In a paper entitled, An Evolvable Organic Electrochemical Transistor for Neuromorphic Applications, lead author Jennifer Gerasimov, from the Laboratory of Organic Electronics at the University, along with four colleagues, describes the creation of an electropolymerized transistor that can alter its conductance and structure resembling the neuroplasticity of neurons seen in the human brain. They describe their invention as “a new paradigm of machine learning based on evolvable organic electronics.”

Organic electrochemical transistors (OECTs) are seen as important additions to the field of neuromorphic computing. Instead of prefabricated circuitry which requires adjustment and supervision, OECTs have the ability to make new connections in a circuit where none existed before.

In their paper the authors describe how they can incorporate OECTs into a learning circuit to take advantage of their capacity to make new or strengthen existing electronic connections. In this way, the OECT becomes a synaptic device communicating with other circuitry and acting identically to our biological synapses that transmit information between nerve cells.

The use of OECTs with neuromorphic chip technology has the potential to bring the advantages of the organized chaos of our human brains to computing. A neuromorphic system is seen to have significant advantages over existing computing technology because it thinks in ways similar to us. Described as fuzzy logic, neuromorphic computers arrive at solutions through the exploration of multiple scenarios, much the way our brains work.

The OECT was built on a silicon substrate, marrying organic chemistry to more familiar computing materials. And one can imagine chips and circuits with OECTs in the near future becoming the central nervous system of robotic systems that learn through observation requiring little in the way of instruction or training.

OECTs, as described above, can alter their structure allowing them to fulfill different form factor requirements so that computing technology can be put to use in many more places than at present.

In Star Trek, bio-neural gel packs were described as operating by making best guess answers to complex problems, including ones related to space-time temporal anomalies, a favourite theme of the science fiction series. The Star Trek biological circuits could catch a cold. In one episode bio-neural gel packs became infected when a virus from a cheese being made on board contaminated the circuits. The cure for this cold was rather novel. Rather than swapping out infected circuits, they were heated to kill the infection.

And finally, using OECTs may make it possible to create computers that can run on as little energy as what it takes to operate our brains, a mere 15 watts. It seems that in developing the OECT technology that may very well be possible. Simone Fabiano, one of the four authors of the paper, is quoted stating, “Our organic electrochemical transistor can… carry out the work of thousands of normal transistors with…energy consumption that approaches the energy consumed when a human brain transmits signals between two cells.” 

Sounds very promising.

What is the living function of ATP?

The “energy currency of life” according to biologists. ATP stores energy in our cells. Cells draw from the stored energy in ATP. When food is oxidized the ATP is replenished. And where do you find ATP? In the mitochondria which many biologists theorize is an organelle that in Earth’s past lived separately from modern cells but somehow was absorbed to become an essential energy generating machine in animal cells today.

Why a biological computer?

Because conventional electronic computers do problem solving sequentially. When you need more computing power than a single silicon processor, you put a whole bunch of them together to tackle a problem in parallel. That’s how IBM built Watson which in its Jeopardy iteration contained 2,880 processors. Computers of this size are not as energy intensive as the supercomputers used for climate modeling and the weather service, but compared to a biological computer, they are energy hogs. A biological computer gets its power from molecules in living cells. You can pack a lot of ATP into a small amount of physical space and as the writers of the paper appearing in Proceedings state, “this approach uses orders of magnitude less energy than conventional computers, thus addressing issues related to power consumption and heat dissipation.”

The components to build a biological computer are relatively cheap. ATP can be had at a negligible cost. Its energy requirements are self-contained. The form factor is so small you can pack a lot of computing power into something book-sized. And it solves  computational problems in parallel just like IBM Watson.

How does it compare to quantum computers?

Instead of qubits, the biological computer uses proteins powered by ATP. The researchers describe the physical appearance of a circuit as looking “a bit like a road map of a busy and very organized city as seen from a plane.”

How much energy does a biological computer consume?

Less than 1% of the energy used by an electronic transistor. Building supercomputers from biological circuits would contribute mightily to energy conservation, an important issue in the fight to reduce our energy dependence on fossil fuels.

How do you build a biological computer?

The researchers manufactured their device using electron-beam lithography at a nanoscale with the material coming from bovine serum albumin (a cow). They experimented with actin filaments and micro tubules and found that computational error rates were far lower with the former than the latter. They accounted for the computational errors because of fabrication problems caused by not working in a clean room environment. The plan going forward is to incorporate the highest standards of clean room facility for future fabrication.

What are the next challenges?

As in most new technology experiments coming from a laboratory, scaling the device is the remaining challenge. To compete with  electronic computers the developers will have to replicate the design on mass or have the actin filaments self replicate. In other words teach the circuit to build itself. All this is doable so it shouldn’t be too long before we see biological-based circuits and possibly a hybrid that blends biology and silicon-based transistors.

Is life imitating art with the publishing of a paper on biological computing?

Absolutely. Where have you seen biological computers in the past? Look no further than an episode of Star Trek Voyager where the ship’s bio-neural technology (see the blue gel pack in the picture below) becomes infected and an induced fever is used to cure it. USS Voyager‘s science-fiction based organic computer consisted of massive number of neural fibers similar to the actin filaments of our real biological computer. The power source isn’t mentioned although we know Star Trek used antimatter and warp technology. But life indeed is eerily resembling art.