Summary: In the 1960s, Nobel laureates Hubel and Wiesel proposed that visual perception is a “bottom-up” process, where the brain builds complex images from simple features like edges and lines. For decades, a fierce debate persisted: does this “feature selectivity” start in the eye’s relay station (the thalamus), or is it constructed in the cortex?
A landmark study has finally settled the score. Using ultra-high-resolution imaging and optogenetics to “mute” specific brain parts, researchers confirmed that the thalamus sends raw, non-specific data, while the cortex acts as the master architect, refining that data into the specific orientations that allow us to see the world.
Key Facts
- Confirming a Legend: The study directly proves the “Stepwise Computation” model proposed by Hubel and Wiesel over 60 years ago.
- The “Raw Data” Feed: Signals arriving from the thalamus are robust but “broadly tuned,” meaning they don’t distinguish between a horizontal or vertical line.
- Cortical Construction: Orientation selectivity—the ability to tell a vertical stripe from a horizontal one—emerges only once the signal is processed within the cortical circuits.
- Learning Divide: In a surprise twist, researchers found that synapses within the cortex show signs of learning and plasticity (calcium signals), while those coming from the thalamus do not.
- New Tech: The team used two-photon microscopy and fluorescent proteins to watch individual synapses fire in real-time in a living brain.
Source: TUM
Already in the 1960s, Hubel and Wiesel proposed a model according to which visual perception is the result of orderly, stepwise computations in the brain – with specialized neurons in the cortex responding selectively to specific features, such as edges or the orientations of moving objects.
While widely celebrated, important aspects of the theory remained an issue of debate: does this feature selectivity already originate in the thalamus, or does it emerge later in the cortex?
The new study addresses this question directly by analyzing signal transmission at individual synapses between the thalamus and the visual cortex – something that had not previously been possible.
The research team, led by Prof. Arthur Konnerth, Dr. Yang Chen, and PhD student Marinus Kloos at the Institute of Neuroscience at the TUM School of Medicine and Health and the Munich Cluster for Systems Neurology (SyNergy), developed a high-resolution imaging approach to measure synaptic activity in the intact brain. Their findings directly confirm core predictions of the Hubel and Wiesel model. The new research results were published in the prestigious journal Science.
“Our results highlight how remarkably accurate and forward-looking Hubel and Wiesel’s insights were,” says Prof. Konnerth. “Modern neuroscience – and even artificial neural networks – continue to build on their principles. Learning from biological systems remains a powerful driver of technological innovation.”
What exactly did the TUM researchers do?
When we see, signals travel from the eye first to the thalamus, a relay station deep in the brain, and from there to the visual cortex at the back of the head. In the first area of this visual cortex, known as the primary visual cortex, simple image features like edges, contrast, and orientation are processed. The TUM researchers specifically examined this segment – the connection from the thalamus to this initial visual area of the cortex – in mice.
Using two-photon microscopy, the researchers visualized individual synapses in the living brains. They employed fluorescent proteins that emit light when synaptic transmission occurs, allowing them to track activity at specific neuronal contact points in real time. At the same time, the animals were presented with simple visual stimuli, such as horizontal and vertical stripes, enabling the team to map which synapses responded to which orientations.
To distinguish direct input from the thalamus from signals generated within the cortex, the researchers used optogenetics. They equipped certain neurons with light-sensitive proteins and could thus temporarily “mute” parts of the cortex with light. So, they could determine whether synaptic activity persisted (indicating thalamic input) or disappeared (indicating intracortical processing).
This approach allowed the team to separately quantify thalamocortical and corticocortical inputs. The results were clear: signals arriving from the thalamus were robust but largely non-specific with respect to orientation. In contrast, orientation selectivity – such as distinguishing horizontal from vertical lines – emerged only through processing within cortical circuits.
These findings resolve a long-standing controversy. The new data show directly that, in mammals, the cortex constructs this information step by step from broadly tuned inputs – precisely as predicted by Hubel and Wiesel.
Implications for neuroscience and beyond
Beyond confirming a foundational theory, the study introduces a versatile method for analyzing synaptic function. According to the researchers, this technique can be applied to a wide range of neuron types and may help identify dysfunctional circuits in neurological disorders such as Alzheimer’s disease.
The study also revealed a fundamental difference between synapse types. Synapses within the cortex (corticocortical synapses) exhibited calcium signals associated with learning and plasticity, whereas synapses from the thalamus (thalamocortical synapses) did not.
“This was an unexpected finding,” Konnerth explains. “It suggests that not all synapses have the same capacity for adaptation and learning, challenging long-standing assumptions in neuroscience.”
Key Questions Answered:
A: It tells us how the brain’s “operating system” is organized. If the thalamus did the heavy lifting, the cortex would just be a display screen. Because the cortex does the work, it shows that the brain is designed for hierarchical processing—taking simple bits of data and building them into complex concepts. This is exactly how modern AI and “neural networks” are designed!
A: Scientists usually assume all synapses can “learn” (change over time). This study found that the “input wires” from the thalamus are basically hard-wired and stable, while the “internal wires” of the cortex are the ones that adapt and change. It’s like your computer having a fixed power cable but a highly flexible processor.
A: They used optogenetics. By engineering specific neurons to respond to light, they could use a laser to temporarily “switch off” the cortex’s internal chatter. This allowed them to see exactly what the thalamus was sending without the cortex “interrupting,” proving the thalamic signals were non-specific.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- Journal paper reviewed in full.
- Additional context added by our staff.
About this visual neuroscience research news
Author: Ulrich Meyer
Source: TUM
Contact: Ulrich Meyer – TUM
Image: The image is credited to Neuroscience News
Original Research: Closed access.
“Thalamic activation of the visual cortex at the single-synapse level” by Yang Chen, Marinus Kloos, Zsuzsanna Varga, Yonghai Zhang, Inken Piro, Tatsuo K. Sato, Bert Sakmann, Israel Nelken, and Arthur Konnerth. Science
DOI:10.1126/science.aec9923
Abstract
Thalamic activation of the visual cortex at the single-synapse level
Deciphering thalamocortical (TC) activation at the level of individual synapses is essential to understanding how the cortex processes sensory information. In this work, we studied TC computation underlying the emergence of orientation selectivity in the mammalian primary visual cortex (V1).
Using two-photon glutamate imaging and optogenetic cortical silencing in vivo, we identified and characterized TC synapses onto mouse V1 layer 4 neurons. We found that TC- but not corticocortical-recipient spines lacked postsynaptic Ca2+ signals.
Our results directly validate the core predictions of Hubel and Wiesel’s feedforward model and reveal distinctive synaptic properties that are critical for cortical computation and plasticity.
