Simulating a virtual worm inside a computer

Let’s say you’re a mad scientist and you want to simulate humans inside a computer and create the most realistic version of The Sims for shits and giggles – first you need to start off with something simpler, like a bug or a worm. This is the goal of the OpenWorm project, which is an international collaborative venture by scientists and programmers, who are trying to see if it is possible to take a bottom-up approach and simulate life, down to the cellular level, inside a computer. This could be the first step in creating artificial virtual human minds.

Simulation of C. elegans in a fluid environment

The human brain has a little less than 90 billion neurons. The African elephant has even more, at about 257 billion. To simulate all those neurons at the same time in an artificial neural network takes immense computational power. Way too much for us to even try running simulations of a human brain. Furthermore, we haven’t yet mapped out our brains. Nerve cells have these things called axons and dendrites, which connect nerve cells to each other to exchange electrical impulses. Mapping all those connections inside a human brain would take a huge effort, something even more monumental than the Human Genome Project (EDIT: I have since come across a project just like that). A simpler organism, let’s say a roundworm, is a much better starting point.

Caenorhabditis elegans is a good choice. This little 1 mm long nematode has only about 959 cells, 302 of them are neurons – a manageable number compared to humans. And what’s even better, the map of all these nerves, known as the connectome, was already completed in the 80s. So far, C. elegans is the only organism to have a completed connectome.

C. elegans

If we have mapped out all the nerve cells of the C. elegans and know how they function, we should be able to simulate this organism inside a computer and there would be no reason why the virtual version wouldn’t think and react exactly like its real-life equal – an idea in AI research called substrate independence (also known in philosophy as multiple realizability), which essentially suggests that there’s no reason a working mind has to be made of living biological cells.

Knowing all these cells and how they’re wired together isn’t enough, however. Connections between nerve cells are called synapses (C. elegans has about 7500 synapses and humans 100 trillion) and the amount one neuron can influence another is called its synaptic weight. And, while we can use modern two-photon calcium imaging to see exactly how neurons physically connect with each other, the technology to see these strengths between neurons is in its early stages. Neurons can also be inhibitory and excitatory and are influenced by special messenger chemicals called neurotransmitters, further compilating things. So, understading how exactly C. elegans neurons work – not to mention simulating them – is no simple task.

Right now, the project is starting out with focusing only on trying to simulate the worm’s 302 nerve and 95 muscle cells and custom software has already been written to accomplish just that.

You can check out the worm’s 3D model in the browser:

Go to OpenWorm, where you can find more information or even donate to the project.