Evolutionary Implications of Self-Assembling Cybernetic Materials with Collective Problem-Solving Intelligence at...

Area of Science:

• Developmental biology
• Evolutionary biology
• Computational neuroscience

Background:

• Biology features a complex multi-scale competency architecture (MCA) with nested homeostatic agents.
• How natural selection creates MCAs is a key research question.
• This study examines MCA's impact on evolution, not its origin.

Purpose of the Study:

• Investigate how decision-making competencies of MCA agents affect evolution.
• Model morphogenesis using neural cellular automata (NCAs) and evolutionary algorithms.
• Analyze the effect of varying cellular agent regulation accuracy on evolution.

Main Methods:

• Used in silico neuroevolution experiments with simulated minimal developmental biology.
• Modeled morphogenesis with neural cellular automata (NCAs).
• Employed an evolutionary algorithm to optimize NCA parameters for self-assembling a 2D pattern.

Main Results:

• Evolutionary processes were significantly faster when optimizing MCA functional parameters versus direct pattern evolution.
• Evolved MCAs demonstrated strong generalization to system parameter changes and modified objectives.
• Agent competency levels were scaled from direct encoding to MCA.

Conclusions:

• Adaptive problem-solving competencies of agential parts in NCAs significantly influence the evolutionary process.
• The findings suggest functional implications of near-ubiquitous competency in living matter.
• MCA facilitates more rapid and robust evolutionary adaptation.