A New Kind of Microchip Mimics the Human Brain in Real Time

A team of scientists in Switzerland has managed to cram 11,011 electrodes onto a single two-millimeter-by-two-millimeter piece of silicon to create a microchip that works just like an actual brain. The best part about this so-called neuromorphic chips? They can feel.

Don't over interpret the word "feel" though. The brain-like microchips built by scientists at the University of Zurich and ETH Zurich are not a sentient beings, but they can carry out complex sensorimoter tasks that show off the network's cognitive abilities. And what's more impressive is that all of this happens in real time. Previous brain-like computer systems have been slower and larger, whereas the Swiss system is comparable to an actual brain in both speed and size. That's exactly what the team was trying to do. "Our goal is to emulate the properties of biological neurons and synapses directly on microchips," says University of Zurich professor Giacomo Indiveri.

The next step for these neuromorphic chips is to take on more and more complex tasks. In a paper published this week by the National Academy of Sciences, the researchers who built the chips suggest that they could connect the neuromorphic chips to sensory systems like an artificial retina. This is somewhat of a fascination for the community of scientists trying to build a brain-like computer. Stanford professor Kwabena Boahen rose to prominence after developing a silicon retina that behaved like a biological retina, and since then, he's been working on ways to mimic the brain using artificial circuits.

Truth be told, it won't be a single scientist or even a team that ultimately accomplishes the task of building a computer that truly behaves like the human brain. The grey matter between our ears is so impossibly complex that so far only supercomputers seem halfway capable of such a feat. And over the next ten years, a whole slew of scientists will attempt to accomplish that task. It will be difficult, and with a budget over $1.6 billion, it will be frightfully expensive. But at least they have some microchips to start with. [Science Daily]

Image via Flickr / Riley Porter