Overview: A Leap Toward Brain-like Computing
Scientists from the University of Massachusetts Amherst have unveiled a groundbreaking artificial neuron designed to closely mimic the electrical activity of natural brain cells. Building on earlier work with conducting filaments derived from electricity-producing bacteria, the researchers have developed a neuron that operates at voltages comparable to biological neurons and with dramatically reduced power consumption. This innovation promises new pathways for computers that emulate living systems, potentially enabling seamless interfaces with biological tissue and more energy-efficient devices.
In the human brain, electrical performance far outpaces typical computer circuits—by more than 100 times in certain aspects. For a task as routine as writing a story, the human brain uses roughly 20 watts of power, while large language models running on conventional hardware can require more than a megawatt. The challenge has been to design artificial neurons that can operate at these biological voltages and interact directly with living neurons without causing damage or excessive heat.
From High Voltage to Low Voltage: A Fundamental Shift
Associate Professor Jun Yao of electrical and computer engineering at UMass Amherst explains a key achievement: earlier versions of artificial neurons demanded voltages up to 10 times higher and consumed far more power than the new design. The team’s latest neuron registers about 0.1 volts, aligning with the natural voltage range of human neurons. This low-voltage operation is not merely a power-saving measure; it is essential for direct, safe communication with biological tissue and for creating compact, wearable, or implantable devices.
Why Low Voltage Matters
Lower voltages reduce heat generation and improve compatibility with living tissue. This makes it feasible to develop neuromorphic systems that can interface with the nervous system or with bio-sensing platforms without requiring aggressive voltage amplification stages. The reduction in amplification steps, in turn, cuts power consumption and simplifies circuit design, leading to smaller, more robust devices that can operate continuously in real-world environments.
The Role of Protein Nanowires
A distinctive feature of the new artificial neuron is the use of protein nanowires derived from Geobacter sulfurreducens, a bacterium known for its natural electricity-generating capabilities. These biological-inspired nanowires act as efficient conduits for electrical signals within the artificial neuron and potentially across neuromorphic networks. The researchers leveraged the bacteria’s protein nanowires to construct devices that can mimic the way real neurons propagate signals, contributing to a more faithful, energy-efficient computational neuron.
The approach marks a shift from synthetic materials to biology-informed components that can be produced at scale with precision. The protein nanowires provide a stable, conductive backbone that supports reliable signal transduction at ultra-low voltages, aligning artificial systems more closely with the behavior of living neurons. This alignment is crucial for long-term integration with bio-electronic interfaces and for exploring new forms of human-machine collaboration.
Potential Applications: From Bio-Inspired Computers to Wearable Tech
The broader implications of this technology span several domains. In computing, ultra-low voltage neurons could enable energy-efficient neuromorphic chips that emulate cognitive processes with significantly lower power demands. In medicine and wearable technology, these neurons open doors to seamless interfaces with the human body—sensors and therapeutic devices that can monitor, respond to, and collaborate with biological systems without bulky power requirements.
Researchers note that current wearable sensing systems face inefficiencies largely due to amplification of body signals before processing. With low-voltage artificial neurons, sensors could analyze raw biological data directly, reducing both power usage and circuit complexity. This could lead to wearable health monitors that last longer between charges and respond more quickly to physiological changes.
Future Outlook: Toward Real-World Integration
While the work is still in a research stage, the development of ultra-low voltage neurons that can interact with living tissue represents an important step toward practical bio-compatible neuromorphic systems. The team envisions further refinements that expand compatibility with diverse biological signals and enable scalable networks of artificial neurons that operate with brain-like efficiency. The ultimate aim is to create computing architectures that parallel the efficiency of natural neural networks while maintaining the precision and reliability required for modern applications.
