Brain Struct Funct (2018) 223:701–712

https://doi.org/10.1007/s00429-017-1516-x

ORIGINAL ARTICLE

How interindividual differences in brain anatomy shape reading

accuracy
Arnaud Cachia1,2,3,4,6 • Margot Roell1,2,5,6 • Jean-François Mangin6,7,8 •
Zhong Yi Sun6,7,8 • Antoinette Jobert7 • Lucia Braga9 • Olivier Houde1,2,3,6 •

Stanislas Dehaene5,6,10 • Grégoire Borst1,2,3,6

Received: 26 April 2017 / Accepted: 7 September 2017 / Published online: 15 September 2017
Ó Springer-Verlag GmbH Germany 2017

Abstract The capacity to read develops throughout the left lateral occipito-temporal sulcus (OTS) hosting the
intensive academic learning and training. Several studies visual word form area (VWFA) predicts reading skills in
have investigated the impact of reading on the brain, and adults; that (b) this effect is modulated by the age of
particularly how the anatomy of the brain changes with reading acquisition; and that (c) the length of the OTS
reading acquisition. In the present study, we investigated sulcal interruption is associated with reading skills.
the converse issue, namely whether and how reading Because the sulcal pattern is determined in utero, our
acquisition is constrained by the anatomy of the brain. findings suggest that individual difference in reading skills
Using multimodal MRI, we found that (a) the pattern can be traced back to early stages of brain development in
(continuous or interrupted sulcus) of the posterior part of addition to the well-established socioeconomic and edu-
cational factors.

Arnaud Cachia and Margot Roell have equally contributed to this Keywords Reading
VWFA
MRI
Neurodevelopment

study. SES
& Arnaud Cachia
arnaud.cachia@parisdescartes.fr
Introduction
1
Laboratory for the Psychology of Child Development and
Education, CNRS UMR8240, Sorbonne, 46 Rue The capacity to read, contrary to spoken language, devel-
Saint-Jacques, 75005 Paris, France
ops throughout intensive academic learning and training.
2
Paris Descartes University, Sorbonne Paris Cité, Paris, France Investigating how the brain is modified by the acquisition
3
Institut Universitaire de France, Paris, France of literacy is crucial to clarify the factors that contribute to
4
Biomarkers of Brain Development and Disorders, Center of this cultural acquisition and its individual differences pat-
Psychiatry and Neurosciences, INSERM UMR894, Paris, terns (Dehaene et al. 2015). In expert literate adults of all
France cultures, reading relies on a left-lateralized network of
5
Cognitive Neuroimaging Unit, CEA DSV/I2BM, INSERM, brain areas that includes the inferior frontal/precentral gyri,
NeuroSpin Center, Université Paris-Sud, Université Paris- the dorsal temporoparietal circuit, and the ventral occipito-
Saclay, Gif-sur-Yvette, France
temporal pathway (Houde et al. 2010; Martin et al. 2015).
6
Ecole des Neurosciences de Paris (ENP), Paris, France Particularly, a region in the left lateral occipito-temporal
7
UNATI, Neurospin, CEA, 91191 Gif-sur-Yvette, France sulcus (OTS), termed the ‘‘visual word form area’’
8
CATI Multicenter Neuroimaging Platform, Neurospin, cati- (VWFA) (Cohen and Dehaene 2004), is consistently and
neuroimaging.com, 91191 Gif-sur-Yvette, France specifically more activated in response to visually pre-
9
SARAH Network-International Center for Neurosciences and sented words in the learned script than to other categories
Rehabilitation, QL 13, Lago Norte, 71, Brasilia, DF 535-005, of stimuli (Dehaene and Cohen 2011).
Brazil Reading acquisition affects how visual and phonological
10
Collège de France, Paris, France information is coded and processed, and induces functional

