- Bio-Compatible Artificial Neuron: Engineers at UMass Amherst have created an artificial neuron that mirrors the electrical behavior of biological brain cells. It operates at ~0.1 volts, the same voltage range as human neurons, allowing it to “speak” the brain’s electrical language scitechdaily.com, umass.edu.
- Direct Communication With Living Cells: This synthetic neuron can interface directly with living neurons, exchanging signals in real time. In lab tests it connected to a biological cell and processed the cell’s signals, effectively interpreting the cell’s state – a first for artificial neurons theregister.com, nature.com.
- Built From Bacteria Nanowires: The device uses a memristor (memory transistor) made with protein nanowires from the bacterium Geobacter. These bio-derived nanowires conduct at ultra-low power, enabling neuron-like spikes with only picojoules of energy scitechdaily.com. Ionic signaling is involved, meaning the artificial neuron responds to chemical neurotransmitters similar to a real neuron nature.com.
- Who & When: The innovation was developed by graduate student Shuai Fu and Prof. Jun Yao at UMass Amherst, with a multi-disciplinary team. Their peer-reviewed study was published in Nature Communications on September 29, 2025 umass.edu, scitechdaily.com, and a university press release announced the breakthrough on September 30, 2025.
- Why It Matters: By operating on brain-level voltages and power, these artificial neurons could revolutionize neuromorphic computing – computers that run on brain-like efficiency – and enable new bioelectronic medical devices. Experts say it’s a game-changer for bridging silicon and biology, potentially leading to ultra-efficient AI hardware and implants or wearables that seamlessly communicate with the nervous system theregister.com, bath.ac.uk.
What Is the Artificial Neuron and How Does It Work?
The UMass Amherst team’s artificial neuron is essentially a tiny electronic neuron that behaves like a real one. Physically, it’s built from a special memristor device – an electrical component whose resistance can change based on past current (hence “memory resistor”). Unlike typical silicon transistors, this memristor was engineered to operate at extremely low voltages and currents, comparable to those in our brain’s neurons nature.com. The key material is a protein nanowire harvested from Geobacter sulfurreducens, a bacteria known for producing electrically conductive nanofilaments. These bio-nanowires give the artificial neuron an ion-friendly, highly efficient conduction channel, letting it fire signals with only a fraction of the power conventional electronics would require scitechdaily.com.
Functionally, the artificial neuron mimics the spiking behavior of natural neurons. Neurons communicate via brief voltage spikes (“action potentials”) and integrate incoming signals over time. The memristor-based design reproduces this: it can integrate weak electrical inputs and, once a threshold is reached, emit a voltage spike of ~0.1 V – about the same amplitude as a biological neuron’s spike umass.edu. Because the device operates in the same voltage range and timescales as real neurons, its output doesn’t overwhelm or misalign with biological signals. In other words, it fires just like a real brain cell, but on an electronic chip.
Another remarkable aspect is that the artificial neuron can be chemically modulated akin to a real neuron. The researchers report that it responds to neurotransmitter-like chemical cues in its environment, adjusting its activity in a way similar to how real neurons are regulated by chemicals (a process known as neuromodulation) nature.com. This means the device doesn’t just copy neurons’ electrical behavior in isolation – it also can interpret chemical “messages” (like ions or neurotransmitters) and alter its firing accordingly. By incorporating this capability, the artificial neuron is truly speaking the brain’s electrochemical language, not just electrical impulses.
In short, the UMass artificial neuron is a bio-inspired, nanoscale electronic neuron. It integrates a Geobacter nanowire memristor that allows it to generate authentic neuron-like spikes at ultra-low energy. This design achieves a long-sought goal in neuromorphic engineering: matching the fundamental parameters of biological neurons – from spike voltage and frequency to energy per spike – within a single device nature.com. Practically, it’s as if a tiny brain cell was recreated using hardware, able to compute and communicate like the real thing.
