Skip to Content

Synaptic Behaviour Captured By New Memristor Circuit Design

When it comes to copying the real behaviour of synapses, two memristors are better than one, according to a new circuit design.

Since the 1970s, electronic engineers have known that there are four fundamental building blocks of electronic circuits: resistors, capacitors, inductors and memristors (essentially variable resistors with memory). Memristors, however, had an air of mythology about them until last year when a group of researchers at HP Labs in California announced they had discovered them for the first time.

Since then, numerous others have claimed to have played with memristance over the years (although none seem to have noticed what they were doing until now). In fact, it turns out that the synapses between neurons behave exactly like memristors. That raises the possibility that memristors can be connected together in a way that truly mimics the wiring of human brains.

One of the defining features of the connections between neurons is that they become stronger when neurons fire together; hence the phrase “neurons that fire together, wire together”, a phenomenon otherwise known as Hebbian learning. Various experiments have shown that this effect is most pronounced early in the learning process, when the increase in connection strength is greatest. Later learning merely reinforces the links

That’s somewhat at odds with the actual behaviour of memristors, say Farnood Merrikh-Bayat and Saeed Bagheri at the University of Tehran in Iran. They say that in a single memristor connecting two neurons, the memristance decreases when a voltage is applied which increases the current which in turn causes the memristance to drop further, in a kind of positive feedback effect.

A lower memristance allows more current to flow so this certainly increases the strength of the connection as expected but there’s a problem. The positive feedback effect means that later signals have a bigger effect on the connection than earlier ones, which is the opposite way round to the way real neurons connect, where earlier signals have the strongest effect.

Merrikh-Bayat and Bagheri have a simple solution: use two memristors in series. Choosing their memristance carefully allows them to reproduce Hebbian-type synapse strengthening more or less exactly.

That may turn out to be a useful insight. The first neuromorphic chips to use memristance to mimic synapse behaviour are already being built. A small change in their design may make a significant difference.

Ref: arxiv.org/abs/1008.3450: Bottleneck Of Using Single Memristor As A Synapse And Its Solution

Keep Reading

Most Popular

Rendering of Waterfront Toronto project
Rendering of Waterfront Toronto project

Toronto wants to kill the smart city forever

The city wants to get right what Sidewalk Labs got so wrong.

Muhammad bin Salman funds anti-aging research
Muhammad bin Salman funds anti-aging research

Saudi Arabia plans to spend $1 billion a year discovering treatments to slow aging

The oil kingdom fears that its population is aging at an accelerated rate and hopes to test drugs to reverse the problem. First up might be the diabetes drug metformin.

Yann LeCun
Yann LeCun

Yann LeCun has a bold new vision for the future of AI

One of the godfathers of deep learning pulls together old ideas to sketch out a fresh path for AI, but raises as many questions as he answers.

images created by Google Imagen
images created by Google Imagen

The dark secret behind those cute AI-generated animal images

Google Brain has revealed its own image-making AI, called Imagen. But don't expect to see anything that isn't wholesome.

Stay connected

Illustration by Rose WongIllustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.