The brain mechanisms of dyslexia

Listen to audio

The course concluded with a brief examination of how these networks are set up in children as they learn to read. We began by recalling the three-stage model proposed by Uta Frith. In the first, logographic or pictorial stage, the shape of a few words is memorized, with no possibility of generalization. In the second, phonological stage, the grapheme-phoneme conversion pathway is put in place. Characterized by the effects of syllabic length and complexity on reading times, this stage sees the emergence of phonological awareness. Finally, a third orthographic stage sees the gradual emergence of expertise in spelling a language, with a gradual disappearance of the length effect and the appearance of morphological and lexical effects.

At the cerebral level, few studies have been carried out in children, and their results are hardly stable. However, in line with the emergence of orthographic expertise, visual activation seems to increase with learning, gradually focusing on the left occipitotemporal region between the ages of 6 and 10. In studies by the Yale group (Shaywitz et al.), the amount of left occipitotemporal activation increases with age and predicts reading scores. However, this region is not irreplaceable, and great plasticity exists during development, since it was possible for a young girl who had undergone left occipitotemporal ablation at age 4 to learn to read normally by using the symmetrical region of the right hemisphere.

In dyslexic children, the left occipitotemporal region does not develop normally and is severely under-activated. However, this anomaly may be merely a consequence of the deficit in learning to read. Depending on the study, other regions may play a more fundamental or even causal role: left temporal and left inferior parietal hypo-activation, and left inferior frontal hyper-activation. The dominant hypothesis is that of an anomaly in phonological processing. In support of this hypothesis, diffusion tensor and gray matter density abnormalities are observed opposite temporal regions under-activated during reading. The possibility of an early abnormality in neuronal migration, initially suggested by a few post-mortem studies, has recently been reinforced by the genetic approach. At least four candidate dyslexia susceptibility genes have been identified, three of which (KIAA0319, DCDC2 and DYX1C) affect neuronal migration (work by Joe Lo Turco et al.), with the fourth (ROBO1) playing a role in axonal pathway control.

Fortunately, research in this field is beginning to show that these disorders can be partially remedied by appropriate rehabilitation. Various strategies focusing on phonology and grapheme-phoneme correspondences are bearing fruit insofar as they improve both behavior and brain activation. Reading scores improve by one or two standard deviations on average, and activation increases in regions close to those initially under-activated, and in the temporal and right frontal regions. Rehabilitation would therefore induce both a restoration of activity and a partial compensation of deficits by other brain regions that remained intact.

The lecture concluded by examining a possibility that seems somewhat neglected in current research - that the visual component of reading is also at the root of some of the learning difficulties encountered by children. According to the neural recycling hypothesis, while some properties of the inferotemporal cortex are advantageous to children's learning to read (such as spatial invariance), others could be detrimental. The sub-temporal cortex has evolved to recognize objects regardless of their orientation. In the macaque monkey, inferotemporal neurons generalize in mirror image and respond in the same way to shapes such as p and q. We can therefore assume that this invariance by symmetry must be "unlearned", and that the difficulty of this unlearning would explain the errors in mirror reading and writing in children, as well as their persistence in some dyslexic children. Experimental data on this question remain fragmentary, but they suggest that all children pass through an early stage when they are able to read and write in mirror image. This competence for left-right symmetry remains present later in dyslexic children than in normal children. In a few rare, well-studied cases, mirror over-generalization and spatial errors do indeed appear to be at the root of a reading deficit.

In conclusion, it cannot be denied that the development of phonological awareness plays a fundamental role in the acquisition of reading, and that an anomaly in phonological processing seems to be at the root of reading difficulties in many dyslexic children (or, at the very least, to be an excellent indicator of the extent of these difficulties). However, learning invariant visual recognition of letters and words is also an essential step in the acquisition of fluent, expert reading. For some children, this visual learning and unlearning of symmetrical generalization could be the primary source of dyslexia.