Interpreting the Brain’s Language: How It Communicates With Real Neurons
A major breakthrough of this development is that the artificial neuron can directly communicate with living biological neurons on their terms. In experiments, the team connected their artificial neuron to a real biological cell and demonstrated two-way signal exchange: the artificial neuron could “listen” to the cell’s electrical signals and interpret them, and potentially stimulate the cell in return theregister.com. According to the researchers, the device processed the cell’s signals in real time and was able to “interpret cell states” from those signals theregister.com. In other words, it could read what a biological neuron was “saying” through its electrical activity – effectively understanding the brain’s language of voltage spikes.
The reason this direct interface is possible comes down to the compatible signal characteristics. Biological neurons operate using ion flows that create voltage changes of only tens of millivolts. Traditional electronics use much larger voltages and purely electron-based currents, making direct coupling difficult – it’s like trying to have a shouting match with a whisperer. Previous artificial neurons required 10× the voltage (≥0.5 V) of real neurons and thus could not seamlessly plug into neural tissue without disturbing it scitechdaily.com, umass.edu. The UMass neuron solved this by bringing the operating voltage down to 0.1 V – roughly the same as real neurons’ spikes umass.edu. As Prof. Jun Yao noted, earlier attempts would have “frightened” living neurons with their higher amplitude, whereas ours works on the same level as the neurons in our bodies umass.edu. This parity means the artificial neuron’s outputs and inputs are recognizable to real neurons — no translation or large amplification needed.
Moreover, because the artificial neuron can respond to ionic chemical signals as well, it’s not limited to just electrical coupling. Neurons in the brain communicate not only via electrical pulses but also via neurotransmitters (chemical messengers). The UMass device’s nanowire memristor can interface with ions and molecules, effectively allowing ion-based signaling within the device nature.com. One can imagine, for example, an influx of calcium or a pulse of dopamine (just as examples of neurochemicals) influencing the artificial neuron’s state, just as it would a real cell. This dual-mode compatibility (electrical and chemical) truly lets the artificial neuron “speak the brain’s language” on all fronts.
The practical upshot is that this artificial neuron could be wired into biological neural circuits and they would understand each other. It’s akin to creating a cyborg circuit: living cells and silicon-based neurons exchanging information seamlessly. This was illustrated in prior conceptual demos – for instance, researchers in 2020 linked silicon neurons with a live rat neuron over the internet, showing that artificial chips could stimulate and respond to a biological neuron in a loop singularityhub.com. Now, the UMass team has achieved such integration in real time in the lab, with one artificial neuron physically connected to a cell. Silicon and biology can now speak the same electrical language, as one report put it ground.news. This capability opens the door to hybrid biological-electronic networks where artificial neurons can slot into a living neural circuit, listening, talking, and adapting just like a native cell.
Who Developed It and When?
This cutting-edge artificial neuron was developed by a team of engineers at the University of Massachusetts Amherst, led by Prof. Jun Yao (associate professor of electrical & computer engineering) and Shuai Fu (a PhD student who is the study’s lead author) umass.edu. They collaborated with several other researchers, including experts in nanotechnology and microbiology. Notably, the co-authors list includes Derek R. Lovley, a microbiologist known for pioneering work on Geobacter bacteria nature.com. Lovley’s involvement makes sense given the use of Geobacter protein nanowires in the device – his prior discoveries of these nanowires’ conductive properties laid groundwork for this bio-electronic innovation.
The team’s findings were peer-reviewed and published in Nature Communications on September 29, 2025 scitechdaily.com. The publication title, “Constructing artificial neurons with functional parameters comprehensively matching biological values,” reflects the essence of their achievement scitechdaily.com. The following day, September 30, 2025, UMass Amherst issued a press release to announce the breakthrough to the public umass.edu. The news quickly caught the attention of science media. SciTechDaily ran an article titled “Scientists Create Artificial Neuron That ‘Speaks’ the Language of the Brain” on October 4, 2025 scitechdaily.com, and Forbes also covered the story, emphasizing that these brain-like neurons run on “human brain voltage” ground.news.
