Understanding mirror invariance inhibition

Writing can be considered one of the greatest inventions of Homo Sapiens, allowing humans to “communicate their thoughts” throw visual support (i.e., by writing & reading).  When children learn to read and write, very often they present a specific visual difficulty with letters: mirror confusion (e.g., b=d). In the past, this mirror confusion was thought to be a sign of dyslexia (reading difficulty). In fact, mirror confusion for letters emerges as a natural consequence of a deeply rooted visual mechanism in our visual system: the mirror invariance mechanism. This mechanism is useful for the recognition of faces, objects, and places, from both the left and right perspectives. This is probably why it was also selected by evolution in other primates, cats, pigeons, and even octopus’s visual systems. However, letters have a single orientation and should not be mirror inversed. Further, there are even mirror-letters in the Latin alphabet (b vs d, p vs q). Thus, mirror discrimination for letters should be learned for fast letter recognition and thus for fluent reading. But traditionally, schools do not provide such specific learning and the correct orientation of letters is learned sparsely, with time.

We have previously shown the relationship between mirror discrimination for letters and reading fluency: better readers show faster mirror discrimination for letters. We also demonstrated that a brain region specialized to recognize visual letters exhibits, in skilled readers, spontaneous mirror discrimination for letters but preserved mirror invariance for images in general. Also, this effect of literacy on mirror discrimination peaks at an early stage of brain responses (< 150 milliseconds), suggesting automaticity of mirror discrimination acquired with reading expertise. Then, we postulated hypothetical neural and cognitive mechanisms of how literacy acquisition inhibits mirror invariance for letters.

In our new study*, we performed a “reverse engineering” of this relationship between reading fluency and mirror discrimination. We applied a specific targeted training to address mirror confusion for letters in 1st graders learning to read. The training was based on our hypothetical mechanism: “multisystem mappings”, i.e., mappings of letter representation across sensory-motor systems.  To keep the training short (7.5 hours in total, across 3 weeks), we associated a physiological learning enhancer: post-training sleep consolidation. We observed massif effects of mirror discrimination learning for letters and a critical role played by sleep consolidation in the magnitude, automaticity, and long-term duration of this specific learning. More importantly, well-consolidated mirror discrimination learning for letters unleashed the reading fluency potential of children that could now read two-times faster than their controls.

Now, I want to better understand the exact mechanisms, critical factors, and the neural basis of such specific learning including via advanced non-invasive neuroimaging techniques and protocols such as the ones I developed with Prof Hans Op de Beeck and Prof Bart Boets during my FWO postdoc fellowship at KU Leuven. I also want to test if dyslexics could benefit from such training because they often present mirror confusion for longer than usual. Given the coronavirus crises, we are obliged to temporarily interrupt our research program at school. But we adapted and we are now starting to test if we can efficiently provide the training via online courses directly for children with their parents at home (https://bdpq.fr/).

The final goal of this research program is both to understand mirror invariance inhibition for reading and to deliver a “cognitive vaccine” able to prevent mirror errors during reading acquisition. I hope it could largely benefit children during the precious moment of learning to read.

*refers to the paper:

“Selective inhibition of mirror invariance for letters consolidated by sleep doubles reading fluency" accepted for publication in Current Biology.