Summary:
Researchers have developed a high-throughput brain-mapping tool that reveals the brain does not simply rely on pre-calculated motion signals from the eye, but also computes the direction of movement independently. The study mapped nearly 46,000 mouse brain cells to uncover a dual visual system in the superior colliculus, separating localized object tracking from global self-motion sensing.
Key Facts:
- Mapping 46,000 Visual Brain Cells: Combining anterograde trans-neuronal fluorescent tracing with MERFISH spatial transcriptomics enabled researchers to map hundreds of eye-to-brain neural connection pairs across nearly 46,000 cells.
- Dual Motion-Sensing Pathways: Inside the superior colliculus, one population of neurons receives pre-calculated direction signals straight from specialized retinal cells, while a second distinct group independently computes motion direction from basic visual inputs, even when retinal direction sensors are disabled.
- Local Objects vs. Global Self-Motion: Retinal direction-selective pathways appear specialized for tracking small, localized targets like prey or incoming objects, whereas brain-computed direction circuits sense broad background movement to track how the body moves through space.
Source: UC San Francisco / Nature
Whether ducking a fast-moving ball, stepping back from an oncoming vehicle, or evading an approaching predator, survival depends on the brainโs ability to determine not just that an object is moving, but its exact trajectory in real time.
For decades, visual neuroscientists knew that specialized neurons in the mammalian retina fire selectively when stimuli move in specific directions, left, right, up, down, or diagonally.
Because these retinal ganglion cells already encode directional vectors, researchers widely assumed the brain simply inherited that pre-calculated motion data passively from the eye. However, no mapping technology existed that could simultaneously trace exact synaptic connections from the retina while identifying the molecular identity of every receiving brain cell.
Now, in a study published in Nature, researchers at UC San Francisco (UCSF) have bridged that technological gap.
Led by co-senior authors Dr. Xin Duan, professor of ophthalmology, and Dr. Evan Feinberg, professor of anatomy, alongside postdoctoral fellow Dr. Fei Wang, the team built a scalable eye-to-brain wiring atlas revealing that specialized circuits in the midbrain compute motion direction completely on their own.
Building an Atlas of Eye-to-Brain Connections
Over the past decade, Duanโs laboratory pioneered anterograde trans-neuronal tracing, a technique in which a fluorescent molecular tracer delivered to the eye crosses synaptic junctions to illuminate only those downstream brain neurons directly receiving retinal input.
To scale this tool from single-circuit tracing into a comprehensive atlas, Wang paired the trans-neuronal tracer with MERFISH (multiplexed error-robust fluorescence in situ hybridization), a spatial transcriptomic method that reads gene activity across thousands of individual cells in a tissue slice simultaneously to pinpoint each neuron’s exact genetic identity.
Applying this combined platform across nearly 46,000 cells in the mouse visual system, the investigators uncovered hundreds of distinct retinal-to-brain connection pairs.
โItโs like an entire atlas weโve built,โ Wang said. โWeโve uncovered hundreds of pairs of connections between the eye and the brain. We ended up focusing on only one page of this atlas, but there are many left to study in the future.โ
Two Parallel Ways to See Direction
Focusing on the superior colliculus, an evolutionarily conserved midbrain hub essential for visual orientation and rapid motor reflexes, Wang, Feinberg, and Duan recorded real-time neural firing while showing mice high-contrast black bars scrolling across a screen in eight distinct directions.
Combining functional recordings with their molecular wiring map exposed two separate motion-processing architectures operating side by side:
- Retina-Inherited Direction Cells: As long hypothesized, one population of superior colliculus neurons received direct, pre-calculated directional signals from direction-selective cells in the retina.
- Brain-Computed Direction Cells: A second, anatomically distinct population of superior colliculus neurons had no direct connection to retinal direction-selective cells. Instead, these neurons synthesized directional motion independently from basic, non-directional visual inputs. Crucially, even when researchers experimentally disabled the eyeโs dedicated direction-sensing machinery, this second group of brain neurons continued accurately detecting which way objects were moving.
โWe think these sets of cells are doing slightly different things from each other,โ Wang explained. โThe eye cells detecting direction may specialize in small, local motions like prey or food. The brain cells computing direction sense a larger background movement.โ
The investigators hypothesize that this global background motion signal informs an organism of its own self-motion and spatial trajectory as it navigates through the environment.
