Open-air lecture on electronics and the brain attracts a crowd in King’s Cross
18 July 2016
Artificial intelligence, supercomputers and the neurons of a shellfish were some of the topics touched on at the IOP’s latest Summer Session held beside the canal in King’s Cross, London, on 13 July.
The open-air lecture attracted an audience of around 150 on the Canalside Steps, as well as being heard by passing cyclists, boat crews and many interested members of the public who stopped to listen.
Science troubadour Johnny Berliner opened the evening with physics-themed songs about dark matter, the possibility of intelligent machines, Newton’s three laws, and the twin paradox – all given a comic twist but firmly rooted in scientific fact. He then took up a ukulele to sing a George-Formby-styled number about neuroscience, “Nobody can tell you how the brain works”.
This was all the warm-up for the talk by Dr Jessamyn Fairfield of Trinity College Dublin, Electronics on the Brain, about building networks from nanowires that can imitate aspects of the way the brain learns. Researchers are using the properties of 1D nanowires to create memristors – junctions that “remember” the charge that has passed through them by changing their resistance – and explore their potential applications.
Fairfield explained that the human brain has vast numbers of synapses that start off by being randomly connected but that the ones that are used most are strengthened as learning takes place. She was working at the level of just a few junctions – fewer than the number of synapses that were found in a shellfish, she said. The challenge was to scale this up.
The human brain runs on only about 30 W of power but supercomputers present huge problems with overheating and the smaller the components are made, the hotter they become, she said. However, unlike the brain, nanowire junctions can potentially be operated using pulses of light, electricity or chemicals, she said.
During a Q&A session Fairfield was asked whether redundant nanowire connections could be pruned. Unlike in the brain, where unused synapses disappear, such connections would remain unless they were removed individually, which would be too much work, she said.
Fairfield said it was a really exciting field of research. “This is a field that’s only been active for about the last five years. There are a lot of people working on the software side but not so many on the hardware side so the field is wide open,” she said.