Professor Megumi Akai-Kasaya, Graduate School of Science

"Beyond science fiction: A step toward the realization of brain-inspired computers"

“I adore the world of science fiction.”

I have loved books since I was a child and read many different genres, but one genre I absolutely adore is science fiction. One science fiction novel I came across when I was in high school, "Solar Wind Intersection" by Akira Hori, changed my world. I was completely fascinated by the world of "hard science fiction," a sub-genre of science fiction with a strong scientific nature based on principles of physics and technical terms. This led me to study physics at university and eventually continue to graduate school, where I worked on surface and interface properties. Now I am a researcher in the fields of materials science and information science, working on the development of neuromorphic devices using conductive polymer wires in order to bring out functions that have not yet been utilized.
I began this work several years ago after meeting a researcher in the field of information science at a materials science conference and hearing him speak about AI. What really struck me was when he said, "We don't yet have a wire that can extend in three dimensions freely through space." This immediately made me think of the conductive polymers I was working on, and when I introduced my material to him, he encouraged me to try to make it happen. That was when it all began.
Today, most of what is called “artificial intelligence” is based on a mechanism called a "neural network" that has been developed to mimic the structure of the human brain. This is run by software in a computer, and what it is really doing is numerical computations. A real brain works differently. In the brain, neurons extend their axons freely in three dimensions, form synapses at contact points with other neurons, and transmit the next neuron information through the synapse. I wanted to replicate this with materials to try to develop a real AI with extending wires in a 3D environment that acts like an actual brain.

Building a 3D conductive polymer network capable of spike-based information processing

A few years ago, we succeeded in developing a conductive polymer that first realized synaptic plasticity in two dimensions [1]. These conductive polymers allow us to control doping and de-doping by changing the electrical conductivity in solution, thereby enabling the use of long-term memory and short-term memory.
Our recent paper is the result of our efforts over the past 10 years or so and we succeeded to build a 3D conductive polymer network capable of spike-based information processing [2]. In the brain, neurons, which are composed of organic materials, grow in 3D, repeatedly destroying and repairing each other and transmitting information via ions. While other studies of synaptic devices, such as devices made by microfabrication of semiconductors, have used pre-fabricated materials, our conductive polymers can freely develop. Polymers have spatial mobility and grow in a directional manner that inorganic materials do not, which may mean that they are the only materials that have the potential to artificially recreate the three-dimensional, ultra-complex network of the brain by design. Thus our goal is not to make a 3D network by connecting two dimensions network, but rather to create a 3D network by growing wires from scratch. Thanks to the passion and efforts of the students and collaborators who worked on this project, we were able to demonstrate that it is possible to grow conductive polymer wires in three dimensions by creating three-dimensional electrodes with sharp tips and using them as lightning rods to concentrate the electric field on the tips. The three-dimensionality also made it possible to control the resistance between the three-dimensionally wired electrodes and develop a technique for learning 3D polymer network circuits, thereby successfully imparting associative memories, such as “banana” and “yellow,” to the network circuits.

Pioneering a future straight out of science fiction through her research

As I was watching the growth of conductive polymers under the microscope every day, I had a feeling of déjà vu, like I had seen this somewhere before. Then one day, it hit me: it looked just like the growing brain computer from Star Trek Voyager, which I used to watch on late-night TV. The approach developed in our study is a big step toward the realization of neuromorphic wetware and closing the gap between the cognitive abilities of humans and computers. My dream is for these polymers to be regarded as a miniature artificial brain, moving us forward to a future in which AI will accompany us in our daily lives as a "partner," rather than just a "tool.” I also hope that my research will pioneer a new field of science that encompasses various computational sciences and chemistry. Nature has much wisdom to impart to us, and I’d like to learn from this wisdom to push my research to the next level.

What would a "future society where life shines brightly" look like for you?

I believe that the future is here and now. There are many people, especially in Japan, who do not see the future as one that is bright. I was just a baby when the Osaka Expo was going on in 1970, and back then we saw such a bright future. I hope we can show the world the same kind of exciting future this time as well.

[1] Hagiwara N., Sekizaki S., Kuwahara Y., Asai T., and Akai-Kasaya M., "Long- and short-term conductance control of artificial polymer wire synapses," Polymers, vol. 13, no. 2, pp. 312(1)-(10) (2021). Akai-Kasaya M., Hagiwara N., Hikita W., Okada M., Sugito Y., Kuwahara Y., and Asai T., "Evolving conductive polymer neural networks on wetware," Japanese Journal of Applied Physics, vol. 59, no. 5, pp. 060601(1)-(9) (2020).

[2] Naruki Hagiwara, Tetsuya Asai, Kota Ando, Megumi Akai-Kasaya (2023). Fabrication and Training of 3D Conductive Polymer Networks for Neuromorphic Wetware, Advanced Functional Materials. DOI: https://doi.org/10.1002/adfm.202300903.
https://resou.osaka-u.ac.jp/en/research/2023/20230701_1

For more information.
https://researchmap.jp/Meg_Aki_Kassy?lang=en
http://www.chem.sci.osaka-u.ac.jp/lab/akai/englishindex.html

Text: Mayumi Mochizuki/Edit: Christopher Bubb