Researchers at the University of Massachusetts Amherst have developed a revolutionary artificial neuron that closely mimics the behavior of biological neurons. It can fire, learn, and respond to chemical signals, offering potential advancements in computing, medicine, and the integration of technology with biological systems.
Neurons play a crucial role in complex processes such as thought, emotion, and movement, communicating through electrical and chemical signals. The new artificial neuron not only imitates the behavior of its biological counterparts but also matches their size, energy requirements, signal strength, timing, and responsiveness to chemical stimuli.
Shuai Fu, the lead author of the study and a graduate student in electrical and computer engineering at UMass Amherst, highlighted the efficiency of biological brains. “Our brain processes an enormous amount of data,” he stated. “But its power usage is very, very low, especially compared to the amount of electricity it takes to run a Large Language Model, like ChatGPT.”
Innovative Design and Functionality
The team constructed their artificial neuron using a memristor, a type of memory resistor made from protein nanowires derived from the microbe Geobacter sulfurreducens. These conductive nanowires allow the memristor to operate at extremely low voltages (approximately 60 mV) and minimal currents (around 1.7 nA), closely resembling the performance of biological neurons.
Jun Yao, PhD, an associate professor of electrical and computer engineering at UMass Amherst and the study’s corresponding author, explained, “Previous versions of artificial neurons used 10 times more voltage – and 100 times more power – than the one we have created. Ours registers only 0.1 volts, which is about the same as the neurons in our bodies.”
To replicate the electrical activity of a neuron, the researchers integrated the memristor into a basic resistor-capacitor (RC) circuit. This setup allows the artificial neuron to undergo the different phases of electrical activity—charge integration, rapid depolarization, and repolarization. It also includes a refractory period, mirroring the behavior of biological neurons after firing.
The team further enhanced the neuron by adding chemical sensors capable of detecting ions like sodium and neurotransmitters such as dopamine. These sensors adjust the circuit’s electrical properties in reaction to chemical signals, simulating the neuromodulation process observed in real neurons.
Potential Applications in Bioelectronics
In a significant step towards integrating this technology with living systems, the researchers connected the artificial neuron to actual human heart cells, known as cardiomyocytes. They successfully demonstrated the neuron’s ability to interpret biological signals in real time, including changes in cardiomyocyte activity when exposed to the drug norepinephrine.
Yao remarked, “We currently have all kinds of wearable electronic sensing systems, but they are comparatively clunky and inefficient. Every time they sense a signal from our body, they have to electrically amplify it so that a computer can analyze it. That intermediate step of amplification increases both power consumption and the circuit’s complexity. Sensors built with our low-voltage neurons could do this without any amplification at all.”
While this discovery represents an important advancement in bioelectronics, it is still in the early prototype stage, having been tested in laboratory environments. The system is not yet suitable for use within living organisms, but it lays the groundwork for future technologies that could seamlessly merge electronics with biology.
The potential applications of artificial neurons are vast. They could aid in repairing or replacing damaged neural circuits, enhance brain-machine interfaces (BMIs), or serve as real-time sensors to monitor cell health and drug responses. Given their low energy consumption and operation at biological signal levels, these artificial neurons could lead to more efficient brain-inspired computing hardware.
The findings of this study were published in the journal Nature Communications.
