Magnified image of a C. Elegans worm

Worm's brain replicated in neural network and taught to do tricks

Image credit: Dreamstime

Researchers at Vienna University of Technology have translated the entire neural system of a simple worm into computer code and taught it to balance a pole on the tip of its tail.

It is a simple creature, a roundworm called Caenorhabditis elegans. It is approximately a millimetre long, with no respiratory or circulatory system. It is just about clever enough to eat bacteria and react to certain stimuli, such as touch. It would be unremarkable if it were not the first organism to have its entire genome sequenced and, later, to become the only creature to have its entire neural system described in a “wiring diagram”.

Its neural system consists of just 300 neurons, a tiny number compared with the 100 billion found in a typical human brain. This simple system can be drawn as a sort of circuit diagram, or translated into code, such that the activity of the worm can be simulated almost perfectly by a computer.

As the digital version of this worm takes into account the hard-wired connections between its nerve cells, the simulated worm is capable of reacting to external stimuli in the same way that a real worm would, such as squirming away when touched.

“This reflexive response of such a neural circuit is very similar to the reaction of a control agent balancing a pole,” said Ramin Hasani, a PhD student at Vienna University of Technology, who was involved in a project to “train” the digital worm to balance a pole on the tip of its tail.

This is a standard control problem: a pole is attached to the end of a moving object. When the pole begins to tip, the object should adjust itself to stop it from falling. Hasani and his colleagues experimented with the digital worm, endeavouring find out whether it could perform this task simply by altering the strength of its connections through machine learning.

According to the researchers, the artificial neural network could be trained and optimised to balance the pole on the tip of its tail.

“The result is a controller, which can solve a standard technology problem – stabilising a pole, balanced on its tip,” said Professor Radu Grosu, a computer scientist. “But no human being has written even one line of code for this controller, it just emerged by training a biological nerve system.”

Recent articles

Info Message

Our sites use cookies to support some functionality, and to collect anonymous user data.

Learn more about IET cookies and how to control them