The root cause of dyslexia, debated for decades, is poor connectivity in language regions in the brain, not the inability of the brain to "represent" day-to-day sounds, a new study  in the 6 December issue of Science reports.
A computer network provides a relevant metaphor in this debate: For a long time, it was thought the information stored in the brain or "server" was somehow degraded in dyslexia, perhaps because it was poorly recorded during development. Now, the researchers from Belgium show that the information on the server is intact but the connection to access this information is too slow, or even degraded.
At a 5 December press briefing held in Leuven, Belgium, the Science researchers emphasized that the significant differences found in connectivity should be taken into account when designing the most appropriate intervention techniques for people with dyslexia.
"With this new knowledge, it is not unconceivable that we could design more focused and effective interventions that target improving the specific connection between frontal and temporal language regions," said Bart Boets, a clinical psychologist and postdoctoral research fellow at KU Leuven. "Though still a theoretical idea, it should be possible to stimulate two brain regions at the same time…and by firing together, the regions will wire together. So essentially, this would improve the relationship between the two regions."
The researchers compared dyslexic adults with normal adults. When the latter listen to spoken words or read text in a book, they mentally map the related sounds onto internal phonemes, little neural pegs that categorize distinct sounds and help make them interpretable. But people with dyslexia - more than ten percent of the world's population - have trouble with this process.
Some scientists suggest that representations of phonemes are fuzzy in the dyslexic brain (several unrelated sounds may be associated with one phoneme). Another theory is that phonetic representations are intact in people with dyslexia, but are hard to access by other brain regions involved in language processing.
"The two hypotheses are very difficult to disentangle," said Boets. "This is because behavioral tasks [to study dyslexic brains] always tap both the representation [of the sound] and the access to this representation. We needed neuroimaging to assess the two in isolation."
"Initially," Boets explained, "our aim was to finally objectively demonstrate that the quality of phonetic representations is impaired in individuals with dyslexia." He and his colleagues used functional magnetic resonance imaging to scan the brains of 22 normal and 23 dyslexic adults as these individuals listened to explicit speech sounds made of vowels and consonants.
As the subjects responded to certain speech stimuli, Boets and colleagues looked at patterns of nerve activity in their brains using a technique known as multi-voxel activity analysis, or MVPA. They noted how consistently different sounds were mapped to their related phonetic representations.
"I was so convinced that we would observe degraded phonetic representations in the dyslexic participants," Boets said. "Yet, their representations turned out to be perfectly intact." Indeed, to the researchers' surprise, phonetic representations in dyslexic readers were just as well-specified, perhaps even more so, than those in normal readers.
Boets and his team thus performed a second analysis to explore whether connectivity in the brain differed in the two groups. They assessed how easily 13 regions involved in language processing could connect to the same phonetic representations they'd analyzed before. They found connectivity to be significantly hampered between certain language regions in the brains of dyslexics. The worse the connection, the poorer a dyslexic individual performed on reading, spelling and other tests.
Taken together, these findings suggest that deficient access to phonetic representations is at the heart of dyslexia. "Apparently, it is not sufficient to have information [such as phonetic representations] available or stored," Boets explained. "In order to apply this information, one also needs fluent access to it."
"It was by combining and applying two different analysis techniques on the same data set that we were able to draw these controversial conclusions," Boets said, explaining the leap forward he and his group have made.
He cautioned that further studies would be needed to confirm their finding holds true in other samples, for example, in dyslexic children.
"We scanned a group of 5-year-old children with a hereditary risk for dyslexia before they started learning to read," said the study's co-author Pol Ghesquière at the briefing, "and we will scan them again at different stages during their reading development in order to sketch the development of the neural reading network. This is a very exciting new scientific adventure. The ultimate goal is to determine early neural markers of dyslexia. If we could do that, we could detect dyslexia at a much younger age and also develop better interventions."