Evolution can explain how self-reproducers come to be adapted to their environment, but they may not necessarily ever be capable of evolving into anything we might regard as living (e.g. autopoietic organisations) because of the way in which their environment is constructed. We therefore need a theory of the necessary and sufficient requirements for the environment to support living organisations, as well as to support evolution.
Recall from Chapter 3 that Bedau has suggested that life should actually be defined in relation to a system that exhibits open-ended evolution (`supple adaptation' in his words) [Bedau 98b]. In this view, the evolving system as a whole (including individuals, interactions and environments) is the primary form of life, and particular components within the system (e.g. individual organisms) qualify as (secondary forms of) life by virtue of their specific relationship with the system. This is a heretical view even among those who define life in evolutionary terms (see Section 2.1.1), as even they usually offer definitions in terms of individual organism, cells, or genes, and not in terms of (ultimately) the biosphere as a whole.
However, Bedau's equation of life with open-ended evolution raises some intriguing issues. Chief among these is the question: To what extent does open-ended evolution imply life (in the common sense of the word, embracing phenomena such as metabolism, autopoiesis, food webs, evolutionary arms races, hierarchical evolution, etc.)? In other words, would it be possible to create a system with the capacity for open-ended evolution which did not exhibit these phenomena? If so, what would open-ended evolution look like without them?
As a specific example, it seems likely that competition for matter and energy would be essential to provide selection pressure for self-maintenance and autopoiesis. However, a purely logical implementation of a suitable evolutionary process might still be sufficient to bring about open-ended evolution. Comparison of the behaviour of logical versus material models might therefore lead to a better understanding of the relationship between the evolutionary and ecological aspects of life.
Regarding the fundamental nature of life, and specifically regarding its associated ecological (as opposed to evolutionary) phenomena, it is common to hear questions such as: Is metabolism a necessary component for a definition of life? (e.g. [Boden 99]). Using this question of metabolism as a specific example, I think it is more useful, given that life is not a well-defined concept, but that metabolism seems to be a ubiquitous feature of terrestrial life, and that there is widespread agreement that complex metabolisms have arisen from simpler origins by a process of evolution, to ask questions such as: How specific a form of evolution (in terms of individuals, interactions and environments) is required such that metabolism is able to (or necessarily) arise(s)? Conversely, is a form of metabolism required if a system is to have the capacity for open-ended evolution? Such questions can obviously also be applied to other features of living systems, such as autopoiesis, food webs, etc. The advantage of questions such as these is that they are not expressed in terms of the imprecise concept of life.
By modelling processes such as evolution and metabolism in artificial life systems and observing the resulting behaviour, we can see how closely this behaviour corresponds to our common conceptions concerning living systems. In this way, artificial life models can give us a better understanding of the relevance and interdependencies of such processes, and, in so doing, lead us towards a better understanding of the essential properties of life.