New Delhi: When the congenital cataracts that had occluded Subhash Kumar Rawat’s vision, limiting his world to a milky white blur, somewhat akin to what you’d see if you walked around with half-cut table tennis balls over your eyes, were removed, the only thing he could vaguely recognize in the room in which he found himself was the ceiling fan.
Though he couldn’t distinguish the shape of the fan, or its colour, or the ceiling of the room for that matter, he could sense a motion somewhere above him. A fan, as he’d been taught through 13 years of near total blindness, rotated with a certain fixed regularity and its blades made a distinctive susurrating noise as they cut through the air. The motion he detected corresponded to the direction from which a breeze wafted down; and with the sound that he’d learned to associate with a fan. It must, he surmised, be a fan.
The chair at the other end of the room posed an even greater challenge. There was nothing in his previous, primarily tactile experience of the chair that would help him correlate the set of colours that he now saw in front of him with the chair. The same applied to the faces he saw around him. A brave new world of sight was opening up for Subhash, but in this explosion of colours, pixels and textures, he was lost. For a brief moment, he says, he wanted to shut his eyes.
New vision: MIT’s Pawan Sinha (wearing a tie) with (from left) Rajdeep Kumar, Bablu Kumar Rawat and Subhash Kumar Rawat, who were treated under the research programme. Ramesh Pathania / Mint
Four months after his operation, Subhash smiles shyly as he sits chatting with Pawan Sinha, associate professor of computational and visual neuroscience in the department of brain and cognitive sciences at the Massachusetts Institute of Technology (MIT)—the man responsible for getting him treated.
“What colour is that shirt?” asks Sinha, pointing to a man standing next to them. Subhash peers quizzically through spectacles perched jauntily on his nose. “White,” he says triumphantly. “Does your brother Bablu look like what you expected him to look like?” persists Sinha mischievously. Subhash looks at Bablu for a confused second; and then both burst out laughing.
In the space of a few months, Subhash has learnt an astounding variety of skills. He can now detect different types of motion, match shapes, name colours and recognize faces. What were once random agglomerations of pixels have now started resolving themselves into objects that are co-related with his previous tactile and auditory experiences. His perception of depth is developing; and where once he couldn’t walk without assistance, Subhash can now cross a road on his own.
Ironically, till a few years ago, someone like him might never have been treated. Conventional medical wisdom held that a child deprived of visual stimulus for the first six years of their lives, would never learn to see. The lack of visual input in those crucial years, it was thought, led to irrevocable under-development of the optic nerves and the brain’s visual cortex.
“What we’ve found however,” says Sinha, his excitement revealing itself in the broad grin that spreads across his face, “is that even children as old as 14 years, can recover significant visual functions.”
During interactions on his occasional visits to India, Sinha realized that there were hundreds, probably thousands of children like Subhash who suffered from curable congenital problems such as cataracts and opacities. But a lack of awareness and the absence of medical facilities meant that most of these children were never treated.
It led to Project Prakash, a unique research project that combines groundbreaking science on the process of visual learning with the humanitarian task of getting thousands of visually disabled children treated.
The scientific opportunity was thrilling. Studying the processes of visual learning and development is difficult; studies on newly born children, the obvious subjects of such a study were complicated by the fact that their brains are developing at the same time. But here was a large population, old enough to have fairly well developed brain functions but young enough to learn, that was ready to start out on the journey of sight.
How did Subhash learn to see so much so fast? In what order did his visual skills develop? Could the pattern of his development help create a post-surgery rehabilitation protocol for similar children? Could it help us build a computational object discovery system?
These were a few of the questions that Sinha set out to answer.
He collaborated with Dr Shroff’s Charity Eye Hospital in Delhi, one of the oldest charitable hospitals in the country and one of the few with a paediatric ophthalmology division.
The hospital had a strong outreach programme and a network of primary clinics and hospitals in Rajasthan, Haryana and Uttar Pradesh. But they soon realized that rather than waiting for these children to be brought to the clinics, the chances of which were very slim, they would have to go in search of them.
The first point of call they decided would be blind schools. These schools are supposed to screen children for curable conditions, but in practice, says Arun K. Arora, the chief executive of Dr Shroff’s Charity Eye Hospital, their screening is cursory. There was a fair chance that they’d find curably blind children here.
In parts of the country such as Mewat, a predominantly tribal district in Haryana, where intermarriages have led to a much higher than normal incidence of congenital diseases, the plan was to go door to door.
In 2009, Project Prakash screened 20,500 children in Mewat, Jaipur, Jodhpur and Saharanpur. Of these, 573 were referred to vision clinics and the secondary hospitals for vision aids and minor treatments, and 80 were brought to New Delhi for sight-restoring surgeries. “That’s a huge number of curably blind children from such a small area,” says Arora.
Contrary to expectations the children’s parents were often reluctant to have them treated, fearing that their condition would worsen. “We’d have to cajole them,” confides Naval Kumar Chauhan, an outreach worker with Dr Shroff’s Charity Eye Hospital. Apart from arranging transportation, the team compensated the parents, most of whom were daily wage workers, for wages lost because of the trip to New Delhi.
A day after the operation, Sinha and his group of researchers from MIT, would put the children through a battery of simple tests to assess their visual acuity, colour and shape detection skills.
In nearly every case, they found that the first thing that children like Subhash noticed, was motion. In tests, the children wouldn’t spot a static triangle or circle, but would recognize it the moment it started moving across the computer screen.
“The algorithm seemed to be,” explains Sinha, “what moves together, goes together.”
Sounds simple, but it was counterintuitive. After all, would it not be easier to recognize objects first and then track their motion?
“The importance of dynamic motion,” says Sinha emphatically, “has been the single important finding of our work so far.”
It has formed the basis of his efforts to create a computational system that can learn object concepts. The first system he created, based on results early on in the project, used sudden transitions in pixel colours to delineate what were termed “salient” patches, precursors to objects. That, Sinha now admits, was a mistake. The new system he is working on relies primarily on dynamic motion to incrementally “create” objects.
This finding might also have implications for treatment of autistic children, whose motion-tracking abilities are underdeveloped. When they try to track a ball moving across a computer screen, their eyes trail the ball, rather than predict where it’s going to be next and focusing on that position.
Immediately after their operations, the children at Project Prakash do the same thing. But in the space of a few months they learn to predict motion. “If we can figure out how that happens,” says Sinha, “we could try teaching the skill to autistic children.”
With each new patient Sinha gets a little closer to some of the answers that he is searching for, but he knows that he has a very long way to go. “How do these children’s previous auditory and tactile experiences get mapped onto their visual experiences?” he asks, pausing for a second before answering his own question, “I don’t know.”