Reproduce “Integrated information increases with fitness in the evolution of animats” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jeffrey Edlund, Nicolas Chaumont, Arend Hintze, Christof Koch, Giulio Tononi, Christoph Adami
Edlund et al. empirically demonstrate that measures of integrated information, particularly the integrated information of the proper main complex (Φ_MC), correlate strongly with evolutionary fitness in artificial agents that develop memory-dependent navigation strategies. This work validates information integration as a robust quantifier of functional complexity in adaptive systems.
Key insights
- Information integration measures, particularly Φ_MC, serve as superior indicators of functional complexity compared to purely predictive or total processing measures when tasks necessitate internal memory and context-dependent decision-making.
- Evolutionary fitness in controlled simulations provides a robust empirical arbiter for evaluating and validating theoretical information-theoretic measures of functional complexity in adaptive systems.
- The successful evolution of memory mechanisms in animats correlated with a notable increase across all measured complexity metrics, highlighting the functional importance of integrating past information.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.