This project builds on years of prior research by Prof. Yao’s group in developing bio-inspired electronics. Earlier, Yao and colleagues had already been experimenting with Geobacter nanowires in various devices – from a sweat-powered biofilm battery to an “electronic nose” that can sniff disease, and even a moisture-powered generator that pulls electricity from thin air umass.edu. Those inventions demonstrated the extraordinary efficiency of nanowire-based electronics. With that background, the team was poised to tackle the “holy grail” of artificial neurons. By combining Yao’s nanowire memristor platform with insights from neuroscience, Shuai Fu and the team managed to crack the problem of matching a neuron’s low-power, spiking behavior. The result was the first artificial neuron hardware that checks all the boxes of biological fidelity (voltage, spike shape, timing, energy, chemical responsiveness) while still being an engineered device nature.com.
Funding for the research came from agencies including the U.S. Army Research Office, the National Science Foundation (NSF), the National Institutes of Health, and the Alfred P. Sloan Foundation umass.edu. Such support underlines the broad interest in this technology’s potential, from defense applications (e.g. brain-machine interfacing for soldiers) to basic science and medicine. The interdisciplinary nature of the team and funding highlights that this wasn’t an isolated effort – it sits at the crossroads of neuroscience, bioengineering, nanotechnology, and computing.
Why It Matters: Implications for Neuroscience, Medicine, and AI
This innovation is significant on multiple fronts, heralding important implications for computing technology, neuroscience research, and medical applications:
- Ultra-Efficient Computing (Neuromorphic AI): The artificial neuron could help computers run with brain-like efficiency. Today’s computers and AI systems guzzle vastly more power than the human brain. (For example, training or running a large language model can consume millions of watts, whereas the human brain uses only ~20 watts to perform complex tasks scitechdaily.com.) By building computers out of brain-mimicking neurons that operate on the same low power levels, we could achieve a new class of “green” AI hardware. “Our brain processes an enormous amount of data, but its power usage is very, very low, especially compared to… a large language model like ChatGPT,” lead author Shuai Fu noted umass.edu. This gap in efficiency has been a grand challenge; as Stanford’s Prof. Surya Ganguli argued in 2024, we need “to rethink the entire technology stack from electrons to algorithms in order to really go from megawatts to watts” theregister.com. The UMass artificial neuron is a tangible step toward that vision – it literally brings computing elements down to watts-from-megawatts scale by using the brain’s strategies of low-voltage analog processing. In practice, arrays of such artificial neurons might form neuromorphic chips that can run AI algorithms with a tiny fraction of the energy of today’s silicon chips, enabling powerful AI in portable devices or reducing the carbon footprint of data centers.
- Direct Brain-Computer Interfaces: Because these artificial neurons can communicate directly with living cells, they offer a new approach to brain-computer interfaces and neuroprosthetics. We can imagine integrating artificial neurons with a patient’s nervous system to repair or enhance function. For instance, in cases of spinal cord injury or neurodegenerative disease, artificial neurons might be implanted to replace damaged neural circuits and restore communication pathways. Previous research had pointed in this direction – in 2019, a University of Bath-led team demonstrated silicon neurons that accurately replicated rat neuron responses and suggested they could be used in medical implants to cure chronic diseases bath.ac.uk. Those chips ran on nanowatts of power, ideal for implantables bath.ac.uk. The UMass device adds the crucial capability of real-time interaction with actual cells. This could translate to “smart” implants that not only stimulate tissue (like current pacemakers or deep brain stimulators do) but also listen and adapt to the body’s feedback signals. For example, a smart pacemaker built with artificial neurons could sense neural feedback and adjust heart stimulation on the fly – just as the natural nervous system does bath.ac.uk. Similarly, neuroprosthetic limbs could achieve more natural control and sensation if the interface neurons share the brain’s native signal properties.
- Wearable and Minimally Invasive Bioelectronics: Even outside of fully implanted devices, this technology can improve external health monitors and interfaces. Wearable sensors today (for EEG, ECG, muscle signals, etc.) often require complex amplification and power-hungry electronics to convert the body’s weak electrical signals into digital data. With artificial neurons that operate on the same level as those weak bio-signals, one could build much simpler wearable electronics. “Every time [today’s devices] sense a signal from our body, they have to electrically amplify it so a computer can analyze it… that step increases power consumption and complexity. Sensors built with our low-voltage neurons could do without any amplification at all,” Prof. Jun Yao explained umass.edu. Imagine a skin patch or a tiny implant that directly converses with your nerves – reading signals and even writing signals back – all with minimal battery drain. This could enable continuous health monitoring or new therapeutic devices that operate for long periods on small power sources (or even harvest body energy). Essentially, bioelectronics would become more seamlessly integrated with our physiology, since the signal translation barrier is removed.
