|
|
Neural-Silicon Hybrids Point to New Era in Technology
by James Cavuoto, editor and Warren Grill, senior technical editor
Direct interfaces between small networks of nerve cells and synthetic devices promise to advance our understanding of neuronal function and may yield a new generation of hybrid devices that exploit the computational capacities of biological neural networks. There are several research teams in the U.S. and Europe that are currently working on so-called neural-silicon hybrid chips.
One of the most celebrated researchers in the field is Ted Berger at the Center for Neural Engineering at University of Southern California in Los Angeles. Berger is also a key player in the newly established National Science Foundation Engineering Research Center devoted to biomimetic microelectronics [NBR Jan04 p1].
Berger has set his sights on building artificial neural cells, initially to act as a cortical prosthesis for individuals who have lost brain cells to neurological diseases such as Alzheimer’s. But eventually, his lab’s efforts may usher in a new era in biologically inspired computing and information processing.
Berger’s strategy in building artificial neurons has been an empirical one. Rather than attempt to determine every aspect of how neurons communicate, he’s chosen to emulate their behavior, bombarding live neurons from rat hippocampus tissue with every conceivable type of electrical input, and observe what output emerges from the cell. His team at USC then built a silicon microcircuit that behaves accordingly, at least in terms of spatio-temporal patterns of electrical inputs and outputs. The USC team has built circuits that model 100 neurons; their goal is to construct a 10,000-neuron chip model for implantation in primate hippocampus.
The Max Planck Institute in Germany is another center of research on neural-silicon hybrids. Recently, RA Kaul and P. Fromhertz from the Institute and NI Syed from the University of Calgary reported in Physical Review Letters on direct interfacing between a silicon chip and a biological excitatory synapse. The team constructed a silicon-neuron hybrid circuit by culturing a presynaptic nerve cell atop a capacitor and transistor gate and a postsynaptic nerve cell atop a second transistor gate.
They applied a voltage to the capacitor, which excited the presynaptic neuron, and this activity was recorded with the first transistor. When the presynaptic neuron fired, it generated excitation of the postsynaptic neuron, presumably via an excitatory synapses, and the activity in the postsynaptic neuron was recorded with the second transistor. Further, short trains of activity in the presynaptic neuron appeared to increase the strength of the excitatory synapse between the cells, creating a memory trace within the circuit.
These results demonstrate the ability to use integrated capacitors and transistors to stimulate and record from cultured neurons. The neuron-silicon hybrid provides a tool to study formation and plasticity within small neural circuits and may lead to novel computational devices. |
|
 |
|
|
|
