Tuesday 29 August 2006

Wired

From the Harvard University Gazette :
Opening a whole new interface between nanotechnology and neuroscience, scientists at Harvard University have used slender silicon nanowires to detect, stimulate, and inhibit nerve signals along the axons and dendrites of live mammalian neurons.

Harvard chemist Charles M. Lieber and colleagues report on this marriage of nanowires and neurons this week in the journal Science.

"We describe the first artificial synapses between nanoelectronic devices and individual mammalian neurons, and also the first linking of a solid-state device -- a nanowire transistor -- to the neuronal projections that interconnect and carry information in the brain," says Lieber, the Mark Hyman Jr. Professor of Chemistry in Harvard's Faculty of Arts and Sciences and Division of Engineering and Applied Sciences. "These extremely local devices can detect, stimulate, and inhibit propagation of neuronal signals with a spatial resolution unmatched by existing techniques."

Electrophysiological measurements of brain activity play an important role in understanding signal propagation through individual neurons and neuronal networks, but existing technologies are relatively crude: Micropipette electrodes poked into cells are invasive and harmful, and microfabricated electrode arrays are too bulky to detect activity at the level of individual axons and dendrites, the neuronal projections responsible for electrical signal propagation and interneuron communication.

By contrast, the tiny nanowire transistors developed by Lieber and colleagues gently touch a neuronal projection to form a hybrid synapse, making them noninvasive, and are thousands of times smaller than the electronics now used to measure brain activity.
...
The group's latest work takes advantage of the size similarities between ultra-fine silicon nanowires and the axons and dendrites projecting from nerve cells: Nanowires, like neuronal offshoots, are just tens of nanometers in width, making the thin filaments a good match for intercepting nerve signals.

Because the nanowires are so slight -- their contact with a neuron is no more than 20 millionths of a meter in length -- Lieber and colleagues were able to measure and manipulate electrical conductance at as many as 50 locations along a single axon.

This has very important implications for neural-external interfaces. If such precision can be applied to brains in situ rather than single cells in a laboratory, we should be able to learn an awful lot more about our thought processes.

OK, there's a long way to go, maybe a century or two before we can manage that. But probably only a decade or so before prosthetic devices, not just manipulators but sensors, become true replacements for missing limbs.

It also allow the possibility of more capable hybrid circuitry, involving both biological and non-biological components. But, as I've blogged about before, there are definite ethical issues here when using brain tissue from higher animals. The danger that the unthinking machines may not be unfeeling ones.