- Advancing Neuroscience Research: From a scientific perspective, having an artificial neuron that so closely emulates real ones gives researchers a powerful model system. Neuroscientists could use these devices to construct hybrid networks (mixing live neurons and artificial ones) to test hypotheses about brain function. Because the artificial neurons can be finely controlled and observed, they provide a stable stand-in for biological neurons in experiments. Researchers could investigate how networks form, learn, or heal by swapping out or adding artificial neurons. Additionally, the fact that the artificial neuron can report on “cell states” theregister.com suggests it might be used as a real-time biosensor for cellular health or activity. For example, one could connect an artificial neuron to a cultured brain cell or tissue slice and have it monitor subtle changes in the cell’s firing patterns or chemical environment, potentially flagging when the cell is under stress or responding to a drug. This could accelerate research in neuropharmacology or disease modeling by bridging electronic readouts with living tissue responses in a more natural way.
- Neuromorphic Networks and Beyond: In the longer term, developments like this blur the line between biological and artificial networks. We might see neuromorphic computers composed of millions of these low-power artificial neurons tackling problems in ways classical computers can’t. They could excel at brain-like tasks – pattern recognition, learning from unstructured data, etc. – with far less energy, enabling advanced AI in mobile or edge devices. On the flip side, in medicine, one could envision neuro-hybrids where doctors introduce networks of artificial neurons into the body to reinforce or repair neural functions (for example, an artificial neural network graft to restore memory circuits lost to Alzheimer’s someday). While such applications are speculative for now, the UMass team’s achievement is a foundational step. It provides the hardware proof-of-concept that artificial and biological neurons can truly speak to each other fluently. This has fueled excitement that we’re moving closer to a future where computing devices and the human nervous system are interoperable at a fundamental level – a future of brain-like computers and perhaps computer-augmented brains.
Expert Commentary and Reactions
The researchers themselves underscore the significance of matching the brain’s own hardware. “Previous versions of artificial neurons used 10 times more voltage – and 100 times more power – than the one we have created,” notes Prof. Jun Yao, pointing out why those earlier attempts couldn’t interface with biology effectively umass.edu. By contrast, the new device runs at “only 0.1 volts, which is about the same as the neurons in our bodies,” Yao says, highlighting the breakthrough in voltage parity umass.edu. This low-voltage operation is what allows the artificial neuron to comfortably connect with living cells without harming them or losing information. Yao envisions broad uses for the technology, from bio-inspired computers to direct body interfaces: “There are a wide range of applications… from redesigning computers along bio-inspired principles, to electronic devices that could speak to our bodies directly,” he told the UMass news service umass.edu.
Shuai Fu, the graduate student who led the work, framed it in terms of the brain’s efficiency versus current AI. “Our brain processes an enormous amount of data… but its power usage is very, very low, especially compared to the electricity it takes to run a large language model, like ChatGPT,” Fu said umass.edu. This comparison underlines the motivation behind the research: closing the immense energy gap between biological intelligence and artificial intelligence. By proving an artificial neuron can operate on the brain’s power budget, Fu and colleagues illustrate a path to more sustainable, brain-like AI systems. It’s a point not lost on outside experts. Surya Ganguli, a Stanford University physicist and AI researcher (who was not involved in the UMass work), has emphasized that achieving brain-level energy efficiency in computing may require reimagining our devices from the ground up. He said we must “rethink the entire technology stack from electrons to algorithms in order to really go from megawatts to watts” theregister.com. The UMass artificial neuron is a perfect example of this principle – by rethinking the basic electronic component (the “electron” level, so to speak) to operate like a neuron, the team took a leap toward the watt-level paradigm that Ganguli and others advocate.
