Biophysically-detailed Recurrent Neural Network Model of the Entire Nervous System of C. elegans

While AI is rapidly developing and demonstrating impressive performance on wide range of tasks, it is unclear that traditional AI models will ever get us to WBE due to the dramatic simplification of biophysical details of neurons compared to the real brain. Incorporating the biophysical details such as ion channel mechanisms, morphology and neurotransmitters will both allow more faithful representation of the nervous system digitally and hold the potential for energy/data-efficient AI in the future. Leveraging the recent breakthrough to train large recurrent biophysical networks ¹, I am building towards the first closed-loop reproducible biophysically detailed nervous system model of C. elegans fitted to real calcium and behavior recording.

Figure 1. Biophysical Network.