5 Orders of Magnitude Smarter: How Artificial Neurons Finally Won the Heart of a Real Brain
Engineers at Northwestern University prove that sometimes, you really have to be soft to connect with someone deeply.
There is a specific kind of ache in the modern world. It is the ache of wanting to be truly understood. We talk to our phones. We train our AI. But do they truly hear us? I’ve been thinking a lot lately about the nature of connection between the synthetic and the organic. This week, I found a story that feels less like a scientific breakthrough and more like a cosmic romance.
Engineers at Northwestern University have managed something that sounds like science fiction but is very much here. They created printed artificial neurons that do more than just mimic life. These devices interact directly with real brain cells. It is a meeting of the minds, or in this case, a meeting of circuits and synapses. The goal was never just to make a copy of the brain. It was to make the technology feel like it belongs there. Mark C. Hersam, who leads this research, explained that the world is dominated by artificial intelligence. We train it on data until it is smart. But this intelligence comes with a massive cost. It consumes energy at a staggering rate.
Think about a relationship. If you have to keep buying new gifts just to keep the conversation going, the connection becomes exhausting. Modern computing is the same way. It is rigid. It is fixed. Mark Hersam, who is also the Walter P. Murphy Professor of Materials Science and Engineering at the McCormick School of Engineering, tells me that the brain is five orders of magnitude more energy efficient than a digital computer. That is a lot of zeros. It makes sense to look for inspiration from the organ that invented efficiency.
We tend to think of the brain as a mysterious organ, but on a cellular level, it is actually quite cooperative. It is a network of neurons, each playing a role, constantly changing and adjusting. Silicon, however, is the opposite. It is rigid. It is two-dimensional. It does not like to change. To build a relationship between electronics and biology, you cannot just force them together. You have to change the materials. You have to be flexible.
The team turned to a process called aerosol jet printing. They used materials that are closer to the brain's make-up than silicon chips. They used nanoscale flakes of molybdenum disulfide or MoS2. This acts like a semiconductor. They also used graphene, which acts as an electrical conductor. It is a combination that feels almost organic in its intent. The challenge has always been how to make the signal feel real. Most artificial neurons produce simple, static signals. They are boring. They lack the spark of a real neuron firing.
Silicon tries to achieve complexity by having billions of identical devices. It is like trying to have a conversation with a wall of identical people. It is not very dynamic. The brain, on the other hand, is dynamic. It forms and adjusts connections as you learn. To fix this, the Northwestern team used a clever trick with the materials. They printed their devices onto flexible polymer surfaces. In the past, this polymer was seen as a flaw. It interfered with the electrical performance. They usually just removed it.
This time, they kept it. Instead of removing it, they partially decomposed it. When you pass current through the device, it drives the decomposition further. It happens in a spatially inhomogeneous manner. This means it creates a conductive filament. The current gets constricted into a narrow region. This narrow path produces a sudden electrical response. It is very similar to a neuron firing. The result is a device that can generate a wide variety of signals. It can do single spikes, continuous firing, and bursting patterns. It finally feels like it is keeping up with the conversation.
But you cannot just build a device and hope it fits. You have to test it on the object of your affection. The researchers partnered with Indira M. Raman, the Bill and Gayle Cook Professor of Neurobiology at Weinberg. She applied the artificial signals to slices of mouse cerebellum. You might think this is cold, but in the world of relationships, you have to put your money where your mouth is. The results were heartwarming. The electrical spikes matched key biological properties. They matched the timing and the duration.
The signals reliably activated real neurons and triggered neural circuits. It was a true connection. Mark Hersam noted that other labs had tried to make artificial neurons with organic materials. Their signals were too slow. Or they used metal oxides, which were too fast. His team achieved a middle ground. It is a temporal range that has not previously been demonstrated. You can see the living neurons respond. It is a satisfying moment.
There is a practical beauty to this research, too. The manufacturing process is simple and inexpensive. It is additive, meaning it places material only where it is needed. This reduces waste. It is a sustainable way to build a relationship. As AI systems grow more demanding, the energy cost becomes a real issue. Large data centers already consume vast amounts of power. They require significant water for cooling. To meet the energy demands of AI, tech companies are building gigawatt data centers powered by dedicated nuclear power plants. It is evident that this massive power consumption will limit further scaling of computing. It is hard to imagine a next-generation data center requiring 100 nuclear power plants.
Furthermore, when you are dissipating gigawatts of power, there is a lot of heat. Because data centers are cooled with water, AI is putting severe stress on the water supply. However you look at it, we need more energy-efficient hardware. We need to come up with systems that perform complex tasks using far less energy. This technology moves us closer to electronics that can directly interface with the nervous system.
Imagine the applications. This is not just about curiosity. It points toward brain-machine interfaces and neuroprosthetics. These are devices that could help restore hearing, vision, or movement. It is the ultimate act of restoration. To find out more about how these materials are changing the landscape of neurotechnology, you can read the study published on April 15 in the journal Nature Nanotechnology. It is a fascinating read for anyone interested in how we might one day heal ourselves.
The study, titled "Multi-order complexity spiking neurons enabled by printed MoS2 memristive nanosheet networks," was supported by the National Science Foundation. It is a collaborative effort that shows the best of what we can do when we combine different fields of study. From the McCormick School of Engineering to the Feinberg School of Medicine, the expertise required to bridge this gap is immense. It requires a team that understands both the hard matter of engineering and the soft matter of biology.
This is a reminder that sometimes the most complex solutions are the ones that mimic the most natural things. We are learning that to build the future, we don't always have to reinvent the wheel. Sometimes we just have to look inside the brain and see how it manages to keep the spark alive. It is a testament to the power of connection. It is a testament to the fact that even in the age of silicon, there is still nothing quite like a biological connection.
First reported by ScienceDaily: Relationships.
More in Love & Relationships
- Clear requests and respectful boundaries belong in everyday relationships
NIH guidance highlights honest conversation, listening and limits as parts of supportive connections.
- How Eye-Tracking Revealed Hidden Social Skills in Severely Disabled Youth
A new study finds that personalized training can unlock social abilities in people often considered untestable.
- Sleep matters: Duration, timing, quality and more may affect cardiovascular disease risk
Healthy sleep includes multiple components, such as number of hours of sleep per night, how long it takes to fall asleep, daytime functionin