Network size required for recurrent neural chemical reaction networks (RNCRNs) to realize full cellular behavior

Determine the minimal number of species or nodes required for recurrent neural chemical reaction networks to compute the full repertoire of cellular behavior, and specify how required size scales with functional demands.

Background

The authors discuss recent work mapping artificial recurrent neural networks onto chemical reaction networks (RNCRNs), which can approximate arbitrary dynamical systems. While small RNCRNs have been demonstrated, it is uncertain how large such networks must be to reproduce the complex, time-dependent behaviors characteristic of living cells.

Quantifying the size threshold would bridge computational expressivity with biochemical feasibility, informing whether lifelike computation can plausibly emerge in prebiotic chemical networks.