SIMPLE mathematical considerations suggest that adaptive systems in general must operate by a process of evolution, in the rather special sense of involving processes of differentiation from within some primitive unity, and of complexity from primal simplicity, under the influence of an environment which is more or less arbitrary.
This conclusion depends on the assumption that adaptability implies a capacity for responding to some set of varied circumstances which are not precisely specified in advance. Some such assumption seems to come very close to our intuitive ideas of what distinguishes living structure from mechanical structure; we can say in advance what mechanical structures will do, while the living organism remains to some extent unpredictable, and must remain unpredictable simply because this is the nature of adaptability, that appropriate reactions are possible in response to an environment which varies in more or less unspecifiable
Collinson JB. Ill-Defined Procedures in Learning and GrowthUnsolvability in Systems of Formal Logic May Provide Analogies With Evolution and Learning. Arch Gen Psychiatry. 1968;19(3):290-299. doi:10.1001/archpsyc.1968.01740090034004