Brains that can't see in grayscale rely more on color first: Project Prakash study

Brains that can't see in grayscale rely more on color first: Project Prakash study


There are many colors in the world and being able to see them is a great source of joy. But a newborn baby sees the world mostly in black and white. The light-sensitive cone cells in a baby's eyes don't mature until they are about four months old. At this time, the brain uses other visual cues to understand the world.

In May, a team of Indian and American researchers reported in the journal Science This delay in the development of color vision is actually important for overall vision development.

“We are able to explain why normal visual development occurs the way it does,” said study co-lead Preeti Gupta, a cognitive neuroscientist. Hindu,

Dr. Gupta leads the research team for 'Project Prakash' in India, a US initiative of IIT Delhi, Dr. Shroff Charity Eye Hospital, New Delhi, and the Massachusetts Institute of Technology.

Project Prakash treats and rehabilitates blind children in India. These children have helped researchers understand how the brain learns to see.

The importance of color vision

Dr Gupta said humans do not need colour vision to recognise objects, but colours may provide advantages for adaptation and survival.

“If you look at food items, you can identify them without color. But if something has gone bad, you need to look at its color to understand it.”

Yet color vision involves much more than what meets the eye.

“They are very fascinated by colours,” Dr Gupta said of the children of Project Prakash. They often describe objects around them by their colour. “Their reliance on colours is a little more than normal children.” This observation gave the researchers an idea: “They should be shown some objects without colour.”

They brought together a group of visually impaired children and young adults aged 8 to 26 and asked them to identify objects in images – a tree, a bus, a chicken, a stack of books, etc. – first in grayscale and then in color. In another test, they asked the group to determine which of two discs presented to them was lighter in color while the researchers adjusted the color.

The children were able to recognise coloured images and discs quite well – even those who had barely been two days old after eye surgery. But they had difficulty recognising black and white images.

On the other hand, children without any visual impairment had no trouble with either color or grayscale images.

The researchers found that the problem wasn't the blind children's color vision. Rather, the problem was an abnormal development of their vision: an overreliance on color.

Project Prakash team members in India (from left): Abhishek Kumar, Rakesh Kumar, Ajay Chavariya, Preeti Gupta, Shakeela B, Suma Ganesh, Ranu Priya, Navia Lal and Dhun Verma.

Project Prakash team members in India (from left): Abhishek Kumar, Rakesh Kumar, Ajay Chavariya, Preeti Gupta, Shakeela B, Suma Ganesh, Ranupriya, Navia Lal and Dhun Verma. | Photo credit: Himanshu Kumar

Mimicking visual development

Normally, a child first perceives the world in grayscale. But when the Project Light children first experienced normal vision, their eyes had already developed enough to see colors, so they skipped the grayscale stage. As a result, their brains processed black and white images differently.

To understand the implications of this issue, the researchers needed a proxy for the brain that they could tweak to learn in response to different visual stimuli. They set up a deep convolutional neural network (CNN) — a computer program that processes information like neurons in the brain's visual cortex. Engineers have previously used deep CNNs in image recognition software.

“They're not perfect models. But they're the best models people have right now,” Dr. Gupta said.

They trained four CNNs, one each on colored and grayscale images, in a particular order: gray-gray, color-color, color-gray, gray-color.

They found that Gray-CNN recognizes both grayscale and color images better than any other model.

The color-hue model — which most closely mimicked visual development among Project Light's children — performed the worst at identifying grayscale images.

The researchers attributed this to the color-color model's heavy reliance on color cues when examining images because its training data consisted of only color images. The gray-color model had learned enough cues from grayscale images and was thus better able to recognize color images.

These findings explain why the children also performed poorly on color tests. “They already start out with much clearer vision than a normal, developing child,” Dr. Gupta said.

Customizing visual development

Dr. Suma Ganesh, co-author of the study and clinical head of Project Prakash, said it is surprising that the brain develops object recognition and colour perception at different times.

Dr. Ganesh is an eye surgeon, pediatric ophthalmologist and medical director of Dr. Shroff Charity Eye Hospital, New Delhi.

“We never thought that if someone [perception] “When one person is removed, it has a huge impact on others,” he said.

However, it is unclear whether this particular study will have any benefit in rehabilitation.

“Right now, I wouldn't say we have achieved anything significant in terms of rehabilitation,” Dr. Gupta said. “But whatever information we have will eventually be used to develop better treatments.”

But he also said there are ideas that could help. For example, children could also be made to experience a colorless room, simulating a black and white or grayscale environment for a few hours. “Maybe this would help their cells normalize better,” he said.

However, he acknowledged that such 'tests' could also raise ethical concerns and their efficacy would need to be tested on animal models first.

Karthik Vinod is doing internship here Hindu,


join us for quick updates

Leave a Reply

Your email address will not be published. Required fields are marked *

Enable Notifications OK No thanks