The nematode worm, C. elegans, is a star in AI, and now it can exist on your desktop computer. "The worm in the machine" is a headline that's hard to ignore!
We have reached a point where we can create and train neural networks with trillions of connections. However, when it comes to something more biologically accurate, we are limited to a few hundred neurons.
Studying real neural networks requires a lot of computing power. This is because, to provide an accurate simulation, you need to mimic the complex behavior of a neuron. Artificial neurons are simple versions of real biological neurons, which are thought to be equivalent to a small group of artificial neurons. Could it be that real neurons perform tasks that artificial neuron networks cannot? One way to find out is to simulate them and observe the results.
The nematode worm, Caenorhabditis elegans, known as C. elegans to its friends and admirers, is one of the few organisms whose neural wiring we fully understand. It has only 302 neurons, and we know exactly how they connect, making it possible to simulate the entire network. This is the first time a complete network of C. elegans, including its body, has been attempted and achieved.
The latest work includes 92 muscle cells and the neural network to see how the simulation compares to the real worm's behavior. The worm shows easy-to-recognize behaviors like crawling, swimming, and foraging, which can be used to assess how natural the simulation's behavior is.