Recycling Neurons for Reading

Accesibility: Intermediate-Advanced

Our brains have evolved to be good at certain things: seeing, hearing, learning language, and interacting with other similar brains, to name a few examples. But say you want it to do something new – look at symbols on a page and map them to language. In other words, you want to teach your brain to read. How would you go about doing this? What parts of the brain would you use?

Unless you plan on developing a completely new region, it makes sense to repurpose the brain regions you already have -- a process that neuroscientist Stanislas Dehaene refers to as “neuronal recycling.” This raises the question -- what regions are recycled? And do the regions that get co-opted become worse at their original function?



Dehaene and colleagues explored this question by scanning adults at different levels of literacy: literates, ex-literates (adults who used to be illiterate but learned to read in adulthood), and illiterate adults. They had several interesting findings:

1. They first looked at whether learning to read changes brain activation when looking at words. Not surprisingly, it does. Reading performance was correlated with increased brain activation in much of the left hemisphere language network, including the visual word form area. And this increased activation appeared to be specific to word-like stimuli.

2. During reading, ex-literates have more bilateral activation and also recruited more posterior brain regions. This is similar to what we find in children, who also show more spread out activation while reading. This suggests that unskilled readers recruit a wider set of brain regions as they are learning to read. As readers become more skilled, their brains become more efficient and recruit fewer regions

3. In literate adults, response to checker boards and faces in the visual word form area was lower in the visual word form area compared to non-readers. This suggests that learning to process words may actually be taking resources away from processing other stimuli.

4. The researchers looked more closely at responses to other faces and houses to see how exactly learning to read competed with other visual functions. They found that activation in the peak voxels for faces and houses did not change with literacy. However, activation in surrounding voxels did decrease.

5. And here's an interesting result. Since reading is a horizontal process (at least in the languages they were testing), the researchers checked to see if the visual system became more attuned to horizontal stimuli. They found that literacy enhanced response to horizontal but not vertical checker boards in some primary visual areas.


Dehaene S, Pegado F, Braga LW, Ventura P, Nunes Filho G, Jobert A, Dehaene-Lambertz G, Kolinsky R, Morais J, & Cohen L (2010). How learning to read changes the cortical networks for vision and language. Science (New York, N.Y.), 330 (6009), 1359-64 PMID: 21071632

Read more...

White Matter and Reading Ability

Accessibility:  Intermediate-Advanced

Hello folks.  Things are pretty busy over here and I might be having to review a lot of papers soon, so there's a possibility that entries here will get shorter and a bit more technical.  But we'll see.

Since reading is by nature a multimodal task involving both visual and language regions, it makes sense to look at brain connections in dyslexia. I've written once about white matter in dyslexia, when I blogged Bernard Chang’s PNH study. Today I'll cover two other studies that look at white matter and reading.



As a quick recap, brain tissue is often categorized into gray and white matter. White matter consists mostly of axons, the parts of neurons that send signals to other neurons. Therefore, white matter tracts carry information between brain regions and diffusion tensor imaging is a technique often used to study white matter.  You can take several measures with DTI, but one common one is fractional anisotropy, a measure of the directionality of water diffusion.  You can think of it as a measure of white matter integrity.

In one study, James Andrews and colleagues measured white matter integrity in preterm*  and term children. They found a correlation between reading skill and fractional anisotropy  in the corpus callosum, the large white matter tract that connects the two hemispheres. They also found a trend toward a correlation between reading skill and fractional anisotropy in the left temporal parietal region, a region often associated with reading. I'm surprised by the corpus callosum finding, and wonder its role might be in reading. Is the corpus callosum connecting language regions to their right hemisphere homologues? I also wonder if this is something general to the population, or a difference unique to preterm children. I guess we’ll have to see if this finding comes up in later studies.

Another DTI study found some more predictable results. Rimrodt and colleagues scanned the brains of children with dyslexia and normal-reading children between the ages of seven and 16 years. They found that children with dyslexia had lower FA in the left inferior frontal gyrus and the left temporoparietal region, both areas previously implicated in reading. Interestingly, they also found that the FA in some posterior areas involved in visual word processing (including the left fusiform) were correlated with speeded word reading.

*mean gestational age 30.5 weeks

ANDREWS, J., BEN-SHACHAR, M., YEATMAN, J., FLOM, L., LUNA, B., & FELDMAN, H. (2009). Reading performance correlates with white-matter properties in preterm and term children Developmental Medicine & Child Neurology, 52 (6) DOI: 10.1111/j.1469-8749.2009.03456.x

Rimrodt, S., Peterson, D., Denckla, M., Kaufmann, W., & Cutting, L. (2010). White matter microstructural differences linked to left perisylvian language network in children with dyslexia Cortex, 46 (6), 739-749 DOI: 10.1016/j.cortex.2009.07.008

Read more...

