Recent experiments have demonstrated that artificial neurons can effectively trigger responses in real neurons within mouse brain slices. This breakthrough highlights a remarkable compatibility between electronic devices and biological neural systems.
Advancing Brain-Computer Interfaces and Energy-Efficient AI
This innovation brings researchers closer to developing electronics capable of directly interfacing with the nervous system. Potential applications include brain-machine interfaces and neuroprosthetics that could restore hearing, vision, or movement.
The technology also paves the way for a new era of computing systems inspired by the brain's architecture. By mimicking neuronal communication, future hardware could execute complex tasks while consuming significantly less energy. Given that the brain is the most energy-efficient computing system known, scientists are eager to apply its principles to modern technology.
The study is set to be published on April 15 in the journal Nature Nanotechnology.
Mark C. Hersam from Northwestern University, who led the study, noted, "Today's world is increasingly dominated by artificial intelligence (AI). Enhancing AI requires training it on vast amounts of data, which raises significant power-consumption challenges. Thus, developing more efficient hardware to manage big data and AI is crucial. The brain's energy efficiency, being five orders of magnitude greater than that of digital computers, makes it a compelling model for next-gen computing."
Hersam is a leading figure in brain-inspired computing and serves in multiple roles at Northwestern University, including as the Walter P. Murphy Professor of Materials Science and Engineering. He co-led the study with Vinod K. Sangwan, a research associate professor at McCormick.
Understanding the Brain's Superiority Over Silicon
Modern computers manage increasing workloads by integrating billions of identical transistors into rigid, two-dimensional silicon chips. This uniformity results in fixed systems post-manufacturing.
Conversely, the brain operates with diverse neuron types, each fulfilling specialized functions within dynamic, three-dimensional networks that adapt as learning occurs.
"Silicon achieves complexity through billions of identical devices, which are static once fabricated," Hersam explained. "In contrast, the brain is heterogeneous, dynamic, and three-dimensional. To emulate this, we must innovate materials and methods for building electronics."
While previous artificial neurons have been developed, they often produced overly simplistic signals. More complex behavior typically necessitates larger networks of devices, which can escalate energy consumption.
Innovative Materials for Brain-Like Functionality
To better emulate real neural activity, Hersam's team engineered artificial neurons using soft, printable materials that closely align with the brain's structure. Their method employs electronic inks derived from nanoscale flakes of molybdenum disulfide (MoS2) and graphene, deposited onto flexible polymer surfaces via aerosol jet printing.
Rather than viewing the polymer in these inks as a flaw, the researchers utilized it to enhance device performance. By partially decomposing the polymer during the manufacturing process, they created a conductive filament that allows for diverse electrical responses, akin to neuron firing.
Evaluating Interaction with Living Tissue
To assess the artificial neurons' ability to interact with living systems, the research team collaborated with Indira M. Raman, a neurobiology professor at Weinberg. Their findings indicated that the artificial signals matched critical biological properties, effectively activating real neurons and mimicking natural brain activity.
Environmental and Practical Benefits
This innovative approach not only enhances performance but also offers practical and environmental advantages, as the manufacturing process is cost-effective and minimizes waste. As artificial intelligence demands escalate, improving energy efficiency becomes increasingly vital.
In light of these advancements, the future of computing may see a transformative shift towards brain-inspired technologies, potentially revolutionizing the landscape of AI and electronic devices.