A tiny circuit, printed from ink made of atom-thin crystals, just fired electrical pulses that a living brain cell recognized and responded to. Researchers at Northwestern University say it is the first time a fully printed artificial neuron has communicated directly with biological neurons, a milestone published in May 2026 in Nature Nanotechnology.
The achievement matters because it bridges two worlds that have resisted merging: flexible, mass-producible electronics and the soft, electrically active tissue of the brain. If the approach holds up under longer testing, it could open a practical path toward brain-machine interfaces, neurological implants, and computing hardware that processes information the way neurons do.
How printed ink becomes a working neuron
Think of a biological neuron as a tiny battery that charges up and, once it hits a threshold, fires a brief voltage spike lasting roughly one to a few milliseconds before resetting. Networks of these spiking cells are what let brains recognize faces, form memories, and control muscles. Neuromorphic engineering tries to build electronic components that spike in the same way, so hardware can process information through pulses of current rather than the steady binary signals used by conventional computer chips.
The Northwestern devices are built from two nanomaterials: molybdenum disulfide (MoS₂) flakes and graphene. Both can be suspended in liquid to create electronic inks, which the team deposited onto flexible surfaces using aerosol jet printing. The technique works by spraying fine droplets through a nozzle, guided by a focused gas stream so they land in precise patterns. The result is a circuit thin and bendable enough to conform to curved, moving tissue, a basic requirement for anything meant to sit on or near a brain.
Once printed, the MoS₂ networks behave as memristive devices, meaning their electrical resistance shifts depending on the history of current flowing through them. That property lets the circuits store and process information in a way that loosely mirrors how biological synapses strengthen or weaken over time. By tuning the arrangement of devices and circuit parameters, the Northwestern team coaxed the networks into generating voltage spikes on their own, without external digital control. According to the paper, the spikes had widths on the order of milliseconds and inter-spike intervals that fell within the ranges typically observed in biological neurons, roughly 1 to 100 milliseconds for spike duration and tens to hundreds of milliseconds between spikes, depending on firing mode.
Crucially, the devices did not just repeat a single monotonous pulse. They switched among several firing patterns, including regular spiking, bursting, and more complex rhythms. In neuromorphic engineering, that versatility is described as “multi-order complexity” and is considered a hallmark of brain-like dynamics, because real neurons constantly shift their firing behavior depending on context.
Talking to living brain cells
The team then placed their printed neurons alongside living brain cells in a controlled laboratory setting and recorded the electrical exchange. According to the Nature Nanotechnology paper and Northwestern’s institutional announcement, the artificial devices produced spikes that the biological neurons detected and responded to, establishing two-way signaling between printed hardware and living tissue.
“This is the first time that printed artificial neurons have been shown to communicate with biological neurons,” said Mark C. Hersam, the Northwestern professor whose research group led the study, as quoted in the university’s press release.
The word “printed” is the key qualifier. An earlier project at the University of Massachusetts, announced through the AAAS-hosted platform EurekAlert!, reported artificial neurons communicating with living cells, but those devices were fabricated through a different, non-printing process. It is worth noting that the UMass work, as of June 2026, has been publicized only through a press release and has not yet appeared in a peer-reviewed journal, so its claims carry less evidentiary weight than the Northwestern results published in Nature Nanotechnology. The two claims appear compatible rather than contradictory: UMass can reasonably claim priority for artificial-neuron-to-cell communication in general, while Northwestern’s advance is specific to devices produced by scalable printing.
Scalability is the practical payoff. Traditional semiconductor fabrication requires expensive cleanroom lithography. Aerosol jet printing, by contrast, uses liquid inks and can deposit material on flexible substrates at relatively low cost. The Northwestern team suggests in its paper that printing-based fabrication could offer a more scalable route to neuromorphic hardware than conventional lithography, though the authors do not provide direct cost or throughput comparisons, and whether the technique can reliably produce devices at volume remains to be demonstrated.
Where the science fits in a larger picture
The Northwestern devices belong to a growing family of neuromorphic electronics, hardware engineered to mimic the brain’s information-processing style rather than the binary logic of conventional chips. A widely cited review of neuromorphic nanoelectronic materials in Nature Nanotechnology catalogs the main device types: memristive, threshold-switching, ionic, and electrochemical. Each can be configured to reproduce spiking, plasticity, or other neural functions. The Northwestern work uses memristive MoS₂ elements, placing it squarely within one of the most active branches of the field.
What sets this project apart is the combination of printing-based fabrication, flexible form factor, and demonstrated biological communication in a single platform. Previous neuromorphic chips have achieved impressive computational feats, but most are rigid silicon devices with no direct biological interface. Conversely, some bioelectronic implants can stimulate or record neural activity but lack the internal spiking dynamics that would let them process information the way a neuron does. The Northwestern prototype attempts to do both.
What still needs to be proven
A proof of concept is not a product, and several large questions remain unanswered.
Long-term stability. Neither the journal article nor the press materials report how the printed neurons perform over days, weeks, or months. For any medical application, durability under warm, wet physiological conditions will matter as much as initial spike quality.
Biocompatibility. The available documents do not detail whether the presence of printed circuitry altered cell health, triggered inflammation, or changed gene expression over time. Short-term communication is encouraging, but it does not guarantee that the interface is safe for extended contact with living tissue.
Performance benchmarks. The study does not include a direct, side-by-side comparison between printed neurons and neuromorphic devices made through conventional lithography. Metrics such as energy per spike, switching speed, and endurance, outlined in the neuromorphic materials review, have not been explicitly mapped onto the printed devices. Until those comparisons exist, it is difficult to say whether printing sacrifices performance for flexibility and cost.
Independent replication. No outside laboratory has publicly reported reproducing the results. That is normal for freshly published work, but it means the findings rest on a single dataset from a single group.
From ink to neuron: what the printing milestone does and does not prove
The long-term promise is a class of electronics that can be manufactured like printed circuits, flex like living tissue, and speak the electrical language of the brain. Such devices could eventually serve as components in next-generation neural implants for conditions like epilepsy or paralysis, where precise, two-way communication with neurons is essential. They could also feed into neuromorphic computing systems that consume far less power than conventional processors, an increasingly urgent goal as artificial intelligence workloads strain global energy supplies.
None of that is guaranteed by a single paper. What the Northwestern team has shown is that the raw capability exists: you can formulate brain-compatible inks, print them onto flexible substrates, and get the resulting circuits to fire spikes that living neurons understand. Turning that capability into durable, safe, and affordable technology is a different challenge entirely, one that will require years of further engineering, biological testing, and independent validation. But as of May 2026, the starting line has been crossed, and the path from ink to neuron is shorter than it was before.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
