Categories

AI’s New Insights on Dyslexia

Findings about the brain for the educator

For the past 100 years, dyslexia was a murky, confounding disorder of the brain. For the one in five people who has it, dyslexia remained chronic and lifelong, impacting learning and reading. There was no way to peer into a dyslexic brain to see underlying problems precisely enough to correct them.

But now we can. An autonomous AI system now generates online activities for each person to probe brain processes during task performance. This allows the AI system to find and measure inefficiencies in these processes. Because language difficulties underlie dyslexia, this disorder can now be defined operationally in terms of inefficiencies in language processing in the brain.

Functional Map of the Brain

With this AI method, we get a functional map of each person’s brain for the areas of interest, at the level of granularity needed for correction. This new method sidesteps difficulties faced by traditional methods, which focused on the physical structure of the brain.

AI’s functional mapping has yielded new insights on the brain.


Inefficient Language Processes are Common

First, inefficient language processes occur commonly, extending beyond the population with dyslexia. This may explain why around 66% of students fail to meet reading standards when tested in 4th, 8th and 12th grades annually—even though only 20% are supposed to have dyslexia.

Efficiency is computed based on the accuracy of the person’s responses (out of 100) and the speed of delivery/response. Speed is a critical factor since the brain has to process natural language very rapidly for speaking, listening and reading.

Efficiencies in the 50-60% range in areas key to literacy development are not good enough. Users of the AI program registering in this range still display reading difficulties at school and thus have dyslexia. Often, struggling readers register in the 20-30% range.

Nevertheless, corrective training on the AI program can improve efficiencies incrementally to nudge them to the 90-100% target. When they perform consistently in this range, within a two-month window, improvements in spelling first become noticeable, followed by reading and learning. This tells us that, to use language, the brain has to process it highly efficiently.

The Linguistic System is Very Fragile

Second, the linguistic system in the brain is very fragile. Contrary to the widely held assumption in Linguistics, language acquisition does not simply unfold naturally and unproblematically for everyone. The fact that at least one in five people have language issues severe enough to display dyslexia should give us pause.

The data generated by the AI program for dyslexia show that inefficiencies in the linguistic system make it glitchy, cause it to slow down or crash. Overtly, the person may read slowly and haltingly at first. Then misreadings occur more and more frequently with every new sentence, until she gets too overwhelmed and frustrated to continue.

The good news, though, is that once the error can be traced back to its source, the problem can be corrected, to enable the system to run smoothly. This explains why some AI users’ reading scores changed greatly, from under the 25th, 10th or even 1st percentile to above the 50th within one year, as happened in actual cases.

New Methods of Brain Mapping Yield Functional Benefits

To use an analogy, before the invention of sonar, we could not see the topography of the ocean floor. But the invention enabled us to discern it in a totally unexpected but functional way. More than that, today’s technologies can give us crystal-clear maps of the brain.