Evidence Suggesting that Specialized Visual Regions Are Formed by Pruning in Early Childhood

There are quite a few specialized visual regions in the brain. For example, the fusiform face area (FFA) activates for faces, and the visual word form area (VWFA) in the left fusiform is consistently active for words.

How do these specialized cortical regions develop? Is it experience dependent? Do regions have a preexisting preference for certain visual features? (For example, perhaps the visual word form region prefers high contrast stimuli with sharp borders). Do these regions form by increasing activation to preferred stimuli, or a decreasing activation to nonpreferred stimuli? Cantlon and colleagues investigated these questions in a recent study.



They tested prereading five year olds and adults in an fMRI experiment. Participants saw faces, letters, numbers, shoes and scrambled images and pressed a button if a green border appeared around the picture. There were two interesting findings.

The first concerned the visual word form area. Both adults and children had a specialized brain region in the left fusiform that activated more for letters than objects. However, while adults activated that region more for letters than for numbers, children had equally high activation for letters and numbers.*

These results support a role for both experience and low level visual features in the development of the visual word form area. Note that these children are nonreaders, but they already activate the left fusiform for letters and numbers. So perhaps there’s something hardwired in the left fusiform that prefers symbol-like, high contrast, visual stimuli. But only adults, who have had extensive experience with letters, show differential activation for words and numbers.

The authors then investigated the relationship between activation level and behavior. They tested children on a face matching task and a letter naming task. Contrary to what you might expect, activation in the fusiform face area did not correlate with face matching skill, and activation in the visual word form area did not correlate with letter naming skill.

Rather, skill was negatively correlated with activation to the nonpreferred category. Face matching performance was inversely correlated with FFA activation to shoes. And letter naming was inversely correlated with VWFA activation to faces. This suggests that that increased skill in face and letter recognition is associated not with enhancing activation to preferred stimuli, but with pruning back activation to unrelated stimuli. **

*Methodological note: ROI selection, 10 strongest voxels within a sphere 10mm radius around peaks of All>scrambled.

**Note that not all nonpreferred stimuli show this inverse correlation. In the face area, there was no correlation between face skill and symbols, and in the VWFA, there is no correlation between letter naming skill and shoe activation. Perhaps these nonpreferred stimuli too far from the preferred stimulus, so no pruning is needed?



Cantlon JF, Pinel P, Dehaene S, & Pelphrey KA (2010). Cortical Representations of Symbols, Objects, and Faces Are Pruned Back during Early Childhood. Cerebral cortex (New York, N.Y. : 1991) PMID: 20457691

Read more...

Multimodal Investigation of Reading in Children: More from Brem and Colleagues

Accessibility: Advanced

Last time we read an article from Brem and colleagues that compared word processing in adolescents (age 15-17) and adults (19-30). In follow-up paper from 2009, Brem expanded the report to include children (9-11).

If you didn’t read the last post, it’s probably a good idea to do that first. I won’t repeat any of the methodological details or background information here, just gonna make few quick notes on their results.



The 2006 paper found that adolescents had higher N1 amplitude than adults. Here, Brem reports that children have an even higher N1 amplitude with adolescents, thus suggesting a steady decrease in N1 amplitude from age 9 onwards.

For all groups, the N1 amplitude was higher for words than symbols. However, the difference between words and symbols declined with age. At first, I found this counterintuitive. I would have expected the opposite, with kids treating words and symbol similarly and the word/symbol difference getting larger as they matured and became better readers. The kids in this study, however, have already been reading for a few years. Perhaps they’re at the stage where they can process the words but are less efficient in doing so, thus resulting in a higher N1 amplitude for words than symbols.

On the fMRI front, Brem found the same  posterior to anterior gradient in the fusiform gyrus, with posterior regions being more responsive to symbols, and anterior regions being more responsive to words. There didn’t seem to be any difference between age groups there.

Brem also increases that a higher signal in anterior fusiform is correlated with slower reading.
(This is opposite of what was reported in other paper, perhaps I’m misreading the paper.)

There were some discrepancies between EEG and fMRI results. The N1 ERP component shows clear difference between words and symbols, but the fMRI analysis doesn’t show differences in the occipital temporal region, the calculated source of the N1. The could be due to temporal resolution. The N1 component only lasts about 100 ms.  EEG has good enough temporal resolution to pick up on the difference, but fMRI may not.


Brem S, Halder P, Bucher K, Summers P, Martin E, & Brandeis D (2009). Tuning of the visual word processing system: distinct developmental ERP and fMRI effects. Human brain mapping, 30 (6), 1833-44 PMID: 19288464

Read more...

Developmental Changes in Word Processing After Adolescence

When does brain development for reading stop? We often focus on school aged children, but what about the later teen years? To answer this question, Brem and colleagues tested adolescents (age 15-17) and adults (19-31) in a study using fMRI and EEG.



Participants were presented with words and symbols strings and asked to detect repeats. It’s an easy task, so it’s not surprising that the two groups had equal reading accuracy and speed. However, there were brain differences.

Brem focused on two early ERP components. The P1 component, a positive peak at 100 ms, is sensitive to low level stimulus characteristics like luminance and size. Brem found that this component had a higher amplitude for symbol strings and for words in both groups.

The N1 component occurs later (140-220ms) and is sensitive to higher level factors like stimulus category. Brem found that the later part of the N1 component was more pronounced to words than symbol strings. Source localization on the N1 component  found that the early part of the N1 localized to the temporal parietal occipital junction, while the late N1 localized to the left fusiform.

There were differences between the two groups. Adolescents had higher P1 and N1 amplitudes than adults. The N1 latency also became faster with age for words but not symbol strings.

Brem also used fMRI to look at the spatial organization of the fusiform gyrus*. Posterior fusiform regions responded more to symbol strings than words, while anterior regions responded more to words than symbol strings.

The left fusiform region seems to be related to reading skill. Bigger N1 amplitude was correlated with fewer mistakes in a reading test.  Higher fMRI signal in the anterior fusiform was correlated with faster reading.

It’s interesting that despite similar behavior between groups, brain measures still differ. I do wonder about differences within the adults as well. 19-31 is a pretty big range, so I'd like to see what happens after age 18.

*Using five regions of interest. 6 mm spheres based on Taleraich coordinates.


Brem S, Bucher K, Halder P, Summers P, Dietrich T, Martin E, & Brandeis D (2006). Evidence for developmental changes in the visual word processing network beyond adolescence. NeuroImage, 29 (3), 822-37 PMID: 16257546

Read more...

Brief Introduction to ERP Components

Accessibility: Basic

EEG (electroencephalography) uses scalp electrodes to measure electrical field potentials that result from brain activity. Many EEG studies focus on event related potentials (ERP), patterns of activity that occur in response to a stimulus or cognitive event (Hence, they’re “event related.”).

Usually, an experimenter averages the brain response over many trials to achieve adequate signal to noise ratio. The end result is a waveform representing the average pattern over trials of a certain type. Peaks and troughs in waveform are known as components. While the naming of components isn’t systematic, they are often named with a letter (P if it’s a peak in the positive direction and N if it’s in the negative direction), and a number that either corresponds to the approximate time of the peak or its order of appearance. Commonly studied components include the N400 and P300. To learn more about how ERPs are used in research, take a look at entries with the EEG label.

Read more...