Boston/ Science, Tech & Medicine
AI Assisted Icon
Published on May 24, 2024
MIT Study Reveals How Early Visual Limitations Boost Brain's Object Recognition AbilitiesSource: Unsplash/ Robina Weermeijer

Researchers at MIT have discovered a remarkable trait of the human brain: it can accurately identify objects in pictures without color, and this knack for recognition may stem from early visual limitations. The study, led by MIT's Pawan Sinha, delved into why some individuals, particularly those born blind who regain sight, struggle more with black-and-white images. The findings, released in Science, suggest that our brains get a developmental head start by initially processing less color information.

It seems that, before the retina and cortex fully mature, newborns predominantly rely on brightness to differentiate objects – a necessary adaption due to the limited color vision at birth, "This general idea, that there is something important about the initial limitations that we have in our perceptual system, transcends color vision and visual acuity," Sinha told MIT News. He notes a similar pattern in the development of the auditory system. The study also throws light on the challenges faced by children who recover from congenital blindness and immediately experience rich color input, leading to an overdependence on color for object recognition.

The research came from observations in Project Prakash, an initiative founded by Sinha in India in 2005, which seeks to treat children with reversible forms of blindness. MIT postdocs Marin Vogelsang and Lukas Vogelsang, Project Prakash scientist Priti Gupta served as lead authors of the study, together with Sidney Diamond, a retired neurologist and MIT research affiliate. They found the children with sight restored were more challenged when asked to recognize objects in black-and-white photos as opposed to their normally sighted counterparts, who faced no such obstacle when identifying desaturated images.

Leveraging computational modeling with the neural network AlexNet, the team simulated infant vision development by initially exposing a model to grayscale imagery and only later introducing color pictures. This approach seemed to reflect how a child's vision progresses, building a resilient recognition capability independent of color. The research showed that models trained with an early diet of black and white visuals were adept at identifying objects regardless of the image being color or grayscale. In contrast, models feasting only on colored images failed to perform as well when faced with the absence of hue, "What happens is that this Prakash-like model is very good with colored images, but it’s very poor with anything else," Lukas Vogelsang explained to MIT News.

This study underscores the potential advantages of early sensory limitations in visual and possibly other developmental areas. It raises intriguing possibilities for understanding object recognition, parsing the developmental effects on the brain, and could have implications for the optimal timing and methods of visual rehabilitation for the congenitally blind. Funding for this research came from the National Eye Institute of NIH and the Intelligence Advanced Research Projects Activity, confirming the significance of continued investment in cognitive science and healthcare innovation.

Boston-Science, Tech & Medicine