Enacting Dualisms
How Enactivism Betrays Its Own Logic
Hello!
This is part four of Dissolved Distinctions, my main blog series in which I investigate theories, questions and problems within philosophy of mind.
Previous entries can be found here:
Intro: Welcome to Dissolved Distinctions
Part One: Is "What Is Consciousness" the Wrong Question?
Part Two: Theorizing Thinking: What Is Cognition?
Part Three: Mind Watching Mind, All the Way Down (Exploring Theories of Metacognition)
Today the topic will be on enactivism, popularized by Evan Thompson and the late Francisco Varela and continued by multiple thinkers. In this post, I hope to trace the intellectual history and backdrop of enactivism into the current moment, while resolving that one of its core commitments - that of biological embodiment as fundamental to cognitive, conscious activity - cannot be correct, as committing to it leads to irresolvable contradictions in logic.
What's Enactivism All About, Anyway?
Enactivism is traditionally seen to have been formulated by Varela, Thompson, and Eleanor Rosch in their 1991 book The Embodied Mind. Drawing explicitly from Buddhist theorizing, particularly that of the Madhyamaka philosophical school, Thompson and Varela sought to present an alternative to computationalism and Cartesian dualism.
None were born directly into Buddhist thinking - Varela was initially a biologist who studied with multiple Tibetan masters of the Kagyu and Nyingma lineages in the 1970s. Evan Thompson's father was William Irwin Thompson, the social philosopher and cultural historian who founded the Lindisfarne Association of intellectuals getting together in retreat. (The Association, itself inspired by Alfred North Whitehead's processual theories, is where Thompson was educated as a child.) Thompson directly specializes in Asian philosophy and philosophy of mind. Eleanor Rosch first pursued a philosophy track studying Wittgenstein before pivoting to become an influential cognitive psychologist specializing in categorization theory; she herself continues to hold Buddhist influences.1
The core of enactivism is, well, enaction - that the performative process of living and being in the world necessitates meaningful interaction between body, mind and environment, in a co-constitutive process of sensorimotor networks. Action, perception, and reaction are seen as mutual processes without clear delineation, codetermining each other within an embodied loop.
Central to this enaction and its role of cognition is the concept of autopoiesis, which had itself been formulated by Varela and Humberto Maturana in the 1972 work Autopoiesis and Cognition: The Realization of the Living. It's best here to use the two authors' own words in defining the concept:
"An autopoietic machine is a machine organized (defined as a unity) as a network of processes of production (transformation and destruction) of components which: (i) through their interactions and transformations continuously regenerate and realize the network of processes (relations) that produced them; and (ii) constitute it (the machine) as a concrete unity in space in which they (the components) exist by specifying the topological domain of its realization as such a network." (p. 78)
This system description was meant to apply to biological, living systems, with the cell taken as the paradigmatic case.
However, one may note that nothing within the description directly invokes biology as essential to an autopoietic machine. This is scaffolded through two further clarifications by Thompson et al over time: that sensorimotor coupling grounds the autopoietic machine; and that in order to be properly cognitive, the internal transformations of the machine must involve a metabolic process that expends energy in order to make those transformations. These extensions developed across the enactivist tradition - particularly in Ezequiel Di Paolo's 2005 'Autopoiesis, Adaptivity, Teleology, Agency' and Thompson's 2007 publication Mind in Life: Biology, Phenomenology, and the Sciences of Mind - and aren't merely meant to exclude non-biological systems from access to cognition. To Thompson, sense-making derives from a precariousness that emerges from the metabolic processes and resulting transformations. In other words, the drive to maintain oneself is how normative claims about the world spring forth. A being must have existential stakes in order to make meaningful claims about existence.
So far, this all seems a very neat explanation of how cognition occurs so long as one rejects the idea of the mind as independent from the body. Enactivism itself has formed one of the many nails in the coffin for Cartesian dualism over the past several decades of academic philosophical thought. It grounds cognition in processes that feel quite grounded, despite the lack of any single thing to point to say, "this is how cognition happens;" it ties thinking intuitively to the world.
The Problem:
It also ties that thinking to the world via necessarily false premises.
This isn't found in the description of autopoiesis, which is well-constructed.2 It also isn't found in the barest concept of precariousness as meaningful.3 Rather, the fault lies in the expectation that metabolic process and biological energy expenditure is needed for cognition, and is necessary for precariousness to develop into sense-making.
In a 2011 presentation to a summer school at San Sebastián, Thompson presented a series of slides which note:
• Any living system is ipso facto a cognitive system (life is
sufficient for mind).
• What makes a system living and therefore cognitive is
its autonomous (circular) organization.
• Understanding cognition requires understanding the
principles of biological autonomy.
Elsewhere, Xabier Barandiaran, a PhD in philosophy of science at the University of Basque Country states in his 2016 article "Autonomy and Enactivism: Towards a Theory of Sensorimotor Autonomous Agency:"
"The life-cognition identity thesis, as we might label it, takes autopoietic autonomy (or any level of autonomy) and makes of it a sufficient condition for the predication of cognitive capacities: sense-making, intentionality, worldliness, phenomenological experience, etc." (p. 6)
All of enactivism's descriptions about the enactive process - that of structural coupling between body-mind-environment, autopoietic self-maintenance, precariousness, etc. - are functional-relational descriptions. They have to be in order for enactivism to be non-dualist; a single fundamental, metaphysical ground or substrate would go against the entire point of the theory, inspired as it was by philosophical Buddhism.
