Team of Researchers Brainstorming to Develop New Electronic Devices

Scientists at Boston University say they have made significant headway toward developing a new generation of high-fidelity instruments through what they call “biologically inspired electronics.”

Physicist Douglas Mar, one of the leaders of an interdisciplinary team of about 50 people working on the project, said they didn’t have to look far for the biological model they want to replicate in electronic circuits: the human brain.

That’s a tall order, because the brain is an extremely complex organism that is still not well understood. Scientists at Caltech have pioneered in the effort to translate human biological systems into electronic devices. They call it “neuromorphic engineering.”

Boston University got into the act when Robert Adams, manager of audio development at Analog Devices Inc. of Norwood, Mass., came to the scientists with an idea.

“Bob is a guru in the audio electronics community, and he had this weird idea that maybe neurons [in the brain] used this special technique called ‘noise shaping,’ ” Mar said. It’s used in compact disc players to give “very high resolution with fairly crummy parts,” and Adams thought that perhaps the brain does the same thing.

It may have seemed weird at the time, but the more Mar thought about it, the more sense it made.

That’s based partly on the fact that the human brain has a stunning capacity to screen out background noise from what it wants to see or hear.

“The brain does some really amazing things for you,” Mar said. “You can pick out the sound of my voice in a crowded room, you can pick out the face of someone you know from an extremely cluttered scene. That kind of processing is very hard to do with a computer.”

But then there’s the problem of people such as Sammy Sosa and Mark McGwire.

The brain’s neurons are biological equivalents of transistors, yet they “fire” slowly in comparison with many of the signals they need to encode. Mar notes that a baseball is in the “hitting zone” for only a few milliseconds, whereas individual neurons often take 10 milliseconds or more between firings.

If that’s the case, the baseball should be in the catcher’s mitt before McGwire even begins his awesome swing.

So how does he do it?

Neurons are pulse-conducting cells, and the human body has more than 10 billion--mostly in the brain. Research shows that the neurons are interconnected, so one neuron can influence the behavior of hundreds of others. That suggests that the number of interconnections is huge, maybe as high as a trillion.

Adams’ idea was that maybe the neurons in the brain work in such a way as to suppress neurons that might be getting in the way when Sosa swings his bat, thus simplifying the task of determining when the ball is in the right place to be belted over the fence.

And if that’s how the brain works--and the jury is still out on that--couldn’t it also work as a hi-fi system?

The Boston team, which includes electrical engineers, biologists and neuroscientists, decided to give it a try. They reported the results of their research in the Aug. 31 issue of the Proceedings of the National Academy of Sciences.

They created a mathematical model of a neuronal network that works by “inhibitory coupling,” in which one neuron momentarily suppresses the level of activity in other neurons in the network.

“Not only does it work, but it seems to work really well,” Mar said.

The next step will be to see if they can translate their model into a functioning electronic device. Details are sketchy. Analog Devices is funding the research, but Mar offered a few insights.

*

If they can make it work, they should be able to build a high-fidelity system, for example, that produces exquisite sounds even though many of its components may be “crummy.”

By mimicking what they think may be going on in the human brain, individual components will suppress those not needed for the immediate task, thus screening out background noise from, say, the sound of the fiddle.

“Where we might be helpful is in connecting circuits that are not high performance, and trying to improve one or more properties of those circuits” through noise shaping, Mar said.

Unfortunately, they probably won’t be able to build a “biologically inspired” electronic system that does everything the brain does.

Neurons in the brain die. Some of them come back from the dead, and some of them don’t, yet the brain continues to function.

In electronics, if a transistor goes belly up, the whole system can collapse.

So the human brain is likely to remain in a class by itself for a long, long time.

*

Lee Dye can be reached at leedye@ptialaska.net.