123
702 Brain Struct Funct (2018) 223:701–712

and structural changes in the brain (Dehaene et al. 2015). suggests that their presence biases the subsequent emer-
For instance, a combined electro-encephalogram gence of functional specialization in the cortex. Another
(EEG)/functional magnetic resonance imaging (fMRI) study provided preliminary evidence that the oral reading
longitudinal study in 6-year-old children revealed that skills in sixteen 8-year-old children (Borst et al. 2016) are
training with grapheme–phoneme correspondences leads to correlated with the sulcal pattern (continuous or interrupted
the rapid emergence of a selective activation of the VWFA sulcus) of the left lateral OTS, the sulcus in which the
(Brem et al. 2010). Reading acquisition also induces VWFA appears during the course of learning to read.
structural changes in the brain: a voxel-based morphometry Importantly, interindividual variations in sulcal patterns
(VBM) study on children aged 9–10 revealed that reading presumably reflect the early brain differences that predate
skills were related to gray matter volume in the VWFA, the the acquisition of a cognitive skills, since the sulcal pattern
anterior part of the left inferior occipital gyrus and the left is determined in utero (Mangin et al. 2010) and their major
thalamus (Simon et al. 2013). In addition, a longitudinal features remain unaffected by neuroplastic processes
MRI study (Myers et al. 2014) showed that an increase in occurring during brain maturation after birth (Cachia et al.
white-matter volume in the left temporoparietal region 2016).
between kindergarten and Grade 3 predicts reading ability In the present study, we aimed to investigate whether
at Grade 3, over and above other measures (i.e., family the sulcal pattern of the left lateral OTS is predictive of
history, socioeconomic status, and pre-reading cognitive reading skills in a large sample of adults of variable
skills). Finally, a diffusion MRI (dMRI) study on children socioeconomic status (SES) and age of reading acquisi-
aged 7–12 demonstrated that the microstructure of white- tion. In addition, we examined whether the position of the
matter fibers connectivity between the VWFA and the sulcal interruption was of consequence. Given that the
anterior and medial temporal lobe increases with reading VWFA is consistently located in the posterior part of the
proficiency and age (Yeatman et al. 2012). left lateral OTS (Dehaene and Cohen 2011; Vinckier et al.
Importantly, the age of reading acquisition may modu- 2007), we reasoned that an interruption in the posterior
late the impact of reading acquisition on the brain. Com- part of left lateral OTS (referred hereafter as ‘posterior
parisons of ex-illiterate adults, who learned to read during interruption’) should have a greater effect on reading
adulthood, with literate adults, who learned to read during fluency than an interruption in the anterior part of the left
their childhood, suggest that reading relies on the same lateral OTS (referred hereafter as ‘anterior interruption’).
brain network in both populations (Dehaene et al. In addition, we evaluated whether the effect of the sulcal
2010, 2015; Pegado et al. 2014), but that structural changes pattern of the OTS on reading fluency varies with the age
in the brain induced by reading acquisition are generally of acquisition of reading (i.e., childhood vs adulthood).
smaller when reading acquisition occurs during adulthood To this aim, we compared whether the effect of the
than during childhood (Carreiras et al. 2009; Thiebaut de interruption of left lateral OTS on reading fluency was
Schotten et al. 2014). The acquisition of literacy in adult- similar for literate adults who learned to read during
hood has been associated with (a) an increase of gray childhood, and for ex-illiterate adults who learned to read
matter in the bilateral angular, dorsal occipital, middle during adulthood. Finally, we directly tested our hypoth-
temporal, left supramarginal, and posterior superior tem- esis that the OTS sulcal pattern is a robust feature that is
poral gyri, (b) a thickening of the splenium or the isthmus not affected by reading acquisition. To this aim, we
of the corpus callosum (Carreiras et al. 2009; Castro-Cal- compared the distribution of interrupted and continuous
das et al. 2009), and (c) an increase in fractional anisotropy left lateral OTS in literate, ex-illiterate, and illiterate
and a decrease in perpendicular diffusivity in the tem- adults. If the macroscopic sulcal pattern of the brain truly
poroparietal portion of the left arcuate fasciculus (Thiebaut reflects the early cerebral constraints on cognitive devel-
de Schotten et al. 2014). opment, then it should not be affected by reading acqui-
While most studies focused on brain changes induced by sition, as suggested by a recent study indicating stability
reading acquisition, a few studies investigated whether during development (Cachia et al. 2016).
individual difference in pre-readers’ brain structure affects
their reading skills later in life. Using diffusion MRI, a
longitudinal study on 14 children revealed that the white- Method
matter connectivity of the VWFA at age 5, before children
learned to read, predicts the location of VWFA peak acti- Participants
vation in fMRI at age 8, after children learned to read
(Saygin et al. 2016). This study fits with the observation Sixty-two Portuguese and Brazilian participants were
that the major fiber tracts of the human brain are present recruited for this study and divided into three groups
since infancy (Dubois et al. 2006, 2008, 2014, 2016), and according to their literacy profile. The sample included 10

123
Brain Struct Funct (2018) 223:701–712 703

illiterate, 20 ex-illiterate, and 32 literate adults. All subjects Anatomical MRI acquisition and analysis
were right-handed as estimated using the Edinburgh
handedness test (Oldfield 1971). Full demographic infor- Structural MRIs were acquired with a 3 T Siemens
mation is reported in Table 1. MAGNETOM Trio MR scanner. A sagittal high-resolution
The illiterate participants were from rural areas or urban structural MRI was obtained for all participants (magneti-
centers in the Brasilia vicinity. Participants were illiterates zation prepared rapid gradient echo, 160 slice T1-weighted
for social reasons, with no history of special difficulty other image, TR 2300 ms, TE 2.98 ms; flip angle 9°, TA
than the lack of access to schools. The illiterate participants 7.46 min, resolution 1 9 1 9 1.1 mm). These MRIs were
did not attend school at all as children. The families of all adapted for sulcus segmentation required for the three-di-
these participants were originally from rural areas, and mensional reconstruction of the fine individual cortical
their parents had been illiterate rural workers. The ex-il- folds.
literates group includes Brazilian and Portuguese partici- Analysis of anatomical data was performed with the
pants who had backgrounds similar to those of the Morphologist toolbox using BrainVISA 4.2 software
illiterates (illiterate parents from rural areas). Like the with standard parameters (http://brainvisa.info/). We
illiterates, ex-illiterates had not attended school during used an automated pre-processing step to skull-strip the
childhood, but they differed in having attended adult lit- T1 MRIs and to segment the brain tissues. The automatic
eracy courses. Finally, the literate group comprised some segmentation of the cortical folds throughout the cortex
participants from the same social community as the illit- was based on the skeleton of the gray matter/cere-
erate group, but with 2–7 years of early education, as well brospinal fluid mask (Mangin et al. 2004). This proce-
as other participants with higher socioeconomic status and dure yields a stable and robust sulcal surface definition
education [for details, see Supplementary Materials in our that is not affected by variations in the cortical thickness
previous study (Dehaene et al. 2010)]. or the gray-matter/white-matter contrast. At each pro-
SES was dichotomized into two levels (‘low’ and cessing step and for each MRI, images were visually
‘high’) based on participant’s occupation: for instance, checked and we observed no motion artefact and no
‘street vendor’ and ‘rural worker’ were classified as ‘low’ segmentation error.
SES, while ‘system analyst’ and ‘geologist’ were classified
as ‘high’ SES. Classification of OTS sulcal pattern
All participants gave their written informed consent (the
consent form was read aloud and explained to illiterates). We first classified the lateral OTS sulcal pattern in each
participant based on Ono’s classification (Borst et al. 2016;
Reading test Ono et al. 1990). Briefly, a three-dimensional mesh-based
reconstruction of the cortical folds was visually inspected
Prior to the MRI session, all participants underwent a to classify the sulcal pattern of the lateral OTS. All MRI
battery of neuropsychological tests to assess their reading data were anonymized, and manual labelling of OTS in left
skills. As a measure of reading performance, the number and right hemispheres was carried out blind to the partic-
of words read per minute was calculated. The word ipant’s demographic characteristics (literacy group, gender,
reading task comprised 36 words (12 simple, 12 complex, SES, etc). The sulcal patterns of the left and right OTS
and 12 irregular) to be read aloud. For illiterates, who were independently classified by two of the co-authors
failed to identify even some single letters, the measure (MR & AC) using a dichotomous variable code for the
was set to 0. continuity or interruption of the sulcus. Reliability was

