Why Gliders Don't Exist: Anti-reductionism and Emergence Pragmatic Anti-reductionism

ALife has always been centrally concerned with the nature and origins of emergent phenomena and their anti-reductionist implications for our understanding of complex systems. I argue that the traditional approach to understanding emergent phenomena in physical systems is still fundamentally reductionist, and outline an anti-reductionist alternative. Contemporary debate about emergence can only be understood as part of the much older debate about reductionism. Indeed much of the importance of, and interest in, the question of emergence in Artiicial Life is because of the light that it can shed on this much wider issue. The central point of this paper is that the usual arguments against reductionism are too weak, that they concede a crucial part of the reductionist case, and that a more radical approach is required. When we apply these arguments to the question of emergence we nd that the usual models of emergent phenomena are awed, and that an alternative is needed. The central claim of reductionism is that if all phenomena are on every occasion physically realised, then the laws governing those phenomena are determined by, and derivable from, the laws governing their constituent parts. Does a materialist have any alternative but to accept this priority of lower level entities over those they comprise? The usual alternative to reductionism is some form of pragmatic anti-reductionism which argues that although reductionism may be correct in principle, it can rarely be used in practise: it is simply not feasible to collect all the data, and perform the calculations necessary, for all but the most trivial of systems. In other words, that the properties of the whole may be determined by those of the parts, but it is (usually) impossible for us to derive them. Basic pragmatic anti-reductionism can be strengthened in various ways. We can borrow from chaos theory and argue that aggregate properties of the system may be sensitive to some properties of a part, such as the infamous sensitivity of weather systems to a butterry's wing. If this is the case then an accurate derivation of a higher level description would require that the properties of the parts are known with unbounded accuracy, and there are various reasons, such as the Uncertainty Principle, why this is not possible. A pragmatic anti-reductionist can also argue that just knowing the properties of the parts is not enough to derive higher level properties; we also have to know the composition …