“Memristors” to replace your neurons? Thanks, but no thanks.
Researchers at the Hewlett-Packard laboratories in California have produced tiny electronic switches called memristors (shortening of memory-resistor) that have the potential to revolutionize computing.
Traditional electronic devices use small switches called transistors as the elements of information storage and transfer. A typical computer may have millions of transistors, which may be on the scale of tens of nanometers. Limits in possible reduction of transistor size serve a great threat to progress in integrated circuit design. Memristors – about 3 nanometers in length – therefore offer a new path for making smaller and denser electronic devices.
The team’s report in last week’s issue of Nature shows off the data, with electric traces that are hauntingly reminiscent of neuronal current-voltage plots and action potentials.
The New York Times quotes Dr. Chua, who envisaged memristors in 1971, as saying that “our brains are made of memristors… We have the right stuff now to build real brains.” But are these inglorious transistors really capable of mimicking biological brains?
Simply thinking of the scale differences suggests that the answer may be… maybe. A neuron cell body is on the order of 10-25 micrometers. Compare that to the 3 nanometers of the memristor. Furthermore, memristors operate on a time scale of nanoseconds, whereas most neurons are much slower, spiking in milliseconds.
So memristors are smaller and faster than neurons. In fact, current transistors are also smaller and faster than neurons. So why haven’t computers taken over the world? For one, computers are designed to do what we tell them. And even maverick computers (if they exist) aren’t nearly as smart as the average human. This is because information is transferred in parallel in the brain; and in series in the computer. Put simply: the brain does many things simultaneously, even if slowly, while the computer does only one thing at a time, very quickly. (Curious readers should see “The computer and the brain” by John Von Neumann).
So while memristors may be found inside your next nano-MacBook or iPod-atomic, don’t expect them to replace your neurons.
ORIGINAL NATURE PAPER: http://www.nature.com/nature/journal/v464/n7290/full/nature08940.html#B15
NY TIMES ARTICLE: http://www.nytimes.com/2010/04/08/science/08chips.html?ref=science
– G. Guitchounts
April 2, 2014
I agree with you 100% that neurophysiological data are key in dovleeping theories about the mind. But I think that we need such a theory before we can start doing any kind of useful analysis of neural circuits. In the case of electrical circuits, for example, concepts like voltage and current and theories like Kirchoff’s laws are absolutely vital for meaningful circuit analysis.I think that until we understand what’s fundamentally going on in the interactions between neurons, all the observational data in the world aren’t going to help us in understanding what kind of information processing is happening. This isn’t to say that empirical data aren’t important; rather, an understanding of how to interpret the data is something that needs to be developed.That said, I’m still at the devouring and consuming research phase myself, and plan to be there for a long, long while.