Table 1 Demographical data

Illiterate (N = 10) Ex-illiterate (N = 20) Literate (N = 32)
and reading performance
Age (years) 53.40 (2.83) 49.75 (2.00) 48.75 (1.58)
Gender (F/M) 05/05 13/7 16/16
SES level (high/low) 0/10 0/20 17/15
Early education (years) 0.21 (0.39) 0.24 (0.69) 8.65 (3.46)
Late education (years) 0.61 (1.05) 2.08 (1.22) 2.37 (1.95)
Reading fluency (words per minute) 0 (–) 21.25 (5.90) 96.43 (4.66)
Early (resp. late) education corresponds to the years of education received by the participant before (resp.
after) the age of 18. Quantitative data are represented as mean (SD)

123
704 Brain Struct Funct (2018) 223:701–712

100% (j = 1) among the two raters for the left and the space using only linear transformations, thus avoiding
right hemispheres. potential biases resulting from the shape deformations
The sulcal pattern of the lateral OTS was characterized that may occur during a non-linear warping process. The
as ‘interrupted’ when the lateral OTS had interruptions and length of the sulcal interruption was decomposed in two
‘continuous’ otherwise. In addition, to investigate the components: an antero-posterior gap and a left–right gap
possible effect of the position of the left OTS interruption, (see Fig. 3).
we identified whether OTS interruption was located in the
posterior part of the sulcus hosting the VWFA (Dehaene Statistical analysis
and Cohen 2011; Vinckier et al. 2007) (‘posterior’ inter-
ruption), or in the anterior part of the sulcus (‘anterior’ Continuous variables (e.g., number of words or pseu-
interruption) (see Fig. 1). We used an anatomical criterion, dowords read per minute, OTS interruption length, etc)
namely the Y-coordinate of the posterior extremity of the were analyzed using linear models. When a significant
brainstem (PEB), as a limit to define the anterior and main effect was detected, analysis was continued by post
posterior interruptions of the left and right OTS. We con- hoc paired comparisons using Tukey’s HSD test with
sidered this anatomical criterion and not the position of the Bonferroni correction. Categorical variables (e.g., ‘contin-
VWFA, because such functional criterion could only be uous’ or ‘interrupted’ OTS) were analyzed using binomial
used for the left OTS and in readers (literates and ex-il- and multinomial generalized linear models.
literates) but not for the right OTS or the non-readers. The We included SES and gender as confounding covariates,
functional validity of this anatomical criterion was a pos- because SES was previously shown to have an effect on
teriori confirmed for the left OTS (see Fig. 2; Supple- reading skills and brain anatomy (Farah et al. 2006; Noble
mentary materials). and McCandliss 2005; Fluss et al. 2009; Noble et al. 2007;
Finally, we measured the length of the OTS inter- Jednorog et al. 2012) and gender on sulcal anatomy
ruption based on the coordinates of the extremities of (Duchesnay et al. 2007). Estimates of the mean values of
the anterior and posterior OTS interruptions. OTS continuous variables were, therefore, linearly adjusted on
interruption length was calculated after normalization to SES and gender.
MNI space, thus controlling for global brain size dif- Main effects and interactions were probed with F tests
ferences due, for instance, to differences in age or in the linear models, Chi-squared tests in the generalized
gender. The T1 MRI was spatially normalized to MNI linear models, and with t tests in the paired post hoc

Fig. 1 Sulcal interruptions of the occipito-temporal sulcus (OTS). blue). The posterior extremity of the brainstem (dashed line) was used
Examples of a posterior interruption (on the left) and an anterior as a limit to define the anterior and posterior interruptions of the OTS
interruption (on the right) of the left lateral OTS (sulci are depicted in

123
Brain Struct Funct (2018) 223:701–712 705

Fig. 2 Sulcal pattern of the left

occipito-temporal sulcus (OTS)
and position of the Visual Word
Form Area (VWFA). 3-D mesh-
based reconstructions of the
white-matter surface (in semi-
transparent white) with left OTS
(in blue). Individual VWFA
peak activation is represented
with a purple sphere. The mean
position of the VWFA in MNI
space (-44; -50; -14) based
on coordinates provided in the
literature (Dehaene et al. 2010)
is represented with a green
sphere

analyses. A two-tailed p value of less than 0.05 was Results

considered statistically significant. The relative impor-
tance of each factor in the linear models was estimated Effect of OTS sulcal pattern on reading
using the ‘lmg’ metric, also known as ‘‘hierarchical
partitioning’’, which provides a decomposition of the Literates
total variance (adjusted R2) and has the advantage to
robustly adjust for other factors in the model (Grömping We first analyzed across the group of literate subjects
2015). whether the sulcal pattern of the lateral OTS was associated
All the statistical analyses were carried out with R 2.9 with reading fluency as measured by the number of words
software (http://www.r-project.org/) and ‘car’, ‘effects’, correctly read per minute. To this aim, we entered fluency
‘nnet’, ‘multcomp’, and ‘relaimpo’ libraries. in a linear model with five categorical factors: the sulcal

123
706 Brain Struct Funct (2018) 223:701–712

To check whether SES level had a differential effect

depending on OTS sulcal pattern, analyses were ran again
including four additional interaction terms between SES
level and the OTS anterior and posterior interruptions in
both right and left hemisphere. No interaction effect was
found between any factor (p [ 0.4).
To estimate the relative importance of the sulcal pattern
of the left posterior lateral OTS and SES on reading flu-
ency, we reran a linear model with only these two factors.
The new linear model revealed that the two predictors
accounted for 64.5% of the variance in the number of
words correctly read, F(2, 29) = 26.43, p \ 0.0001,
adjusted R2 = 0.645, with the sulcal pattern of the left
posterior lateral OTS explaining 4.25% and SES 60.25% of
the variance.

