Quantifying Non-trivial Open-Ended Evolution Reveals Necessary and Sufficient Conditions

Adams, Alyssa M., Sara I. Walker, Hector Zenil, and P. C. W. Davies. “Quantifying Non-trivial Open-Ended Evolution Reveals Necessary and Sufficient Conditions.”
URL1 URL2

One of the most remarkable features over the last ⇠ 3.5 billion years of life on Earth is the apparent trend of innovation and open-ended growth of complexity. However, this trend
does not have a satisfactory explanation in terms of currently known principles. Here, we demonstrate that a variant of CA are capable of open-ended evolution and innovation by implementing state-dependent dynamical rules. To quantitatively evaluate potential for open-ended evolution and innovation, we present formal definitions of open-ended evolution as patterns that are non-repeating within the expected Poincare re- ´currence time of an equivalent isolated system, and of innovation as trajectories not observed in isolated systems. We show that a small subset of state-dependent systems satisfy both definitions. We compare these systems to a set of controls including CA evolved according to fixed rules and randomly evolved rules and show that state-dependent systems are statistically more reliable at producing both open-ended evolution and innovation. We further show how state-dependent CA allow for sustained growth of complexity, demonstrating that both the complexity and percentage of open-ended cases increases with increasing environment size. Our results indicate that uncovering the principles governing open-ended evolution and innovation in the biosphere will likely require removing the segregation of states and (fixed) dynamic laws characteristic of the physical sciences in attempts to model biological complexity.

Related articles