Noise Exclusion Deficits in Dyslexia

Accessibility:  Intermediate-Advanced

The human visual system includes two pathways, magnocellular and parvocellular, deriving from two types of retinal ganglion cells that project to different layers of the lateral geniculate nucleus. Generally speaking, the magnocelluar pathway is specialized for movement while the parvocellular pathway is specialized for color and detail.  Some researchers have found dyslexia to be associated with magnocelluar impairment, although evidence has been mixed.

A paper from Sperling and colleagues argues that magnocelluar deficits in dyslexica may actually be a deficit in noise exclusion.  The authors tested children with and without dyslexia using stimuli that were designed to activate the magnocellular or parvocellular pathways. The magnocellular stimulus was a patch with white bars that alternated rapidly between light and dark. The parvocellular stimulus had thin light and dark bars that did not alternate.



In addition to the two stimulus types, there was a high noise and low noise condition. In the low noise condition, one of the stimuli appeared to the left or right of the fixation mark. In the high noise condition, noise patches appeared on either side of fixation and the stimulus was overlaid onto one of the noise patches. In both cases, child had to say on which side the stimulus appeared.

The authors calculated contrast thresholds (the amount of contrast needed between the light and dark bars for accurate detection) for both groups of children. They found no difference in the contrast thresholds for the low noise condition. In the high noise condition, dyslexic children had higher contrast thresholds (more difficulty detecting) for both the magnocellular and parvocellular stimuli. In addition, thresholds in the high noise condition were correlated with language measures.

These are interesting results. While one study cannot rule out the magnocellular theory of dyslexia, this does open the possibility that many of the results that pointed to a magnocellular deficit were actually cases of noise exclusion deficit.   I do remember one paper about motion perception and dyslexia that can't be explained by noise, so I'll see if I can write about that later.

Another question is, how does noise exclusion lead to dyslexia? It could be that a noise exclusion deficit results in difficulties building phonological categories, which in turn affect reading. The authors also mention that noise exclusion could affect learning in the visual modality by making it harder to extract regularities from different fonts and scripts.


Sperling, A., Lu, Z., Manis, F., & Seidenberg, M. (2005). Deficits in perceptual noise exclusion in developmental dyslexia Nature Neuroscience DOI: 10.1038/nn1474

Read more...

Sensitivity and Specialization in the Occipitatemporal Region: Differences in Dyslexic Children

Accessibility: Advanced/intermediate

Early research on the role of the occipitotemporal region in reading often focused on characterizing a single region in the mid fusiform, commonly called the visual word form area. Since then, focus has gradually shifted from a single region to the entire length of the occipitotemporal region, looking at how the sensitivity and tuning changes as you move from posterior to anterior regions.



Van der mark used an approach like this to look at dyslexic and control children aged 9-12 years. Eighteen normal reading and twenty four dyslexic children performed a phonological lexical decision task in the scanner. Children saw words, pseudohomophones (words that sounded like real words but spelled differently, like “taksi”), pseudowords (pronounceable nonwords), and false fonts. The children were asked to decide whether something sounded like a real word. For example, the correct response would be “yes” for words and pseudohomophones and “no” for pseudowords and false fonts.

The children with dyslexia did worse for pseudohomophones and pseudowords and performed similarly to the controls for words and false fonts.

The authors report two main findings. First, the control children showed a gradient of print specialization in the occipitotemporal region, with more activation to false fonts in posterior regions and more activation to real letters and anterior regions. The control children did not show this trend.

Second, control showed more activation for pseudowords and pseudohomophones than words, while children with dyslexia didn't.

This is a nice study that takes a more nuanced approach to dyslexia brain differences. Brem and colleagues also got similar results with the words and false fonts.

By now there's quite a bit of literature on the specialization of the visual word form area. My own struggle, as I’m also doing this type of research, is the question of what does it all mean? We have all the studies now showing brain differences between control and dyslexic children, but what does it mean to have more or less activation? That the brains of dyslexic children process words differently? I could've told you that before we stared.

So what would help? Perhaps the next step in dyslexia research, now that we've mapped out the basic differences, is to zoom in as much as we can on the relationships between brain differences and behavioral differences. Perhaps more fine grained behavioral measures would help, or more interventional studies that looked at brain activation before and after training. It may also help to look at functional connectivity and how different brain regions interact. Anyone else have ideas?



van der Mark S, Bucher K, Maurer U, Schulz E, Brem S, Buckelmüller J, Kronbichler M, Loenneker T, Klaver P, Martin E, & Brandeis D (2009). Children with dyslexia lack multiple specializations along the visual word-form (VWF) system. NeuroImage, 47 (4), 1940-9 PMID: 19446640

Read more...