Differences between literates and ex-illiterates

We then investigated whether the effect of the left posterior

OTS sulcal pattern on reading fluency differed between
literates and ex-illiterates. We, therefore, ran the regression
Fig. 3 Sulcal interruption of the left posterior OTS. The length of the model on the data from literate and ex-illiterate subjects
sulcal interruption was decomposed in two components: an antero- (excluding illiterates), and included a categorical group
posterior gap (Dy) and a left–right gap (Dx)
factor (‘literate’ vs ‘ex-illiterate’) and its interactions.
Analyses revealed a significant main effect of group,
pattern of the lateral posterior OTS in the left hemisphere F(1,41) = 58.47, p \ 0.00001, as well as a main effect of
(‘continuous’ vs ‘interrupted’), the sulcal pattern of the SES, F(1,41) = 54.83, p \ 0.00001, on reading fluency.
lateral anterior OTS in the left hemisphere (‘continuous’ vs Importantly a significant interaction between group and left
‘interrupted’), the lateral posterior OTS in the right hemi- OTS posterior interruption was detected, F(1,41) = 8.81,
sphere (‘continuous’ vs ‘interrupted’), the lateral anterior p \ 0.005, indicating that the effect of left OTS sulcal
OTS in the right hemisphere (‘continuous’ vs ‘inter- pattern on reading fluency was different in literate and ex-
rupted’), and SES (‘low’ vs ‘high’). This analysis revealed illiterate groups. Post hoc Tukey revealed that reading
that the number of words correctly read was associated fluency was found to be significantly greater (T = 2.67,
with the sulcal pattern of the left posterior lateral OTS p = 0.01) in participants in the literate group with a pos-
(p \ 0.025) and SES (p \ 0.0001), but not with the left terior interruption in the left OTS (102.90 ± 6.74 words
anterior lateral OTS (p = 0.49), right posterior lateral OTS read per minute) compared to literate participant with a
(p = 0.63), nor with right posterior lateral OTS (p = 0.64). continuous posterior left OTS (79.21 ± 4.62 words read
Post Hoc Tukey’s tests showed that participants with a per minute), while no significant difference was found
posterior interruption in the OTS had significantly better (T = 0.49, p = 0.9) between participants with
reading fluency scores (mean ± SD 113.01 ± 7.67 words (37.02 ± 6.52 words read per minutes) or without
per minute) than participants that had a continuous OTS (41.50 ± 7.04 words read per minutes) interruption in the
(89.94 ± 4.60) (t = 2.53; p = 0.02) (see Fig. 4). Impor- posterior part of the left OTS in the ex-illiterate group (see
tantly, no other variables regarding OTS sulcal pattern (i.e., Fig. 5).
anterior interruption of OTS in left hemisphere, anterior Again, these findings were specific to the sulcal pattern
and posterior interruption of OTS in right hemisphere) had of the left anterior OTS, as complementary analyses
any significant effect on reading fluency. These results involving the left posterior, right anterior, or right posterior
replicate and extend a previous study in children (Borst OTS pattern did not reveal any significant interaction with
et al. 2016) by showing that the interruption of the OTS the group factor on reading fluency, all ps [ 0.2).
affects oral reading skills not only in children but also in
adults, and additionally that the effect of the OTS inter- Effect of OTS interruption length on reading
ruption on oral reading skills is restricted to interruption of
the posterior part of the left lateral OTS, hosting the We further investigated the effect of left posterior OTS
VWFA. anatomy on reading skills by analyzing the effect of the

123
Brain Struct Funct (2018) 223:701–712 707

Fig. 4 Effect of OTS sulcal interruption on reading ability. The posteriorly (‘below’ the back dashed line) or anteriorly (‘above’ the
number of words reads correctly on average in literate adults with black dashed line). Error bars depict one standard error of the mean,
continuous (in plain gray) or interrupted (in hatched gray) OTS in left *p \ 0.05. Data were linearly adjusted on gender and SES levels
and right hemispheres. The sulcal interruption can be located

Fig. 5 Differential effect of posterior sulcal interruption of OTS on continuous (in gray) or interrupted (in white) OTS in left hemisphere
reading ability in literates and ex-illiterates. The number of words (inside VWFA). Error bars depict one standard error of the mean,
reads correctly on average in literate and ex-illiterate adults with *p \ 0.05. Data were linearly adjusted on gender and SES levels

123
708 Brain Struct Funct (2018) 223:701–712

length of the sulcal interruption on reading fluency of Discussion

readers (i.e., literates and ex-illiterates). The analysis
focused on readers (i.e., literates and ex-illiterates) with a Several studies have investigated the impact of reading on
posterior interruption in the left lateral OTS (N = 20). the brain, and particularly how the anatomy of the brain
We ran a linear model with the number of words read changes with reading acquisition (Dehaene et al. 2015).
correctly as the outcome variable and the antero-posterior The aim of the present study was to investigate the con-
gap and the left–right gap as continuous variables. Note verse issue, namely whether and how reading acquisition is
that SES was not entered as variable in the model, since all constrained by the anatomy of the brain. We found that the
subjects but three literates were ‘low’ SES. The number of sulcal pattern of the left OTS, a marker of early neurode-
words correctly read was associated with the antero-pos- velopmental constraints, predicted reading skills in adults.
terior gap (F(1,17) = 4.02, p = 0.007) but not with the These results replicate and extend in a larger, older, and
left–right gap (F(1,17) = 0.00, p = 0.99). As shown in more variable sample our preliminary findings found in a
Fig. 6, the greater the antero-posterior gap, the greater the small sample of sixteen 8-year-old children (Borst et al.
number of words correctly read (R2 = 0.36). 2016). Specifically, adults who had an interrupted left OTS,
in particular in its posterior portion hosting the visual word
Effects of reading on OTS sulcal pattern form area (VWFA), had better reading fluency than adults
who had a continuous left OTS. Furthermore, we found a
Finally, to determine whether learning to read modifies the linear association between the length of the posterior
sulcal pattern of the lateral OTS, we evaluated the literacy interruption of the left OTS and reading skills, namely the
effect on the distribution of interrupted and continuous greater the interruption, the greater the number of words
posterior left lateral OTS (see Table 2). Multinomial and correctly read per minute. Our results also suggest that the
binomial regression analyses with SES and gender as position of the sulcal interruption of the OTS is critical:
confounding covariates revealed that the distribution of the only OTS interruption located in the posterior part of the
sulcal interruption of the left posterior OTS (‘continuous’ sulcus, hosting the VWFA, but not its anterior part,
vs ‘interrupted’) was similar in literates, ex-illiterates, and affected reading fluency.
illiterates (v = 3.95, df = 2, p = 0.14), and did not differ In addition, comparison of adults who learned to read
(v = 2.83, df = 1, p = 0.09) between readers (literates during childhood (literates) and adults who learned to read
and ex-illiterates) and non-readers (illiterates). For these during adulthood (ex-illiterates) revealed that age of
two analyses, main effects of SES and gender were non- reading acquisition modulates the effect of OTS sulcal
significant (ps [ 0.17). pattern on reading skills: interruption of the posterior left

