Neuromorphic Microchips
From: Scientific American - 05/2005 - Vol. 292, No. 5, P. 56
By: Kwabena Boahen

The human brain is superior to the computer in terms of operational
efficiency and functions such as vision, hearing, pattern recognition, and
learning; the key to this superiority appears to lie in the organization of
the brain's neural system, which engineers are attempting to duplicate
electronically. Such a breakthrough could yield implantable electronic
retinas that restore sight or sound processors that restore hearing, as well
as smart visual, audio, or olfactory sensors for robots, writes University of
Pennsylvania bioengineering professor Kwabena Boahen. Power-efficient
microchips patterned after the neural system could form the basis of such
advanced technologies, and University of Pennsylvania researcher Kareem
Zaghloul has created a silicon retina that is 1,000 times less power-hungry
than a PC. His Visio1 chip uses four types of silicon ganglion cells that
mimic the way in which voltage-activated ion channels induce biological
ganglia to discharge spikes. Boahen says hardware customization is common to
both the brain and neuromorphic chips, and morphing the customization
mechanism would make reverse-engineering the brain's circuits unnecessary.
Research into neural development led to the realization that sensory neurons
wire themselves in response to sensory inputs, and accept signals from
neurons that are consistently active when they are active. This process was
morphed, imperfectly, into the Neurotrope1 artificial tectum chip, and
Boahen's team reasoned the system could perhaps be refined through deeper
investigation of cortical connections. Researchers have successfully emulated
in silicon the visual cortex's process of responding preferentially to object
edges of a certain orientation, but Boahen notes that integrated circuits
with many more transistors per unit area are needed if all six cortex layers
are to be morphed.

Links:

Kwabena Boahen
http://perception.upenn.edu/faculty/pages/boahen.php
http://yoda.seas.upenn.edu/boahen/people/boahen/fs_Boahen.htm

Brains in Silicon
http://yoda.seas.upenn.edu/boahen/
