Nanowire networks can exhibit short-term and long-term memory like the human brain
(Nanowerk News) An international team led by scientists at the University of Sydney have demonstrated nanowire networks can exhibit short-term and long-term memory like the human brain.
This research has been published today in the journal Science Advances (“Neuromorphic learning, working memory and metaplasticity in nanowire networks”), led by Dr Alon Loeffler, who received his PhD in the School of Physics, with collaborators in Japan.
“In this study we found that higher-order cognitive functions, which we normally associate with the human brain, can be replicated in non-biological hardware,” said Dr Loeffler.
“This work builds on our previous research in which we demonstrated how nanotechnology can be used to build brain-inspired electrical devices with neural network-like circuits and signaling like synapses.
“Our current work paves the way towards replicating brain-like learning and memory in non-biological hardware systems and suggests that the traits underlying brain-like intelligence may be physical in nature.”
A nanowire network is a type of nanotechnology typically made of tiny, highly conductive silver wires invisible to the naked eye, covered in a plastic material, which are spread over one another like a net. The wires mimic aspects of the physical structure of human brain tissue.
Advances in nanowire networks could mark many real-world applications, such as enhanced robotics or sensory devices that need to make quick decisions in unpredictable environments.
“These nanowire networks are like synthetic neural networks because the nanowires act like neurons, and the places where they connect to each other are akin to synapses,” said senior author Professor Zdenka Kuncic, from the School of Physics.
“Instead of implementing some sort of machine learning task, in this study Dr Loeffler has actually taken it one step further and tried to show that nanowire networks exhibit some sort of cognitive function.”
To test the networking capabilities of the nanowires, the researchers gave them a test similar to a common memory task used in human psychology experiments, called the N-Back task.
For one person, the N-Back task might involve recalling a particular cat picture from a series of cat pictures presented in sequence. An N-Back score of 7, averaged across people, indicates that the person can recognize the same image that appears seven steps back.
When applied to a network of nanowires, the researchers found it could ‘remember’ the desired end point in an electrical circuit seven steps back, meaning a score of 7 in the N-Back test.
“What we’re doing here is manipulating the tip electrode voltage to force the pathway to change, rather than letting the tissue do its own thing. We force the path to go where we want,” says Dr Loeffler.
“When we implemented it, the memory had much higher accuracy and didn’t really diminish over time, indicating that we had found a way to strengthen the pathway to push it where we wanted it, and then the network remembered it.
“Neurologists think it’s how the brain works, certain synaptic connections strengthen while others weaken, and it’s thought to be how we remember things, how we learn, and so on.”
The researchers say as the nanowire network continues to be strengthened, it reaches a point where that reinforcement is no longer needed as the information is consolidated into memory.
“It’s like the difference between long-term memory and short-term memory in our brains,” said Professor Kuncic.
“If we want to remember something for a long period of time, we really need to keep training our brains to consolidate it, otherwise it will fade over time.
“One task demonstrated that nanowire networks can retain up to seven items in memory at a much higher rate than the odds without reinforcement training and near-perfect accuracy with reinforcement training.”