Fig. 6 Left posterior OTS

sulcal interruption length and
reading skills. Linear
association between the length
of the sulcal interruption of the
posterior OTS (antero-posterior
gap in mm) and the number of
words read per minute. Only
subjects with a sulcal
interruption on the left posterior
OTS were represented (i.e.,
antero-posterior gap [0)

123
Brain Struct Funct (2018) 223:701–712 709

Table 2 Distribution of the posterior and anterior sulcal interruptions of the lateral occipito-temporal sulcus (OTS) in the left and right
hemispheres in non-readers (illiterate) and readers (ex-illiterates and literates)
Non-readers (N = 10) Readers (N = 52) Group difference* Non-readers vs readers*
Illiterates (N = 10) Ex-illiterates (N = 20) Literates (N = 32)

Left hemisphere
Posterior
Interrupted 2 11 9 v = 3.95; p = 0.14 v = 2.83; p = 0.09
Continuous 8 9 23
Anterior
Interrupted 5 12 19 v = 0.17; p = 0.91 v = 0.14; p = 0.71
Continuous 5 8 13
Right hemisphere
Posterior
Interrupted 5 9 9 v = 1.50; p = 0.47 v = 0.08; p = 0.77
Continuous 5 11 23
Anterior
Interrupted 7 12 12 v = 5.46; p = 0.07 v = 0.76; p = 0.38
Continuous 3 8 20
*Distribution difference between groups (illiterates vs ex-illiterates vs literates), or between non-readers and readers, was evaluated using
multinomial, or binomial, generalized linear models with gender and SES as confounding covariates

lateral OTS affected reading skills in literates but not in ex- matter bundles and the tension they exert constrain cortical
illiterates. This finding probably reflects the fact that the folding (Hilgetag and Barbas 2006; Van Essen 1997;
level of reading competence in our sample of ex-illiterates Dehay et al. 1996). More specifically, the sulcal pattern of
was highly variable, from good fluency to near-illiteracy, the left OTS could be associated with oral reading accuracy
and that the primary determinant of this variability is via the connectivity between the VWFA and the anterior
probably the highly variable duration, intensity, and quality and medial temporal lobe (Yeatman et al. 2012; Bouhali
of the reading education that those adults received (De- et al. 2014). This interpretation is consistent with a recent
haene et al. 2010). Thus, this negative finding does not study that showed that the white-matter connectivity of the
exclude that an effect of the anatomical OTS sulcal pattern VWFA at age 5, before children learned to read, is pre-
would eventually be observed on the asymptotic perfor- dictive of VWFA location at age 8, after they learned to
mance of ex-illiterates adults, once they received extensive read (Saygin et al. 2016). A larger interruption of the left
education—just like the effect of genetic variability on OTS might confer an advantage in reading by reflecting an
school-based achievements increases with age, once underlying difference in the size of white-matter bundles,
developmental and educational variability recedes (Plomin indicating a higher number, greater fiber diameter, and/or
and Deary 2015). greater myelination in the bundles connecting the VWFA
Finally, the similar distribution of the left OTS sulcal (Bouhali et al. 2014). This interpretation is supported by
interruption found in readers and non-readers provides the previous studies showing that the connectivity between
preliminary evidence that the qualitative sulcal pattern is the VWFA and the anterior and medial temporal lobe in
robust to neuroplastic processes underlying brain devel- 7–12-year-old children increases with reading efficiency
opment and education, in line with our previous study and age (Yeatman et al. 2012), and that the anisotropy of
reporting a stability during the development of the quali- the posterior part of the arcuate fasciculus varies with lit-
tative sulcal pattern of the anterior cingulate cortex (ACC) eracy (Thiebaut de Schotten et al. 2014). Another not
(Cachia et al. 2016). We, therefore, propose that differ- necessarily incompatible hypothesis is a putative increase
ences in reading skills in individuals with interrupted or in VWFA surface area. OTS sulcal interruption might,
continuous left OTS are likely to arise from the early stages therefore, be an indirect marker of an increase in the
of brain development, when the major sulcal patterns amount of cortical tissue and connections available for
appear. We suspect that the relation between the sulcal neuroplastic processes underlying reading acquisition.
pattern of the left OTS and reading accuracy might be a Finally, cytoarchitectonic differences may also contribute
consequence of the association between sulcal pattern and to the observed functional differences in reading skills in
white-matter connectivity, whereby the thickness of white- participants with continuous or interrupted left OTS.

