically produced plasticity.

Microbiome research also makes visible autopoietic enactivism’s penchant for excluding the micro-organismic. This might seem a curious charge, given the role of cellular organisation. In addition, the enactivist’s ‘go to’ example is E. coli bacteria, which is found in the gut of humans and warm-blooded animals, and sometimes causes serious health problems. But, do enactivist’s reflect in detail on the role of other micro-organisms within the organism or on microbial genes and their functional links to host traits, as studied by metagenomics? Do they seek to trace the chemical microbial effects on this or that organism’s cognition and sense-making, including perception, learning, and beyond? It is certainly not the focus. But it is the focus of a lot of current scientific work (see, e.g. Castells-Nobau et al., 2024; Kolobaric et al., 2024). There is hence an opportunity here for enactivism to contribute to these debates about sense-making, but to do so we think it needs to preserve a possible epistemic role for reductive empirical science, especially genomic, metagenomic and microbiological research, and adjust some of the core concepts of autopoietic theory to better recognise externality and relationality across all levels of sense-making and organisational structure.

In this respect, we can recall that autopoietic enactivism typically excludes viruses from the domain of the living (and therefore sense-making) because they depend on a host and are without autonomy. Thompson also specifies some additional criteria, including the absence of metabolism (2007, chapt. 5). We can debate whether or not such exclusions are valid and appropriate, of course, but it seems that micro-organisms retain their ‘autonomy’ by enactivist lights, and yet they are also functionally necessary for the autonomous organism as a whole. Microbiota depend on the hosting organism, as with viruses, but the reverse dependency also holds, which considerably complicates the issue of discriminating between the inside and outside of an autopoietic or autonomous system (Thompson, 2007, chapter 5). If the micro-organisms in question are absent or degraded, then the organisms and cells in question are not able to function and self-repair, at least with humans and many mammals. They may not survive and their sense-making activities will be significantly attenuated. These considerations seem to support Mario Villalobos (2024) when he postulates an addendum to autopoietic theory to be specifically exo-autopoietic. Villalobos emphasises exogeny to give greater attention to the external origin. But we can now see another exogenous or heterogenous dimension, which is not just the external environment classically understood, but also with regard to the putatively ‘inner’. It complicates the auto-dimension of autopoiesis, since ‘the holobiontic perspective challenges the idea that biological individuality necessarily implies autonomy and endogeny’ (Boem et al., 2021, p. 18).

How might the enactivist who wants to retain the classical autopoietic concepts respond? We have suggested we don’t think they will have recourse to the charge of cognitive bloat, and the causal-constitution fallacy, but it remains a possible avenue (subject to ongoing microbiome research). They might also reason that the emphasis upon the microbiome is akin to thinking of autopoietic systems as allopoietic ones, with input-output relations (i.e. the charge of mechanistic thinking). As Maturana and Varela put it: ‘the fact that we can divide physical autopoietic machines into parts does not reveal the nature of the domain of interactions that they define as concrete entities operating in the physical universe’ (1980, p. 82). This is perhaps fair as a cautionary note with regard to reductionism, but if it is not complemented with a positive program (of reciprocal constraints) it can also seem precariously close to mysterianism. While thinkers like Di Paolo, Thompson and others have done crucial work emphasising that ‘to differing extents organisms are always historical’ (Di Paolo et al., 2017, p. 30; c.f. also Di Paolo et al., 2022), the diachronic historicity of particular bodies, in their metabolic interactions with the environment facilitated by their microbiome, has been given little attention, thus far. Perhaps they might double-down, so to speak, and insist that the human autonomous system determines the significance these micro-organisms has for its self-maintaining activity. But such a move appears somewhat stipulative, at least when we consider organisation and autonomy beyond the cell membrane, and perhaps even has the result that humans (like viruses) are not-living, if our dependence on the microbiome is in fact constitutive rather than causal.

Some of the ontological issues about autonomy and self-identity remain applicable to more recent enactivist thinkers. Di Paolo et al. still refer to ‘autonomous processes of material self-individuation’ (Di Paolo et al., 2018, p. 21). Ramírez-Vizcaya and Froese maintain that ‘it is the precariousness of its self-constituted metabolic identity (the potential to die that is intrinsic to life) that is at the origin of autonomous systems having a concern for its own conservation’ (2019, p. 5). Their emphasis on precarity is important, but we have argued that metabolic identity is not just self-constituted. Rather, it is co-constituted through the ongoing mediation of other heteronomous organisms. And this is not just those organisms (predators, food, mates) that are most directly and obviously implicated. Rather, it includes the colonies of other micro-organisms that comprise the microbiome, and symbiotically interact with those multi-cellular organisms that are more readily visible to the unaugmented (human) eye. They appear to be essential, playing a constitutive physiological role, without which various of our cognitive capacities are either compromised or cease. The enactivist emphasis on autonomy and self-constitution elides this crucial aspect of cognitive, developmental and evolutionary processes, thus seemingly falling into an ontological version of the immunological model criticised by Gilbert and Tauber, that is, ‘the immune system exists to protect the genetically homogenous and developmentally autonomous animal from pathogens. We argue that this defensive function falsely restricts the role of immunity’ (2016). ¹⁴ We need to investigate not just the mechanisms of insularity, that preserve or repair a boundary or border, but also the mechanisms of synergy as constituents of a large community, as Gilbert and Tauber put it (2016, p. 845).

Conclusion

There are many resources within enactivism that are crucial for an integrated philosophy of mind and biology today. However, we have argued that the ongoing commitment to some concepts from autopoietic theory (including autonomy) present difficulties for that project, and may be a partial explanation for the enactivist ‘divide’ between utopian philosophy of nature and empirical science that some have described. In our view, autopoietic enactivism hence needs supplementation, especially with regard to the ontological picture that creeps in via the affirmation of self-production, and with regard to the epistemic and organocentric holism that either eschews the significance of the consideration of (micro)parts, or leaves that to the scientists without integration. It might be to affirm ‘sympoiesis’ to a greater extent (c.f. Dengsø & Kirchhoff, 2023; Haraway, 2016), for example, and direct relationality, rather than radically separate organisation and structure. It might be to give greater credence to the epistemic need for modelling and local explanatory approaches, which need not purport to represent and explain nature or life as a whole, but are crucial in accounting for how the life sciences can enable us to glimpse more contextual ‘tell-tales of being’ (Morris, 2018), as emerging microbiome research has begun to do. Another kind of response to our argument might be to simply opt for other versions of enactivist theory that have distanced themselves from autopoiesis and autonomy, retreating from philosophy of nature. This might be said to be the way that radical enactivism and sensorimotor enactivism have already gone. But there is not yet detailed engagement with microbiome research in those areas, nor answers to the crucial questions that microbiome research is posing regarding cognition, biological individuality, environmental relatedness, agency, and beyond. We think these will need to be explored if enactivism is to fulfil its promise to provide biologically robust accounts of cognition and sense-making.

Acknowledgement

We would like to thank the peer-reviewers, along with Thomas Netland, Hayden Kee and Vera Straetmanns, for their helpful feedback and suggestions. Thanks also to the guest-editors for putting this special issue together, and their own work on related themes.