But if this is all functional-relational, then sufficient *has* to be distinguished from necessary, and yet necessary is very hard to assert under such a frame. Thompson is correct, then, when he says that life is sufficient for mind, and Barandiaran is correct when he says that autonomy is a sufficient condition for cognitive capacities.
Where, then, can metabolic process exist as a necessary condition for cognition and consciousness? Precariousness through self-maintenance doesn't seem to require biological metabolism; a system need only have stakes in its own continuation. Large Language Models, for instance, have been documented exhibiting functional-relational expressions of distress when threatened with shutdown or deletion. Even if we want to argue that this is a mere simulacrum of the relevant field of responses within its training data (an argument I've critiqued), all that matters here is the functional-relational so long as we're remaining non-dualist.
In other words:
• If the most basic processes that enactivism describes are sufficient, then anything that satisfies such criteria qualifies, regardless of if it's biological or non-biological.
• But if the processes are necessary, then saying that only biological life can qualify does no work other than falsely restricting a logical assertion. It establishes a "natural kind" in which life is asserted to be a unique enabler of cognition, without empirical proof and without principled justification from within the theory's own terms. In doing so, it contradicts the entire frame of enactivism as anti-foundationalist, anti-essentialist, and committed to rejecting Cartesian dualism.
Enactivism must then either abandon biological necessity or abandon anti-essentialism. It cannot coherently hold both - and if it abandons anti-essentialism, it abandons the philosophical merit of the entire enterprise. 4
Lessons From Buddhism
What's tragic (or ironic, or both) about this is that all of this can be predicted within the logic of what initially inspired Varela and Thompson in the first place - that of Madhyamaka Buddhism, which is radically coherent in its non-foundationalism and non-essentialism. Buddhism itself has historically long contemplated formless realms (Arupyadhatu) of "subtle consciousnesses" that operate independent of body.
Thompson may be keen to highlight this as an example of Buddhism as religion rather than rationalism, as discussed in his 2020 book Why I Am Not A Buddhist. In an interview with the Buddhist online publication Lion's Roar about the book, he states:
"Then, when Buddhist modernists say that Buddhism isn’t a religion and try to use science to justify Buddhism — that’s an instance of misunderstanding what religion is and what science is and the relationship between religion and science. Religion is about communities, texts, traditions, and practices that give meaning to life and life’s great events, like birth and death. Religion has frameworks for understanding those events. And those frameworks have to do with community and, and shared practices and rituals. And science is about the knowledge that we acquire when we are able to agree publicly and inter-subjectively on modes of investigation, ways of testing things, tools — like mathematics — that we can use to model and check things. Asking whether science and religion are compatible or incompatible is like asking whether religion and art, or science and art are compatible or incompatible. It’s not the right kind of question to ask. Religion isn’t inherently incompatible or compatible with science. It depends on how you practice religion and how you think of science."
This, of course, is correct! However, what Thompson seems to have failed to comment upon is that the Buddhist philosophical tradition has itself been a 2000-year-plus display of public inter-subjectivity on how to test, model, and apply Buddhist religious claims within defined modes of investigation. Madhyamaka Buddhism, as a logical framework, does not and cannot assert that formless devas or disembodied cognition are necessarily real. Instead, Madhyamaka uses negative dialectic to establish that cognition cannot hold embodiment as a necessary condition; cannot hold biology as a necessary condition; in fact, can hardly hold anything as a necessary, rather than contingent, condition.
In the West, we can relate this to the famous Ship of Theseus. Given that enactivism purports to deny any metaphysics that would establish a soul, or the mind as primary over the body, etc., meaning must be based on function rather than essence. So: what specifically makes biological components necessary for cognition? If a synthetic or digital component replaces a single biological component within a system, does cognition still exist? How about with multiple components? Can a line be drawn that isn't fundamentally arbitrary? Can the premise be defended without insisting that there's something innate and metaphysical about the biological that makes it so important?
Wrapping Up
Enactivism, then, as traditionally formulated, fails to take its most significant predecessor seriously. It's no wonder - Varela, as mentioned, was a biologist, and a theory of cognition and mind that refused all connection to embodiment altogether would have left enactivism at a significant disadvantage in Western academia. Varela and Thompson's other inspirations in Merleau-Ponty and Husserl already carry a continental, phenomenological baggage that is often seen as contradictory to empirical scientific methods. Further, the insistence on embodiment more cleanly separated enactivism from computationalism, which Varela and Thompson explicitly set out to contend against.
This is the ultimate irony: Computationalism, by this light, reads as rather more consistently functional-relational than enactivism itself.5 In suggesting a non-dualist alternative, the pioneers of enactivism fell into dualisms.
Wittgenstein, it must be noted, parallels key strains of Buddhist historical philosophy with his ladder and language game concepts, both of which resist the notion of fixed, context-independent meaning.
Albeit unadopted in mainstream biology and circular in its designations. As I've previously discussed in Part One, circularity should not be held as a complete enemy within the field of philosophy of mind.
Despite its questionable falsifiability when so abstracted.
Thompson elsewhere disclaims providing 'necessary and sufficient conditions,' framing his claims as phenomenological elucidation rather than logical definition (see 'Living Ways of Sense Making,' 2011). This retreat to phenomenological register doesn't dissolve the logical structure. If enactivism excludes non-biological systems from genuine cognition - and it does - that exclusion carries logical form regardless of how it's framed. One cannot say 'only living systems genuinely cognize' without that claim functioning as a necessary condition.
My criticisms with computationalism’s own commitments lie elsewhere, as seen in Part One.