123
710 Brain Struct Funct (2018) 223:701–712

Indeed, several studies suggest not only a relationship home literacy factors such as number of books available in
between cortical folding and cerebral function, but also the household and parental involvement in the school
with the underlying cytoarchitecture (Fischl et al. 2007; (Aikens and Barbarin 2008). SES-related differences in
Zilles et al. 2013). In particular, one study showed that the reading tend to be stark at the initial kindergarten assess-
sulcal interruption of left posterior OTS is located within ment and continue to grow through school (Aikens and
the cytoarchitectonic area 4 of the lateral fusiform gyrus Barbarin 2008). This would be due to the compounding
(FG4) which overlaps the VWFA (Weiner et al. 2017), effect of low-SES environments: low-SES children grow up
supporting, therefore, a correspondence between cytoar- in home environments poor in literacy and enter schools
chitectony (FG4), functional segregation (VWFA), and that have higher proportion of poor children, both factors
sulcal folding pattern (OTS interruption). In the future, it being associated with poorer reading outcomes (Xue and
will be important to examine whether this model also Meisels 2004). Parental SES can also affect an individual
applies to other cognitive domains. There are suggestions from very early development in utero (Hackman et al. 2010;
that the plasticity in FG2, where face and word responses Tomalski and Johnson 2010). Indeed, stress, nutrition,
emerge during development, is greater than in other regions parental care, and cognitive stimulation mediate the impact
such as the area FG3, which responds to places (Weiner of SES on both brain structures and cognitive functions
et al. 2017; Gomez et al. 2017). It thus seems possible that across development (Raizada and Kishiyama 2010; Hack-
the neurodevelopmental mechanisms that constrain VWFA man and Farah 2009; Jednorog et al. 2012). Fourth, a clear
function differ from other cortical regions whose selectiv- limit is that we did not quantify the percentage of the
ity may be less dependent on experience (Saygin et al. VWFA cluster located within or outside of the OTS sulcal
2016). interruption or the selectivity of the voxels within vs. out-
Several issues of the present study call for caution when side the interruption. Indeed, this information would have
interpreting the results. First, we only studied a single provided a more direct test of the possible effect of the left
macroscopic variable, sulcal interruption, but other more OTS interruption on the functional activation of the VWFA.
subtle sulcal patterns may not be so strictly immutable. In our study, we identified the VWFA-related cluster of
Indeed, although the major patterns are in place at birth, each participant using a coarse-to-fine approach (see Glezer
changes may still be observed in the tertiary gyration during and Riesenhuber 2013 for a similar rationale) in which we
infancy (Leroy et al. 2011). Thus, it remains to be seen determined the nearest local maxima in regards to the mean
whether quantitative (e.g., surface, depth, length, etc) and position of the VWFA in MNI space (-44; -50; -14,
smaller scale measures of sulcal anatomy than the one Dehaene et al. 2010) from a whole-brain analysis of the
studied here might be impacted by education to literacy, individual contrast ‘written strings’ vs ‘checkers’. Because
numeracy, or other domains such as music (Mongelli et al. this contrast elicited activations with very variable peak
2017). Second, although our findings are suggestive of a intensities from one participant to the other—likely due to
causal role of sulcation in determining later reading skills, a the high variability of participants’ reading fluency, early
direct causal link has yet to be evidenced. To provide such and late education as well as SES (see Table 1)—we used a
evidence, a longitudinal study would be needed to demon- participant-specific statistical threshold to detect the
strate that pre-reading sulcation constrains future reading VWFA-related cluster (see Supplementary Materials), thus
skill in the same individuals—similar to the longitudinal preventing us from comparing the spatial extent of this
demonstration that infero-temporal connectivity constrains cluster between participants. Note that analyses based on
VWFA location (Saygin et al. 2016). Third, it is also worth the cluster peak position, as the one run in the present study,
remembering that the impact of the OTS sulcal pattern were used in the previous studies investigating the relation
accounted for only *4% of the variability in reading flu- between fMRI activations and individual sulcal morphol-
ency, as compared to *65% for SES. The latter variable is ogy, e.g., in visual areas (Dumoulin et al. 2000), orbito-
likely to summarize several other relevant variables such as frontal cortex (Li et al. 2015), or anterior cingulate cortex
presence of books in the family, quality, and duration of (Amiez and Petrides 2014).
schooling, etc. The previous studies have established a In conclusion, we provided evidence that the sulcal
strong correlation between SES and language skills, pattern of the posterior part of the left lateral OTS, hosting
including vocabulary, literacy, phonological awareness, and the VWFA which is consistently activated in good readers
syntax (Farah et al. 2006; Noble and McCandliss 2005; during the visual presentation of written words, affects
Fluss et al. 2009; Noble et al. 2007; Jednorog et al. 2012). reading fluency in adults. In particular, adults with an
For instance, socioeconomically disadvantaged children are interruption in the left posterior lateral OTS had better
at risk for reading difficulties (Whitehurst and Lonigan reading fluency than adults with continuous left lateral
1998; Lee and Burkam 2002; Evans 2004). SES family OTS. This effect was modulated by the age of reading
environment also impacts reading skills through different acquisition with a stronger effect of the sulcal pattern of the

