11 February 2026
Visual experience triggers the formation of a web of neural connections in different brain areas in order to make sense of the world – and in particular, of feedback connections, which send information from higher-level visual centres back to earlier ones. But is the organisation of these connections generic, or does it reflect the experience itself? New research suggests the latter.
The visual system is hierarchically organised into different areas. The lower visual areas see small parts of the visual field, and they are sensitive to very simple features, such as edges and their orientation. Higher up the hierarchy, the visual areas start encoding more abstract representations of the world, expanding their visual field to respond to stimuli such as objects and faces.
At the same time, the areas that see “the big picture” send back information to the lower visual areas, called “feedback” connections. Feedback connections are considered essential for integrating contextual information, namely by providing information to lower visual areas about the broader scene, rather than just the small, specific part of the image they are looking at.
For example, a neuron in a higher visual area that responds to tables would send feedback to a neuron in a lower level area that encodes just a part of the table, such as its legs.
However, surprisingly, neurons in higher visual areas of the brain can also sometimes send back information to neurons in lower visual areas that have nothing to do with tables.
But do feedback connections reflect relationships between the "parts" and the "whole " learned from experience?, asks Leopoldo Petreanu, principal investigator of the Champalimaud Foundation's Cortical Circuits lab.
In a previous study in mice, Petreanu and his team had shown that the organization of feedback connections depended on having visual experience.
To do this, they compared the feedback connections in normal mice with those in mice reared in the dark, and found that in the second case, due to the lack of visual experience, the organization of the feedback connections was disrupted. This showed that the organisaton of feedback depended on experience, consistent with the hypothesis that it might reflect a learned relation between the big picture, encoded in high-order areas, and the lower-order features, encoded in the lower areas.
But one thing that was still not clear was whether this organization is passive or what the researchers call “instructive”. In other words, does any experience just trigger the same organization of feedback connections, or do different experiences result in different neural wiring? A new study by the same team, published today (11/02/2026) in the journal Current Biology, aimed at tackling this question.
Another way to phrase the issue is to ask whether feedback connections wire themselves to specific, content-related subsets of neurons in lower visual areas, or whether they do so generically, independently of context. The new study shows that these connections do, in fact, play an instructive role. Their organization is not generic.
Mice with little goggles
To show this, the researchers reared mice fitted with miniature goggles that biased their perception of the visual world. One group of animals only saw edges oriented at a certain angle, while the other saw edges oriented at a different angle. “For some mice, the world looked like elongated lines in some direction, and for the other, it also looked elongated, but in a different way”, says Petreanu.
Using a microscopy technique called dual-color two photon imaging, first co-authors Radhika Rajan and Rodrigo Dias measured the tuning properties and the organisation of feedback inputs from a higher visual area back to a lower one.
What they found is that the animals with different visual experiences had very different tuning properties and patterns of organisation in these feedback connections, shaped by what they had seen since birth. “The feedback connections reflected the visual experience of the mice, supporting the idea that these connections capture associations between visual features, represented in different areas, formed through experience”, says Petreanu.
A way to detect novelty?
In their previous study, the authors noticed something puzzling: feedback inputs often linked neurons that represented unrelated features – neurons that would not normally activate together.
Now they have developed a model that actually explains this counterintuitive property. “With our co-authors from Germany, we tried to determine the rules that would allow us to explain all the changes we see that are induced by experience”, says Petreanu.
In the model, feedforward connections – which carry information from lower to higher visual areas – follow a classic “Hebbian” rule: connections between neurons strengthen when they fire simultaneously ("cells that fire together, wire together"). Feedback connections behave differently. They follow the opposite, “anti-Hebbian” rule, meaning connections weaken when neurons fire together.
This arrangement may help the brain ignore what is familiar and instead highlight what is unexpected. "Once you wire the cortex this way, you create a circuit that detects and amplifies surprising stimuli", Petreanu suggests. "It’s a way of storing knowledge about the world and noticing when something doesn’t fit and needs to be updated".
Original scientific paper from Current Biology.
