Document: Modelling Autopoiesis: Harder Than It May Seem!

``Artificial Life'', or ALife , is a newly named, if not newly established, domain of intellectual enquiry. It has strong ties to the Santa Fe Institute through the original series of ALife Workshops organised at the Institute by resident faculty member Chris Langton (Langton et al., 1992; Langton, 1989). There is a complementary European conference series, the first of which was jointly initiated by Francisco Varela and Paul Bourgine (Varela & Bourgine, 1992). More than one ``popular'' book has been published on the subject.

At the very least, ALife qualifies as hype; but is there still ``proper'' science behind the hype?

The guiding inspiration of ALife is to realise ``life-like'' phenomena in artificial media (especially computers), and thus both to improve our understanding of natural living systems and to open up entirely new technological opportunities for artificial living systems.

ALife is then, in major part, a synthetic enterprise--the attempt to create artificial life. A typical report of research in the field may consist in the claim that a certain (life-like) phenomenon will be exhibited by a certain configuration or arrangement of some artificial system; or, more specifically, by the execution of a certain class of computer model.

Because of this synthetic aspect of the work, the nature of scientific testing or criticism can acquire a somewhat different flavour from more traditional science. It is sometimes the case that ALife models are directly related to particular natural living phenomena; and in those case, claims arising from ALife research may be testable by comparison with empirical biological data. However, there is also a class of ALife model in which the relationship to specific living systems is very remote or abstract. In this case, the objective is not so much to explain a particular natural phenomenon, as to claim that a certain kind of phenomenon will be robustly exhibited by a broad variety of systems having a certain abstract or formal structure. In this case, critical testing need not involve comparing with pre-existing empirical phenomena, but may rather require independent synthesis of artificial systems, conforming to the stated abstract structure, and testing whether they do, in fact, demonstrate the claimed phenomena.

It should be clear that, in principle at least, ALife can thus be just as ``scientific'' as its longer established ancestors; but equally, the practise and methodology of this synthetic kind of critical testing may need adjustment.

3 A Case Study: Computational Autopoiesis

Modelling Autopoiesis: Harder Than It May Seem!

1 Prologue: The Scientific Enterprise

Document: Modelling Autopoiesis: Harder Than It May Seem!

mcmullin@eeng.dcu.ie

2 ALife: Hype or Science?