Experts in bioelectronics and neuroscience are also enthusiastic. Although not speaking directly about the UMass device, pioneers in artificial neuron research have long anticipated such advances. Back in 2019 when a different team built silicon-based neurons, Prof. Alain Nogaret of University of Bath called it “paradigm changing” and noted the huge potential for medical implants given the ultra-low power requirement bath.ac.uk. The sentiment in the field is that artificial neurons that truly behave like real ones could transform both neuromorphic engineering and healthcare. “Designing artificial neurons that respond to electrical signals from the nervous system like real neurons has been a major goal in medicine for decades,” that 2019 Bath press release stated bath.ac.uk – a goal now reached in full by the UMass accomplishment.
Outside commentators also highlight the exciting convergence of biology and technology. Tech writers have described the UMass neuron as bringing us closer to “cyborg” interfaces, since it moves beyond clunky electrodes toward actual synthetic neurons communicating with human cells theregister.com. There’s a palpable sense that this is a foundational building block for future innovations. As one coverage noted, “Silicon and biology can now speak the same electrical language.” ground.news This harmony could enable everything from advanced brain-machine interfaces to new kinds of biological computing. The achievement has thus been met with optimism that we are entering a new era where the gap between living tissue and electronic systems diminishes.
In summary, the expert consensus is that UMass Amherst’s artificial neuron marks a significant milestone. It validates a long-sought concept (a true bionic neuron equivalent) and opens practical avenues. The developers underscore its unprecedented bio-compatibility and efficiency, and independent experts laud the implications for reimagining computing and healing the body. It’s a development that resonates with grand ambitions – smarter AI, better treatments for neurological conditions, and deeper understanding of the brain itself – all made possible by an engineered cell that speaks the brain’s mother tongue.
Context: Similar Breakthroughs and Ongoing Research
The quest to merge the strengths of neurons with electronics has been underway for years, and the UMass advance builds upon several important milestones:
- Silicon Neurons for Medicine (2019): In late 2019, a team led by University of Bath announced the first artificial neurons on silicon chips that behave just like real ones bath.ac.uk. They successfully reproduced the complete electrical behavior of certain rat neurons (like heart and brain cells) in chip form. Importantly, those silicon neurons ran on only 140 nanowatts of power – about one-billionth the power of a typical microprocessor, mirroring the incredible efficiency of biology bath.ac.uk. This breakthrough, published in Nature Communications as well, was hailed as having “enormous scope for medical devices to cure chronic diseases” bath.ac.uk. By responding to inputs like real neurons, such chips could potentially replace damaged neurons; for example, the team envisioned smart pacemakers that respond to real-time feedback from the nervous system bath.ac.uk. The Bath study tackled the challenge through sophisticated mathematical modeling of neuron dynamics and analog circuit design bath.ac.uk. However, those early silicon neurons, while ultra-low-power, did not yet demonstrate direct coupling to living cells. The UMass neuron extends this line of work by achieving biocompatible signaling and chemical responsiveness, which could make actual integration into biological systems much more seamless.
- Hybrid Artificial-Biological Networks (2020): A notable proof-of-concept came in 2020 when an international team linked artificial and biological neurons across three countries, creating a hybrid neural network. Using a “silicon spiking neuron” in Switzerland, a memristor-based artificial synapse in the UK, and a living rat neuron in Italy, they demonstrated that signals could be passed and processed through a chain mixing live and artificial nodes singularityhub.com. Remarkably, this network was able to mimic a basic learning rule (a form of long-term potentiation, a cornerstone of memory) entirely via distributed artificial and biological components singularityhub.com. This experiment showed that artificial neurons and synapses can indeed communicate with real neurons to perform a cognitive function. It relied on standard internet connections to bridge distances, digitizing the biological signals for transport singularityhub.com. The UMass advancement brings this concept closer to reality in a localized setting: rather than a virtual link, we now have a physical artificial neuron that can plug into a biological circuit on-site. The 2020 hybrid network was a hint that such integration is possible; the 2025 UMass neuron makes it practical by matching the biological signal characteristics so closely that extensive signal conversion isn’t necessary.