123
Brain Struct Funct (2018) 223:701–712 711

left posterior lateral OTS on reading fluency for literate Dehaene S, Pegado F, Braga LW, Ventura P, Nunes Filho G, Jobert
than ex-illiterate adults. In addition, the length of the A, Dehaene-Lambertz G, Kolinsky R, Morais J, Cohen L (2010)
How learning to read changes the cortical networks for vision
interruption of the left posterior lateral OTS was associated and language. Science 330(6009):1359–1364. doi:10.1126/
with reading fluency, with a longer interruption leading to science.1194140
better reading fluency. While sulcation—an aspect of brain Dehaene S, Cohen L, Morais J, Kolinsky R (2015) Illiterate to literate:
anatomy which is primarily affected by early neurodevel- behavioural and cerebral changes induced by reading acquisi-
tion. Nat Rev Neurosci 16(4):234–244. doi:10.1038/nrn3924
opmental factors—predicts subsequent cognitive develop- Dehay C, Giroud P, Berland M, Killackey H, Kennedy H (1996)
ment, it is worth remembering that socioeconomic and Contribution of thalamic input to the specification of cytoarchi-
educational factors have a much greater, indeed, dominant tectonic cortical fields in the primate: effects of bilateral
effect, on ultimate reading fluency, and are also the ones on enucleation in the fetal monkey on the boundaries, dimensions,
and gyrification of striate and extrastriate cortex. J Comp Neurol
which we may intervene by modifying our educational 367(1):70–89
policies. Dubois J, Hertz-Pannier L, Dehaene-Lambertz G, Cointepas Y, Le
Bihan D (2006) Assessment of the early organization and
Acknowledgements We thank S. F. Pegado, P. Ventura, G. Nunes maturation of infants’ cerebral white matter fiber bundles: a
Filho, G. Dehaene-Lambertz, R. Kolinsky, J. Morais, and L. Cohen feasibility study using quantitative diffusion tensor imaging and
for participant recruitment. tractography. Neuroimage 30(4):1121–1132
Dubois J, Dehaene-Lambertz G, Perrin M, Mangin JF, Cointepas Y,
Author contributions AC, GB, SD, and OH designed the study. AJ, Duchesnay E, Le Bihan D, Hertz-Pannier L (2008) Asynchrony
LB, and SD provided MRI and cognitive data. MR, AC, GB, AJ, of the early maturation of white matter bundles in healthy
J-FM, and ZYS analyzed the data. AC, GB, SD, OH, J-FM, and MR infants: quantitative landmarks revealed noninvasively by diffu-
interpreted the results and wrote the manuscript. sion tensor imaging. Hum Brain Mapp 29(1):14–27. doi:10.
1002/hbm.20363
Dubois J, Dehaene-Lambertz G, Kulikova S, Poupon C, Huppi PS,
Hertz-Pannier L (2014) The early development of brain white
References matter: a review of imaging studies in fetuses, newborns and
infants. Neuroscience 276:48–71. doi:10.1016/j.neuroscience.
Aikens NL, Barbarin O (2008) Socioeconomic differences in reading 2013.12.044
trajectories: the contribution of family, neighborhood, and school Dubois J, Poupon C, Thirion B, Simonnet H, Kulikova S, Leroy F,
contexts. J Educ Psychol 100(2):235 Hertz-Pannier L, Dehaene-Lambertz G (2016) Exploring the
Amiez C, Petrides M (2014) Neuroimaging evidence of the anatomo- early organization and maturation of linguistic pathways in the
functional organization of the human cingulate motor areas. human infant brain. Cereb Cortex 26(5):2283–2298. doi:10.
Cereb Cortex 24(3):563–578. doi:10.1093/cercor/bhs329 1093/cercor/bhv082
Borst G, Cachia A, Tissier C, Ahr E, Simon G, Houdé O (2016) Early Duchesnay E, Cachia A, Roche A, Riviere D, Cointepas Y,
cerebral constraint on reading skills of 10-years-old children. Papadopoulos-Orfanos D, Zilbovicius M, Martinot JL, Regis J,
Mind Brain Educ 10(1):47–54 Mangin JF (2007) Classification based on cortical folding
Bouhali F, Thiebaut de Schotten M, Pinel P, Poupon C, Mangin JF, patterns. IEEE Trans Med Imaging 26(4):553–565. doi:10.
Dehaene S, Cohen L (2014) Anatomical connections of the 1109/TMI.2007.892501
visual word form area. J Neurosci 34(46):15402–15414. doi:10. Dumoulin SO, Bittar RG, Kabani NJ, Baker CL Jr, Le Goualher G,
1523/JNEUROSCI.4918-13.2014 Bruce Pike G, Evans AC (2000) A new anatomical landmark for
Brem S, Bach S, Kucian K, Guttorm TK, Martin E, Lyytinen H, reliable identification of human area V5/MT: a quantitative
Brandeis D, Richardson U (2010) Brain sensitivity to print analysis of sulcal patterning. Cereb Cortex 10(5):454–463
emerges when children learn letter-speech sound correspon- Evans GW (2004) The environment of childhood poverty. Am
dences. Proc Natl Acad Sci USA 107(17):7939–7944. doi:10. Psychol 59(2):77
1073/pnas.0904402107 Farah MJ, Shera DM, Savage JH, Betancourt L, Giannetta JM,
Cachia A, Borst G, Tissier C, Fisher C, Plaze M, Gay O, Riviere D, Brodsky NL, Malmud EK, Hurt H (2006) Childhood poverty:
Gogtay N, Giedd J, Mangin JF, Houde O, Raznahan A (2016) specific associations with neurocognitive development. Brain
Longitudinal stability of the folding pattern of the anterior Res 1110(1):166–174. doi:10.1016/j.brainres.2006.06.072
cingulate cortex during development. Dev Cogn Neurosci Fischl B, Rajendran N, Busa E, Augustinack J, Hinds O, Yeo BTT,
19:122–127. doi:10.1016/j.dcn.2016.02.011 Mohlberg H, Amunts K, Zilles K (2007) Cortical folding
Carreiras M, Seghier ML, Baquero S, Estevez A, Lozano A, Devlin patterns and predicting cytoarchitecture. Cereb Cortex. doi:10.
JT, Price CJ (2009) An anatomical signature for literacy. Nature 1093/cercor/bhm225
461(7266):983–986. doi:10.1038/nature08461 Fluss J, Ziegler JC, Warszawski J, Ducot B, Richard G, Billard C
Castro-Caldas A, Nunes MV, Maestu F, Ortiz T, Simoes R, (2009) Poor reading in French elementary school: the interplay
Fernandes R, de La Guia E, Garcia E, Goncalves M (2009) of cognitive, behavioral, and socioeconomic factors. J Dev
Learning orthography in adulthood: a magnetoencephalographic Behav Pediatr 30(3):206–216. doi:10.1097/DBP.
study. J Neuropsychol 3(Pt 1):17–30. doi:10.1348/ 0b013e3181a7ed6c
174866408X289953 Glezer LS, Riesenhuber M (2013) Individual variability in location
Cohen L, Dehaene S (2004) Specialization within the ventral stream: impacts orthographic selectivity in the ‘‘visual word form area’’.
the case for the visual word form area. Neuroimage J Neurosci 33(27):11221–11226. doi:10.1523/JNEUROSCI.
22(1):466–476. doi:10.1016/j.neuroimage.2003.12.049 5002-12.2013
Dehaene S, Cohen L (2011) The unique role of the visual word form Gomez J, Barnett MA, Natu V, Mezer A, Palomero-Gallagher N,
area in reading. Trends Cogn Sci 15(6):254–262. doi:10.1016/j. Weiner KS, Amunts K, Zilles K, Grill-Spector K (2017)
tics.2011.04.003 Microstructural proliferation in human cortex is coupled with

