In the BayesBrain project, engineers and biomedical researchers at Eindhoven University of Technology will build the world’s first computer that combines brain cells and silicon microchips.
Computers and the microchips that run these machines have brought us great prosperity. We google, app and stream our way around.
However, these computers also use a huge amount of energy, in the form of electricity. The calculations of these processors must be more economical, researchers at the Eindhoven University of Technology believed (together with many other research groups, see what we wrote about it earlier).
They focus on the human brain, which has evolved into an organ that requires relatively little energy to solve complex calculations.
Under the name BayesBrain, they will now build a device where real human brain cells work together with a silicon-based computer. This brain-on-a-chip AI computer – it claims to be the world’s first – should be able to solve problems in real-time and in an energy-efficient way.
Brain cells are difficult to combine with silicon
It sounds quite ambitious because human brain cells are difficult to combine with silicon components. “Silicon is inorganic and a hard material. Brain cells do not like direct contact with this,’ says Regina Luttge, researcher in the Neuro-Nanoscale Engineering chair at Eindhoven University of Technology and one of the leaders of the project together with colleague Bert de Vries (Institute of Electrical Engineering).
To keep alive
The first step in the project is therefore to keep brain cells alive on a chip so that they can form a neural network of around a thousand brain cells. This brain-on-a-chip is then coupled to the silicon part of BayesBrain.
‘The idea is that an algorithm runs on the traditional computer part that performs a calculation. At a certain time, this algorithm sends a signal to the brain network. It performs an operation and then sends a signal back to the silicon,’ says Luttge.
Need a new language
However, that still lies in the future, because the research must first show how the brain network will behave exactly. Luttge: ‘Perhaps the signal goes out, or the biological system will never come to rest. What is also possible is that something happens, but no meaningful signal comes back.’
In addition to the brand new hardware, some researchers on the team will also write the necessary new software. ‘The two separate systems will somehow start talking to each other and communicating. If the network of brain cells emits a certain signal, what does it mean? It requires a new language’, says Luttge.
Healthy network of brain cells
The project is a long-term project. In the short term, Luttge expects to be able to build a healthy, functioning three-dimensional network of brain cells on a microfluidic chip. It requires, for example, channels that supply water and nutrients, otherwise the cells would die. The right temperature and pressure are also necessary for the cells to flourish.
In a few years, BayesBrain is expected to perform some simple calculations. ‘In the longer term, we hope that we can mathematically describe how our system works.’
This means that the research in Eindhoven could not only produce an energy-efficient computer, but also give biologists more insight into how the brain works.
The first PhD student within the BayesBrain project starts on 1 April.
Opening image Shutterstock
If you found this article interesting, you can subscribe to our weekly newsletter for free.