- Advanced Neuromorphic Materials (2023): Other research groups are exploring different approaches to artificial neurons aimed at computing improvements. In 2023, scientists from University of Oxford, IBM Research Europe, and University of Texas developed atomically thin 2D-material neurons that can process both electrical and optical signals ox.ac.uk. By stacking ultrathin layers (graphene, MoS₂, WS₂), they created devices that incorporate both feedforward and feedback pathways, akin to complex brain circuits, and demonstrated these in neural network models ox.ac.uk. Published in Nature Nanotechnology, that work wasn’t about interfacing with biological neurons, but rather about pushing the envelope of neuromorphic computing hardware – making artificial neurons more powerful and brain-like in their computation (for example, enabling learning via feedback signals, not just one-way flow) ox.ac.uk. The common thread with the UMass neuron is the drive to capture the multifaceted abilities of real neurons (low power, analog signaling, rich dynamics) in man-made devices. The Oxford/IBM team’s success with 2D materials shows another facet of the field: using novel nanomaterials to achieve brain-like functions. Meanwhile, the UMass team used a biological nanomaterial (bacterial wire) to achieve brain-like efficiency. Together, these efforts highlight a vibrant, converging research area where material science, biology, and computer engineering meet.
- Bioelectronic Interfaces: The broader field of bioelectronics has seen various devices that aim to translate between electronic and biological signals – from cochlear implants that interface with auditory nerves to brain-computer interface electrodes. What sets the new generation of artificial neurons apart is that they themselves function like neurons, rather than just stimulating or recording neurons. This could mean a leap from electrodes (which are essentially passive wires) to active neuron replacements or additions in neural circuits. The UMass team’s focus on eliminating the voltage and power mismatch addresses one of the biggest hurdles noted in biointerface research: the signal incompatibility and energy inefficiency when linking wet biology to dry electronics umass.edu. By solving this, they have paved the way for more sophisticated integration. Future research will likely build on this by connecting networks of artificial neurons with real neural networks, testing function in living organisms, and scaling up the technology.
In conclusion, the creation of a brain-like artificial neuron at UMass Amherst stands on the shoulders of prior breakthroughs while significantly advancing the state of the art. It combines the medical promise seen in the 2019 silicon neurons (biologically faithful and low-power) with the interactive capability hinted by the 2020 hybrid experiments (two-way communication with real cells). Add to that the cutting-edge materials approach, and it’s clear this achievement is a cornerstone in the emerging landscape of brain-inspired technology. Around the world, researchers are racing to unlock the secrets of the brain’s efficiency and adaptability. With this development, scientists now have a compelling proof that we can build artificial neurons which are every bit as frugal and fluent as the real ones scitechdaily.com, theregister.com. The implications – smarter AI, better brain repairs, and deeper human-machine synergy – make this a truly exciting leap in science and engineering.
Sources and Further Reading:
- Shuai Fu et al., Nature Communications (Sept 2025) – “Constructing artificial neurons with functional parameters comprehensively matching biological values” scitechdaily.com, nature.com.
- UMass Amherst News (Sept 30, 2025) – “UMass Engineers Create First Artificial Neurons That Could Directly Communicate With Living Cells” umass.edu.
- SciTechDaily (Oct 4, 2025) – “Scientists Create Artificial Neuron That ‘Speaks’ the Language of the Brain” scitechdaily.com.
- The Register (Sept 30, 2025) – “Cyborg dreams move closer to reality with low-power artificial neuron” theregister.com.
- Forbes (Oct 3, 2025) – “Scientists Create Brain-Like Neurons That Run On Human Brain Voltage” (summary accessible via Ground News) ground.news.
- University of Bath News (Dec 3, 2019) – “World first as artificial neurons developed to cure chronic diseases” bath.ac.uk.
- Singularity Hub (Mar 10, 2020) – “Scientists Linked Artificial and Biological Neurons in a Network — and Amazingly, It Worked” singularityhub.com.
- University of Oxford News (May 5, 2023) – “Artificial neurons mimic complex brain abilities for next-generation AI computing”ox.ac.ukox.ac.uk.