Precision Blueprints for Vision Restoration
Beyond resolving a foundational question in visual neuroscience, the high-resolution wiring atlas carries direct translational implications for regenerative medicine and neuroengineering. As researchers advance toward optic nerve regeneration, whole-eye transplantation, and high-resolution visual brain-machine interfaces for patients with glaucoma, trauma, or multiple sclerosis, restoring functional sight requires reconnecting damaged axons to their exact functional targets.
โThis is a fundamental mechanism that has been understudied from a basic science perspective, but it also has applied value,โ said Duan. โIf we ever want to repair or restore vision in people, we need to know which neural connections need to be rewired for functional vision.โ
โIn addition to providing a blueprint for revealing structural maps of the brain, our study illustrates how the resulting maps can be used to answer longstanding questions regarding brain function,โ added Feinberg.
The team notes that their scalable tracer-plus-MERFISH platform can be adapted to map how the eye regulates circadian rhythms, and generalized to chart cell-to-cell connectivity across any organ system in the body.
Questions and Answers
Does the eye do all the work of detecting which way things move?
No. While specialized cells in the retina detect motion direction and send that data to the brain, UCSF researchers discovered a second group of neurons in the brain’s superior colliculus that independently calculates motion direction using basic visual inputs, even when the eye’s own direction-sensing cells are turned off.
Why does the brain have two different systems for sensing motion?
Researchers believe the two pathways handle different visual jobs: the eye’s direction-sensing cells appear specialized for tracking small, local moving targets (like food, prey, or a flying ball), while the brain’s independent motion-computing cells detect large-scale background movement to help track how your own body is moving through space.
How could this brain-mapping tool help restore vision in the future?
To successfully develop eye transplants, optic nerve repairs, or visual brain-machine interfaces for conditions like glaucoma or multiple sclerosis, scientists need an exact wiring diagram showing which eye cells connect to which functional brain circuits. This new atlas provides that precise rewiring blueprint.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- Journal paper reviewed in full.
- Additional context added by our staff.
About this neuroscience and vision Research:
- Media Contact:ย Laura Kurtzman
- Source:ย UCSF
- Image Credit:ย Image credited to Neuroscience News
- Original Research is Open Access:ย Nature (Nov 18, 2026). โTransA-MERFISH reveals divergent post-retinal direction-selective circuits.โ Authors: Fei Wang, Lin Zhang, Siyuan Chang, Can Liu, Jitao Hu, Mengya Zhao, Nicole Y. Tsai, Kushal Nimkar, Shaobo Zhang, Jennifer L. Shadrach, Xiaoke Chen, Julia A. Kaltschmidt, Lin Gan, Shenqin Yao, Karthik Shekhar, Evan H. Feinberg & Xin Duan.
- DOI:ย 10.1038/s41586-026-11235-7
Abstract
TransA-MERFISH reveals divergent post-retinal direction-selective circuits
A major goal in neuroscience is to map the transformation of information as it flows through the brain. For example, neurons at many levels of the visual hierarchy are attuned to motion direction. However, links between motion-selective computations and cell types at various stages of the visual hierarchy remain unidentified. A key first step in delineating these relationships is to map the wiring diagrams that connect different motion-selective cell types.
Here we establish a systematic, cell-type-specific circuit-mapping method, combining an anterograde trans-neuronal fluorescent tracer (mWGAโmCherry) with spatial transcriptomics (multiplexed error-robust fluorescence in situ hybridization (MERFISH)) to establish TransA-MERFISH. We used TransA-MERFISH to comprehensively map the synaptic target cell types of four retinal ganglion cell subclasses in superior colliculus (SC) at cell-type resolution.
These experiments unexpectedly revealed that retinal onโoff direction-selective ganglion cells (ooDSGCs) innervate several direction-selective SC cell types, but do not innervate a SC cell type that was previously reported to display strong direction-selective responses.
We validated these results using orthogonal approaches and demonstrated the sensitivity and specificity of TransA-MERFISH. Functional studies of SC activity before and after ablation of retinal direction-selective circuits confirmed that these computationally distinct, ooDSGC-dependent and -independent direction-selective cell types are commingled in SC.
Together, our study establishes TransA-MERFISH as a robust framework for mapping wiring diagrams at the cell-type level and uncovers divergent computational mechanisms of motion detection in early visual processing.

