For nearly a hundred years, neuroscientists have argued about a deceptively simple question: how did the brain's outer sheet of tissue — the cortex, seat of thought, perception and imagination — grow from something simple into the sprawling "smart hub" of a primate?
One camp said it expanded outward from the ancient cortex, the regions handling emotion and memory. The other said it grew out from the primary sensory cortex, where sight, sound and touch arrive. In April 2026, a China-led team published an answer in Science: both were right, and both were incomplete. The cortex grows from two poles at once — and the team has the whole-brain map to prove it.
The atlas behind the answer
The work was led by the Chinese Academy of Sciences' Center for Excellence in Brain Science and Intelligence Technology (CEBSIT, 脑科学与智能技术卓越创新中心), with BGI's Hangzhou research institute and Monash University in Australia, as part of China's brain-science big-science programs.
Its foundation is a first: the world's first single-cell, whole-brain, multimodal atlas of the marmoset — a small primate that has become the standard model animal in national brain projects worldwide. The team fused three kinds of data that usually live in separate silos:
- MRI imaging for large-scale structure and function;
- Stereo-seq, BGI's proprietary spatial transcriptomics technology, which reads gene expression cell by cell while keeping each cell's physical address in the tissue;
- neural connectivity data, tracing how regions talk to each other.
Aligning molecules, cell types, circuits and anatomy in one 3D reference frame is what let the researchers see the pattern nobody could see before.
Two poles, one "innovation zone"
The pattern is a double gradient — a pair of opposing molecular axes the team calls the Pr-Al axis. One pole sits in the ancient cortex, the brain's "old town," handling emotion and memory. The opposite pole sits in the primary sensory cortex, the "industrial zone" processing raw input from eyes and ears. The cortex expanded outward from both ends simultaneously, and where they collided, evolution built the high-order association cortex — the "innovation district" responsible for thinking, deciding and imagining.
The gradient does not stop at the cortex: the team showed the same axis runs through the thalamus, striatum and other deep brain structures, giving the whole organ a shared developmental logic. And the blueprint is drawn at birth — genes provide the map, experience then refines it.
The finding resolves the century-old dispute by unifying the rival hypotheses, and it holds across mammals, including humans. Peer reviewers, cited by Chinese state media, called it an original breakthrough likely to resonate far beyond neurobiology.
Why a brain atlas matters to brain-computer interfaces
The most practical payoff, ironically, belongs to the hardware field this site covers. Traditional brain atlases divide the cortex with hand-drawn lines that often match real function poorly. A continuous, molecularly grounded "topographic map" changes the engineering:
- Electrode targeting becomes more accurate. Implantable brain-computer interface (脑机接口) devices live or die on whether an electrode lands on the functional territory it intends to read — motor cortex versus one gyrus over. A gradient-based map reduces placement error.
- Cross-species translation gets a common reference. Because the dual-pole framework holds from marmosets to humans, findings from primate studies can be mapped onto human targets with more confidence.
- Disease localization sharpens. The team and commentators point to better mechanistic anchoring for conditions such as autism and schizophrenia, where genes implicated in these disorders cluster along the cortical gradients.
The research team itself framed the atlas as infrastructure: not a cure, but the coordinate system future cures will be written in.
The bigger picture
China has spent the past decade investing in brain-science infrastructure — whole-brain imaging pipelines, spatial-omics technology like Stereo-seq, primate colonies and national programs. This paper is the clearest sign yet that the investment strategy of "build the instruments and the maps first" is producing results that set the field's reference standards. For a technology site tracking China's brain-machine future, the significance is straightforward: the same ecosystem that maps the brain is the one building interfaces to it.
Honest limitations
- This is a marmoset atlas, not a human one. The dual-pole pattern appears conserved across mammals, but human-specific features of cortical organization still require human data; direct clinical use is years away.
- "Resolves a century-old debate" describes how the field received the paper, not permanent settlement — science revises, and replication in other labs and species will continue.
- The BCI implications are directional, not demonstrated. No electrode has yet been placed using this atlas in a published clinical case; the link between gradient maps and surgical targeting remains an argument, not a practice.
- Details of the findings here are drawn from the paper's abstract and coverage in Chinese state and institutional media; readers should consult the Science paper itself for technical specifics.
What readers can do now
- Read the paper if you work in neurotech or computational neuroscience — the atlas is a reference resource, and early familiarity with gradient-based brain coordinates will pay off as datasets spread.
- Follow China's brain-atlas programs as leading indicators for BCI. Where atlases go, electrode targeting, and eventually regulatory review of implants, will follow.
- For clinicians and BCI teams: watch for the atlas's public data release and cross-species mapping tools; they are the pieces most likely to become standard references in targeting workflows.