Reading induced epilepsy

Accessibility:  Intermediate

Just a short entry today.  Clinical research is not my specialty, but I ran across a case study today on reading induced epilepsy.

Seizures began during silent reading with the feeling of no longer being able to understand what she was reading (a- or dyslexia). After looking up from the page, she then continued to see letters and words despite actual disappearance of that image from either visual field (palinopsia). She had a feeling of strangeness. She could then have right hemi-body jerks and secondary generalisation. Seizures usually occurred soon after the onset of reading (less than 10 min). All seizures occurred during silent reading. She had not abandoned reading altogether but had developed a distinct style of reading to try to avoid the onset of seizures, in that she read only for short periods and tended to scan the page diagonally.

Not surprisingly, clinical tests revealed that these seizures started in the occipitotemporal region.

Gavaret, M., Guedj, E., Koessler, L., Trebuchon-Da Fonseca, A., Aubert, S., Mundler, O., Chauvel, P., & Bartolomei, F. (2009). Reading epilepsy from the dominant temporo-occipital region Journal of Neurology, Neurosurgery & Psychiatry, 81 (7), 710-715 DOI: 10.1136/jnnp.2009.175935

Read more...

fMRI of Letter Processing in Children and Adults

Accessibility: Intermediate-Advanced

How is letter processing different from word processing? Since letters compose words, many reading models have letter processing earlier in the reading stream, but there is still room for more imaging work.

Turkeltaub and colleagues compared the neural basis of letter processing in children (age 6-11) and adults (age 20-22). The participants were scanned while naming either letters or line drawings out loud. Here are four of their findings.



1. Adults had more activation than children in visual regions. This appeared to be driven mostly by differences in letter naming*. This suggests that object processing might be more adult-like in kids at this age.

2. Areas showing a change in letter processing with age were posterior to regions found in other studies to respond to words. Since visual processing moves from back to front, this fits with a model in which letters are processed before words.

3. The authors found no left hemisphere dominance for letters. This very different words, which are heavily left lateralized. This is also different from Cantlon 2010 which did find letter processing to be left lateralized. I wonder if the results here could be different if the authors had used another method to pick their analysis region**.

4. The authors also found that no regions activated more for letters than for objects. This is consistent with what I also find in my data. Objects are more visually complex than letters, so it's not surprising that you get more activation for objects. I should note that Cantlon found regions that responded more to letters than objects, but Cantlon only used shoes, which as a set are more uniform than line drawings of different objects.

*although there is no interaction between objects and letter naming

**ROI selection based on activation for all tasks.


Turkeltaub PE, Flowers DL, Lyon LG, & Eden GF (2008). Development of ventral stream representations for single letters. Annals of the New York Academy of Sciences, 1145, 13-29 PMID: 19076386

Read more...

A Meta-Analysis of Dyslexia Brain Imaging Studies

Accessibility: Advanced

fMRI experiments, with their small sample sizes, can easily fall victim to variability within the subject pool. This is especially true for patient studies. So it’s nice to step back and look at the big picture once in a while, and see where different studies agree and disagree.



Richlan and colleagues recently did meta-analysis of dyslexia brain imaging studies. They used an algorithm called Activation Likelihood Estimation (ALE), which models foci of activation as Gaussian probability distributions. (The software is called GingleALE. Ha.)

Richlan and colleagues picked studies  with the following criteria:
1) Uses words, pseudowords or single letters as stimuli
2) Uses reading or reading related task in the scanner, and
3) Group comparisons are done in a standard stereotactic space.
The studies included PET and fMRI studies.

The take away message is that people with dyslexia underactivate posterior reading regions and may overactivate  frontal regions.

The authors found underactivation in regions associated with the phonological reading pathway (reading by sounding out words), including the superior temporal gyrus and inferior parietal lobule. Interestingly, they found no difference in the angular gyrus, a region that has often been reported to be important to reading.

They also found underactivation the pathway associated with automatic whole word reading, including the left fusiform, inferior temporal and middle temporal regions.

At a less conservative threshold, the authors found that people with dyslexia overactivated the left inferor frontal region. This is typically interpreted as frontal regions being brought in to compensate for posterior reading regions.

They did find one posterior region as well that was overactivated in people with dyslexia : the left lingual gyrus, a lower level visual region. Perhaps again, a case of compensation.

All in all, a nice summary of dyslexia results. Again, I wonder about the relative variablility of dyslexics and controls, and how they affected the results.


Richlan, F., Kronbichler, M., & Wimmer, H. (2009). Functional abnormalities in the dyslexic brain: A quantitative meta-analysis of neuroimaging studies Human Brain Mapping, 30 (10), 3299-3308 DOI: 10.1002/hbm.20752

Read more...