New paper published in Nature Communications
What happens to the visual brain after early transient blindness? Our new paper in Nature Communications explores this through the study of a rare population: people born with dense bilateral cataracts— a transient blindness occurring during a critical window of visual development.
We used fMRI to study how these “cataract-reversal” individuals process visual categories — faces, bodies, houses, tools, and words — and compared their brain responses to those of sighted controls. The focus: how early deprivation affects both low-level and high-level visual processing, from EVC (early visual cortex) to VOTC (ventral occipito-temporal cortex).
Representational similarity analyses revealed: • EVC activity patterns matched low-level visual models less in cataract-reversal participants than controls • VOTC showed normal categorical representation comparable to controls. This suggests selective impairment in early visual coding, but preserved category representation downstream.
We then tested within-category decoding: Can the brain tell apart individual images within a category? Cataract-reversal participants showed reduced decoding accuracy in EVC across all categories tested. This indicates a broad low-level visual impairment, not tied to any specific category.
We also tested whether these effects were due to the participants’ current visual quality (e.g., reduced acuity, nystagmus). Using the deepMReye toolbox, we tracked and controlled for eye movements — even in participants with nystagmus — ensuring that our brain results weren’t driven by differences in gaze behavior.
In addition, we ran a control experiment where typical participants viewed altered images to mimic the visual deficits of the cataract group. Result: this degraded vision disrupted both EVC and VOTC — unlike in the cataract-reversal participants, where only EVC was affected.
To dig deeper, we trained deep neural networks on degraded visual input to model how vision develops after early deprivation. The networks mirrored our brain findings: – Early layers (V1-like) stayed impaired – Higher layers (VOTC-like) recovered normal categorical coding
In sum Brief blindness right after birth permanently alters low level visual processing in EVC — but higher-level areas (VOTC) remain surprisingly resilient. This means that although early deprivation disrupts basic feature encoding (edges, spatial frequencies), the brain can develop normal categorical representations (faces, bodies, tools…) later in life.
Together, these results reveal different sensitive periods across the visual system: • Early visual areas need input early on to develop normally • Higher-level regions can recover from early loss if later visual experience is rich enough
This challenges the classic “cascade” model of visual development — the idea that if early vision is impaired, higher areas must fail too. Instead, our data show that the ventral visual stream can reorganize and recover.
The message? The brain is both fragile and resilient. A brief blindness at birth leaves lasting traces — but doesn’t prevent the emergence of complex visual categorisation.
This work was only possible thanks to the dedication of an incredible team — across labs, countries, and expertise. A huge thank you to all the co-authors who made this project possible.