Dev Kapadia, ’23, Neuroscience, 4/24/20
Figure 1: This is an example of an electrostatic gripper holding two silicon nanowires. Interacting with the environment, these nanowires become reduced (gain electrons) and subsequently deliver electrons to the cell. The use of nanowires was instrumental in the work of the team at University of Massachusetts Amherst, who reduced the voltage necessary for memristor operation to better mimic neuronal activity.
Source: https://commons.wikimedia.org/wiki/File:Microgripper_holding_silicon_nanowires.jpg
Some people think that scientists should study the 95% of the world’s oceans that are unexplored before turning to space. On that point, there is an opportunity for research even closer to us than our oceans.1 There are marvels of the human body that have yet to be discovered, and one of the most fascinating to scientists is the nervous system and the human brain. There is much to be learned about how these systems interact with the rest of the body to enable vision, thoughts, hearing, and even consciousness; these questions have spurred a whole scientific discipline known as “neuromorphic engineering.” Neuromorphic engineering is a highly interdisciplinary practice that integrates mathematics, physics, biology, electrical engineering, and computer science to produce systems that mimic functions of the nervous system.2
One of the many extraordinary features of the brain that neuromorphic engineers have had trouble reproducing are the electrical messages sent between neurons, called “action potentials.” These electrical signals are given off at a fraction of the voltage found in classical computer systems, which are usually used to model action potential and other electrical interactions. Most computers run on a series of 1s and 0s (called bits) that are physically represented by transistors. They essentially function as switches, receiving small currents and generating a larger one depending on whether the bit is in the “on” or “off” position.3 These transistors send and receive voltages using voltages of one or greater, which consequently seemed to be the lower limit for neuromorphic engineers hoping to mimic this memory storage process through a device called “memristors.” Memristors, essentially a neuromorphic engineer’s take on transistors, are biological electrical filaments that mimic transistor function.4 However, researchers at the University of Massachusetts Amherst broke through this voltage barrier by developing a memristor that can send signals ranging from 40-100 millivolts, a range containing the voltage of action potentials between neurons.5
The key to this discovery was the integration of protein nanowires into the construction of memristors. Protein nanowires are found on the surfaces of cells and are used in order to exchange electrons with the environment.6 The researchers, led by Tianda Fu, used nanowires developed by co-author Derek Lovely that were sheared off the bacteria Geobacter sulfurreducens. Because of their metal ion reactivity, common for nanowires interacting with various ions in the human body, the researchers developed a switch by linking a tiny metal thread with the memristor. Because the nanowires were accustomed to conducting electricity, as on-and-off pulses of electricity were delivered to the memristor, the nanowires made new branches and connections. This behavior demonstrates the capacity of the memristors to learn – altering their rate of reduction and therefore increase power efficiency. Branching caused the voltage required by the memristors to function to drop; falling to the low levels that the brain functions at.4
While the team plans to explore the dynamics of protein nanowires in memristors further, they see this discovery as highly motivational for other engineers wanting to develop biological systems with high power-efficiency. The team shattered what was thought to be the theoretical limit of voltage strength in biological systems and now expect far more research to be conducted in the field of neuromorphic systems.4
Bibliography:
[1] Kershner, K. (2020, January 27). Do we really know more about space than the deep ocean? Retrieved April 24, 2020, from https://science.howstuffworks.com/environmental/earth/oceanography/deep-ocean-exploration.htm
[2] Boddhu, S. K., & Gallagher, J. C. (2012). Qualitative Functional Decomposition Analysis of Evolved Neuromorphic Flight Controllers. Applied Computational Intelligence and Soft Computing, 2012 ed., 1–21. https://dio.org/10.1155/2012/705483
[3] Woodford, C. (2019, June 29). How do transistors work? Retrieved April 24, 2020, from https://www.explainthatstuff.com/howtransistorswork.html
[4] University of Massachusetts Amherst. (2020, April 20). Electronics that mimic the human brain in efficient learning: Researchers advance neuromorphic computing. ScienceDaily. Retrieved April 24, 2020 from www.sciencedaily.com/releases/2020/04/200420084249.htm
[5] Fu, T., Liu, X., Gao, H. et al. (2020). Bioinspired bio-voltage memristors. Nat Commun 11, 1861. https://doi.org/10.1038/s41467-020-15759-y
[6] Lovley, D. R., & Walker, D. J. F. (2019). Geobacter protein nanowires. PeerJ Preprints, 7:e27773v1. https://doi.org/10.7287/peerj.preprints.27773