The Possible Irreducibility of Artificial Software Life - Bruce Edmonds

2 Reducibility as Computability

In the most general terms, the reduction of a theory is a translation of it from an initial descriptive framework to a "more basic" framework. This translation process can be done in many ways. Sarkar [6] suggests that the type of reduction you use needs to be related to the task at hand, so this becomes a sort of pragmatic decision. It has also turned out to be difficult to formalise actual reductions done by scientists.

In theoretical biology, debate about reduction has been often been centred around reduction to a Turing machine, i.e. taking reducibility as computability (especially by opponents of reductionism, e.g. Rosen [5] or Pattee [4]). Given this, it is a foregone conclusion that (hardware glitches, and interaction with a real environment apart) that any artificial life would be reducible. Below I argue that this characterisation of reducibility is too weak for any deliberate reduction.

Using this criteria of reducibility the reductionist and holist positions can be pictured like this (after Rosen [5]).

Diagrams of some holist and reductionist views

The Possible Irreducibility of Artificial Software Life - Bruce Edmonds - 20 MAY 97

[Next] [Previous] [Top] [Contents]

Generated with CERN WebMaker