Memristors are the key to creating a cybernetic brain
When the “electronics missing link” was finally built in 2008, it justified the 30-year prediction. Now, it looks like a so-called memristor can behave in the same way as junctions between neurons in the brain.
A memristor is a device that, like a resistance, resists the passage of current. But memristors also have memory. The resistance of a memristor at any moment depends on the last voltage it experienced, so its behavior can be variable.
Memristors are now being used in US military projects that are trying to build a computer brain, says Wei Lu, team leader at Michigan State University at Ann Arbor.
The existence of the memristor was predicted in 1971 when Leon Chua of the University of California at Berkeley discovered a gap in the capabilities of basic electrical components. It wasn’t until 2008 that Stanley Williams of Hewlett-Packard Labs in Palo Alto, California made the first titanium dioxide memristor.
Lu and his colleagues have already presented the first demonstrations that it is possible to create an electronic version of the brain. What’s more, their memristors were built with materials already used in the manufacture of computer chips.
Lu’s team used a mixture of silicon and silver joined by two metal electrodes at the intersection. The transition mimics the behavior of the synapses of neurons, which allows you to master new models and, presumably, store memories.
In the brain, electrical signals in two neurons affect the ease with which late messages can jump across the synapse between them. If the pair acts in sync, the synapse has a better chance of subsequent messages. “The cells that send momentum together are linked together,” Lou says.
Lu’s device mimics the behavior of neurons. When the gap between the signals from the two electrodes is 20 milliseconds, the resistance of the current flowing between them is about half that after the signals are separated by 40 milliseconds. “The memristor mimics synaptic actions,” Lu said, adding that the next step would be to create circuits with tens of thousands of memristor synapses.
Williams is pleased that the long-predicted work is having results. “I am happy to see that our work is promising,” he says.