5 Diversification

Maturana and Varela themselves continued to explore the general paradigm of autopoiesis, of course. Milestones included the publication of Autopoiesis and Cognition: The Realization of the Living in 1980 [23], and the more popular or accessible account, The Tree of Knowledge: The Biological Roots of Human Understanding in 1987 [24]. Also significant was the appearance, in the early 1980's, of two collections of papers by diverse authors, both edited by Milan Zeleny [52,51]. These demonstrated a broadening interest in autopoiesis from researchers in a range of fields.

In the late 1980's there was also a return to the question of a minimal model of (molecular) autopoiesis--but this time not via computation but through real, ``wet'', chemistry. This direction of research was initially introduced in 1988, in a collaboration between Varela and Luisi [20], and quite quickly led to the exhibition of concrete chemical models [19].

Over that same period, from about 1988 to 1992, a strong philosophical interest, and critique, of autopoietic ideas emerged in the literature of general systems theory. This was particularly associated with the work of Fleischaker [4,5,6]. A special issue of the Journal of General Systems, edited by Fleischaker in 1992 [7], provided a forum for a robust debate, from a variety of specialists, over the scope, and, indeed, precise meaning, of autopoiesis.

For my purposes here, the most significant aspect of this discussion is the question of possible ``spaces'', or ``domains'', in which autopoiesis can be realised--and especially the status of purely computational implementations. Fleischaker has the merit of taking a very clear position on this: she holds that, as least insofar as autopoiesis is taken to be definitional or criterial of ``life'', then it must be interpreted as restricted ``entirely to the physical domain'' [5, p. 43]. In turn I take her to mean that computational models of autopoiesis must be regarded only as ``simulations'', and not ``realisations'' of life.

I am not fully persuaded by this position. In particular, it seems to me that we should distinguish between a purely formal description of a computation (e.g., the original ``algorithm'' of Varela et al. [46], or even a specific programme implementing this algorithm) and a concrete, physical, computer which is actually executing such an algorithm. The latter is clearly a very different kind of system from a conventional chemical, or molecular one; but, equally, it is certainly still a physical one. If it hosts autopoietic entities--albeit removed by a significant number of ``levels'' from the underlying ``hardware''--then those entities are surely also still physical. Thus, I would find it hard to categorise such an entity as ``not-living'' solely on a putative basis of its being ``not-physical''.

On the other hand, I am also unconvinced that debate on this issue will be particularly fruitful in any case. I would argue that--within the field of Artificial Life at least--the substantive issue at the current time is not whether a computationally-based autopoietic entity is ``really'' alive, but rather whether we can exhibit computationally-based autopoietic entities with substantially richer, more life-like, phenomenology. The early work of Zeleny was arguably developing in this direction; but, as we have seen, the concrete implementations of that work are not longer available, and it is now very difficult to critically appraise its substance.

I will therefore now return to consider a second phase of (attempted) elaboration of the phenomenology of computational autopoiesis.

Copyright © 2004 All Rights Reserved.
Timestamp: 2004-06-14