Synthetic Neurocomputers: Advancing Scientific Research through Computing with Living Neurons

Abstract:

Synthetic neurocomputers consist of evolutionarily designed devices—living biological neurons cultured in vitro—and conventional in-silico computing, integrated into an engineered architecture. This forms a unique computational structure for achieving scientific progress. This position paper provides a concise overview of the field's current state, potential opportunities, and strategic directions for future research and development. Key research challenges include advancing culturing and biofabrication techniques, engineering improved biological-silicon interface techniques, and broadly exploring architectural and algorithmic principles. We recommend increasing engagement through open-access and remote-access pathways, and maintaining interdisciplinary collaboration with an emphasis on symbiotic research. We propose that neurocomputers will continue to mature through modular engineering principles, coupled with data-sharing and careful testing, with a particular focus on reliability. Meanwhile, bioethics and social reception should be emphasized as neurocomputers move towards broader commercialization. Neurocomputers should target wide-ranging markets while generating interest in academia, government, and industry. The complex dynamical processes in biological neurons endow synthetic neurocomputers with the potential to tackle a wide range of scientific questions requiring sophisticated information processing, while also providing valuable insights for future computational technologies. Synthetic neurocomputers are poised to drive scientific research and computational technology with their interdisciplinary applications and innovative capabilities, holding great promise for advancing our understanding of biological, computational, and general sciences.

Note:

White paper accepted for inclusion in the U.S. Department of Energy Analog Computing for Science Workshop, September 2024.