123
712 Brain Struct Funct (2018) 223:701–712

the development of face processing. Science 355(6320):68–71. abilities. Dev Sci 10(4):464–480. doi:10.1111/j.1467-7687.
doi:10.1126/science.aag0311 2007.00600.x
Grömping U (2015) Variable importance in regression models. Wiley Oldfield RC (1971) The assessment and analysis of handedness: the
Interdiscip Rev Comput Stat 7(2):137–152 Edinburgh inventory. Neuropsychologia 9(1):97–113
Hackman DA, Farah MJ (2009) Socioeconomic status and the Ono M, Kubik S, Abarnathey CD (1990) Atlas of the cerebral sulci.
developing brain. Trends Cogn Sci 13(2):65–73. doi:10.1016/j. Georg Thieme, New York
tics.2008.11.003 Pegado F, Comerlato E, Ventura F, Jobert A, Nakamura K, Buiatti M,
Hackman DA, Farah MJ, Meaney MJ (2010) Socioeconomic status Ventura P, Dehaene-Lambertz G, Kolinsky R, Morais J, Braga
and the brain: mechanistic insights from human and animal LW, Cohen L, Dehaene S (2014) Timing the impact of literacy
research. Nat Rev Neurosci 11(9):651–659. doi:10.1038/nrn2897 on visual processing. Proc Natl Acad Sci USA 111(49):E5233–
Hilgetag CC, Barbas H (2006) Role of mechanical factors in the E5242. doi:10.1073/pnas.1417347111
morphology of the primate cerebral cortex. PLoS Comput Biol Plomin R, Deary IJ (2015) Genetics and intelligence differences: five
2(3):e22 special findings. Mol Psychiatry 20(1):98–108. doi:10.1038/mp.
Houde O, Rossi S, Lubin A, Joliot M (2010) Mapping numerical 2014.105
processing, reading, and executive functions in the developing Raizada RD, Kishiyama MM (2010) Effects of socioeconomic status
brain: an fMRI meta-analysis of 52 studies including 842 on brain development, and how cognitive neuroscience may
children. Dev Sci 13(6):876–885. doi:10.1111/j.1467-7687.2009. contribute to levelling the playing field. Front Hum Neurosci
00938.x 4(3):1000
Jednorog K, Altarelli I, Monzalvo K, Fluss J, Dubois J, Billard C, Saygin ZM, Osher DE, Norton ES, Youssoufian DA, Beach SD,
Dehaene-Lambertz G, Ramus F (2012) The influence of Feather J, Gaab N, Gabrieli JD, Kanwisher N (2016) Connec-
socioeconomic status on children’s brain structure. PLoS ONE tivity precedes function in the development of the visual word
7(8):e42486. doi:10.1371/journal.pone.0042486 form area. Nat Neurosci 19(9):1250–1255. doi:10.1038/nn.4354
Lee VE, Burkam DT (2002) Inequality at the starting gate: social Simon G, Lanoe C, Poirel N, Rossi S, Lubin A, Pineau A, Houde O
background differences in achievement as children begin school. (2013) Dynamics of the anatomical changes that occur in the
Economic Policy Institute, Washington, DC brains of schoolchildren as they learn to read. PLoS ONE
Leroy F, Glasel H, Dubois J, Hertz-Pannier L, Thirion B, Mangin JF, 8(12):e81789. doi:10.1371/journal.pone.0081789
Dehaene-Lambertz G (2011) Early maturation of the linguistic Thiebaut de Schotten M, Cohen L, Amemiya E, Braga LW, Dehaene
dorsal pathway in human infants. J Neurosci 31(4):1500–1506. S (2014) Learning to read improves the structure of the arcuate
doi:10.1523/JNEUROSCI.4141-10.2011 fasciculus. Cereb Cortex 24(4):989–995. doi:10.1093/cercor/
Li Y, Sescousse G, Amiez C, Dreher JC (2015) Local morphology bhs383
predicts functional organization of experienced value signals in Tomalski P, Johnson MH (2010) The effects of early adversity on the
the human orbitofrontal cortex. J Neurosci 35(4):1648–1658. adult and developing brain. Curr Opin Psychiatry
doi:10.1523/JNEUROSCI.3058-14.2015 23(3):233–238. doi:10.1097/YCO.0b013e3283387a8c
Mangin JF, Riviere D, Cachia A, Duchesnay E, Cointepas Y, Van Essen DC (1997) A tension-based theory of morphogenesis and
Papadopoulos-Orfanos D, Scifo P, Ochiai T, Brunelle F, Regis J compact wiring in the central nervous system. Nature
(2004) A framework to study the cortical folding patterns. 385(6614):313–318
Neuroimage 23(Suppl 1):S129–S138 Vinckier F, Dehaene S, Jobert A, Dubus JP, Sigman M, Cohen L
Mangin JF, Jouvent E, Cachia A (2010) In-vivo measurement of (2007) Hierarchical coding of letter strings in the ventral stream:
cortical morphology: means and meanings. Curr Opin Neurol dissecting the inner organization of the visual word-form system.
23(4):359–367. doi:10.1097/WCO.0b013e32833a0afc Neuron 55(1):143–156. doi:10.1016/j.neuron.2007.05.031
Martin A, Schurz M, Kronbichler M, Richlan F (2015) Reading in the Weiner KS, Barnett MA, Lorenz S, Caspers J, Stigliani A, Amunts K,
brain of children and adults: a meta-analysis of 40 functional Zilles K, Fischl B, Grill-Spector K (2017) The cytoarchitecture
magnetic resonance imaging studies. Hum Brain Mapp of domain-specific regions in human high-level visual cortex.
36(5):1963–1981. doi:10.1002/hbm.22749 Cereb Cortex 27(1):146–161. doi:10.1093/cercor/bhw361
Mongelli V, Dehaene S, Vinckier F, Peretz I, Bartolomeo P, Cohen L Whitehurst GJ, Lonigan CJ (1998) Child development and emergent
(2017) Music and words in the visual cortex: the impact of literacy. Child Dev 69(3):848–872
musical expertise. Cortex 86:260–274. doi:10.1016/j.cortex. Xue Y, Meisels SJ (2004) Early literacy instruction and learning in
2016.05.016 kindergarten: evidence from the early childhood longitudinal
Myers CA, Vandermosten M, Farris EA, Hancock R, Gimenez P, study—kindergarten class of 1998–1999. Am Educ Res J
Black JM, Casto B, Drahos M, Tumber M, Hendren RL, Hulme 41(1):191–229
C, Hoeft F (2014) White matter morphometric changes uniquely Yeatman JD, Dougherty RF, Ben-Shachar M, Wandell BA (2012)
predict children’s reading acquisition. Psychol Sci Development of white matter and reading skills. Proc Natl Acad
25(10):1870–1883. doi:10.1177/0956797614544511 Sci USA 109(44):E3045–E3053. doi:10.1073/pnas.1206792109
Noble KG, McCandliss BD (2005) Reading development and Zilles K, Palomero-Gallagher N, Amunts K (2013) Development of
impairment: behavioral, social, and neurobiological factors. cortical folding during evolution and ontogeny. Trends Neurosci
J Dev Behav Pediatr 26(5):370–378 36(5):275–284. doi:10.1016/j.tins.2013.01.006
Noble KG, McCandliss BD, Farah MJ (2007) Socioeconomic
gradients predict individual differences in neurocognitive

123
Brain Structure & Function is a copyright of Springer, 2018. All Rights Reserved.