South Korean scientists from the Department of Materials Science and Engineering at Pohang University of Science and Technology appear to have cleared the largest obstacle to the feasibility of building brain-like computers: power consumption. In their paper “Organic core-sheath nanowire artificial synapses with femtojoule energy consumption,” published in the June 17th edition of Science Advances, the researchers describe how they use organic nanowire (ONW) to build synaptic transistors (STs) whose power consumption is almost one-tenth of the real thing.
Ever the since the discovery of the role that neurons and their synaptic connections play in cognition, scientists have wanted to create massively parallel computing systems modeled after the human brain. For many decades the best they could do was software emulation: running artificial neural network (ANN) programs on single processor computers. Recently, the rapidly accelerating decrease in both microprocessor price and size has made it possible to run artificial neural networks in a truly parallel fashion, but several factors such as size, shape, and most importantly, power consumption, still hindered the creation of a large scale brain-like computer.
The researchers broke their overarching goal of designing and fabricating a viable artificial synapse down into several smaller goals:
- Choose a material that closely resembles the morphology of the nerve fibers in synaptic junctions.
Organic nanowire (ONW) was chosen because of its shape, flexibility, and scalability. ONWs can form 3D grids of criss-crossing wires, much like the neuronal structures of the human brain.
- Using the chosen material, design a single device that replicates the functionality of a synapse.
In order to maximize plasticity, the researchers fabricated a poly(3-hexylthiophene-2,5-diyl) (P3HT) core nanowire sheathed in polyethylene oxide (PEO).
- In addition to functionality, the artificial synapse should meet or exceed biological neurons in power consumption.
The one-time activation of a single biological synapse uses just 10 femtojoules ( joules) of power. So even though the human brain has somewhere between 100 and 1000 trillion synapses, it still only uses about as much power as a single household lightbulb. Until now, the best synapse-like devices needed pico () and nano () scale levels of power – orders of magnitude higher than their biological counterparts.
In the end, the scientists were not only able to create an artificial synapse that looked and acted like a biological synapse, theirs used just 1.2 femtojoules ( joules) per firing, almost an order of magnitude lower than the real thing!
If you want to dig deeper into the experiments and results, the full paper is available at the Science Advances web site: “Organic core-sheath nanowire artificial synapses with femtojoule energy consumption.”