What is Autopoiesis?

Autopoiesis is a concept introduced by Chilean biologists Francisco Varela and Humberto Maturana (V&M), discussed in their book Autopoiesis and Cognition (which is where all the following quotes come from). They originally posited it as a definition of life, claiming it as a necessary and sufficient criteria; however, there was disagreement between them regarding its applicability to society on a larger scale. In his preface to the book, Stafford Beer takes the stance that the following are autopoietic: "Firms and industries, schools and universities, clinics and hospitals, professional bodies, departments of state, and whole countries" (p70). Niklas Luhmann would take this to further extremes, using the concept as the basis for his own theory of society, which claims that communication is autopoietic.

Autopoiesis literally translates to "self creation". It is defined by V&M as follows.


Autopoiesis and Cognition is a really dense book, and I have a hard time understanding what they're talking about a lot of the time. Here, I will present my own understanding of the concept, in a hopefully more readable manner. My thoughts are a bit messy, though, as I'm still trying to wrap my head around everything.

The given definition has two parts (labeled i and ii), which I will call autoproduction and autodistinction.


Autoproduction is, simply put, the production of the self. Homeostasis is a negative-feedback process which maintains a particular stable state by counteracting any drift away from said state. V&M refer to a "relations-static" system; that is, a homeostat which has its own organization as that which is maintained. The autopoietic system, they say, is a network of relations between components, and its function is to hold these relations constant via the production of new components (to replace old ones). The process is autopoietic, not the components which realize it, as they may be swapped out for new components.

Luhmann essentially does away with this particular relations-stasis concept. The systems Luhmann is concerned with are made up of operations, rather than a set of components and relations. (What are the 'components' and 'relations' of communication?) Autoproduction to him is simply that the operations produce more of the same kind of operation.

The point, I think, is that the system produces itself. If the system is a set of components and relations, then autoproduction is the system's production of components to maintain the relations. If the system is a bunch of operations, then autoproduction is just the system's production of more of its own operations. Whatever the system is, to be autoproductive (and possibly, to be autopoietic) is to produce itself.

I will suggest that this description of autoproduction can be seen as a uniting of the concepts of positive and negative feedback. A homeostatic (negative feedback) process maintains (produces) a particular identity. A positive feedback process rockets itself in a particular direction. The two seem very distinct at a glance, but when actually labeling them in nature, one may notice that they tend to appear together, and the lines between them begin to blur. A planet's gravity can be seen as a positive feedback process: the more mass the planet has, the stronger its gravity, which pulls in more mass. But, pull a bit of mass away, and it will tend to fall back- this is negative feedback. The same process (gravity) can be seen as either negative or positive feedback, depending on the way one describes it. A positive feedback process is, just like negative feedback, the production of an identity- that is, the 'identity' of the process itself. The exact relation between the two is something I'm still trying to understand.


Autodistinction is my term for that second part of the definition, the 'constitution as a concrete unity in the space' and 'specification of the topological domain of realization as such a network' (I told you this book is hard to read). V&M are inspired by Spencer-Brown's calculus of indication, which takes the concept of a distinction as its fundamental element (laid out in his book Laws of Form, and written about by Varela here). For there to be something, the something must be distinguished from what it is not. A "clean cut", separating the system (a unity) from its environment (background).


V&M put a lot of emphasis on the role of the autopoietic system as a unity. There is a rigid distinction between it and its environment, and that distinction is actively maintained. The autopoietic unity is the set of components/relations/operations which produce themselves; that which is not self-produced is not a part of the unity. Something may play a role in production, but not be produced (eg food). Something else may be produced, but not play a role in production (eg excrement; this is sometimes called allopoiesis). Only that which produces and is produced (by that very same production) is a part of the autopoietic system, which is a unity. The production of the self requires that the system have some 'knowledge' of what the 'self' is, in order to produce it. In doing so, it inherently draws a distinction between itself and its environment.

It seems to be important that the parts of the system (which it generates) are not generated outside of autopoiesis. This was apparently Maturana's issue with Luhmann's theories- Maturana did not accept that communication was autopoietic (as Luhmann claims), because he didn't accept that only communication can create more communication. Apparently Luhmann and Maturana had long discussions about this, as mentioned on page 79 of An Introduction to Systems Theory.

The 'cleanness' of the cut, or the discreteness of the distinction between the unity and its background is something that had been argued against, but V&M have maintained it. There is no ambiguity between the inside and outside. Similarly, there is no 'degree' or 'magnitude' of autopoiesis, V&R insist. A system either is or is not autopoietic. I don't know exactly how they justify this (outside of simply pointing to the postulates of Spencer-Brown's calculus), but they seem to be committed to it.

Luhmann makes heavy use of the concept of a distinction, but it's been argued that his interpretation of Spencer-Brown's metaphysics is somewhat peculiar. He assumes, for example, that the 'crossing' of a distinction fundamentally takes time (p47 of Introduction to Systems Theory), despite Spencer-Brown never claiming such a thing (according to this person, anyway). Luhmann also diverges from V&M in that he does not limit himself to spatial distinctions, such as the boundary of a cell, as V&R do. Different kinds of communication may be fully distinguished from one another, despite overlapping spatially.

The line between autoproduction and autodistinction may be blurred by the status of distinction in Spencer-Brown's metaphysics. If something is defined by the distinction between 'it' and 'not it', then it is defined as a distinction, and thus the production of the self is the production of the distinction. In this case, I think an autopoietic system can be defined as a system which both produces its distinction, and produces the production of its distinction. I'm not sure how well this fits with my above notes on positive/negative feedback.

My difficulty understanding this subject may largely stem from my not understanding the unique metaphysical foundations the theory lays on. V&M say things like:


I think they are just taking really seriously the 'distinction' metaphysics as the basis of everything, including their own existence and cognition as observers. Luhmann speaks in a similar way. I haven't really been able to understand this yet.


I will finish by linking some other resources which may be of interest.
- Notes on Autopoiesis
- Semiology of Autopoiesis (II) (also see the preceeding/following parts)
- Metaphysics: Deleuze│Spencer-Brown
- Autopoiesis and Eigenform