NeuroImage 233 (2021) 117911

Contents lists available at ScienceDirect

NeuroImage
journal homepage: www.elsevier.com/locate/neuroimage

Failure of resting-state frontal–occipital connectivity in linking visual

perception with reading fluency in Chinese children with developmental
dyslexia ✩
Xiujie Yang a,1, Jia Zhang b,1, Yaping Lv b, Fang Wang c, Guosheng Ding b, Manli Zhang d,
Xiangzhi Meng e,∗, Yan Song b,∗
a
Faculty of Psychology, Beijing Normal University, Beijing 100875, China
b
State Key Laboratory of Cognitive Neuroscience and Learning and IDG/McGovern Institute for Brain Research, Beijing Normal University, Beijing 100875, China
c
Laboratory of Applied Brain and Cognitive Sciences, School of Business and Management, Shanghai International Studies University, Shanghai, China
d
Maastricht Brain Imaging Center, Department Cognitive Neuroscience, Faculty of Psychology and Neuroscience, Maastricht University, Maastricht, the Netherlands
e
School of Psychological and Cognitive Sciences, Beijing Key Laboratory of Behavioral and Mental Health, Peking University, Beijing 100871, China

a r t i c l e i n f o a b s t r a c t

Keywords: It is widely accepted that impairment in visual perception impedes children’s reading development, and further
Texture discrimination task (TDT) studies have demonstrated significant enhancement in reading fluency after visual perceptual training. However,
Reading fluency the mechanism of the neural linkage between visual perception and reading is unclear. The purpose of this
Resting-state functional connectivity
study was to examine the intrinsic functional relationship between visual perception (indexed by the texture
Primary visual cortex
discrimination task,TDT) and reading ability (character reading and reading fluency) in Chinese children with
Left middle frontal gyrus
Developmental dyslexia developmental dyslexia (DD) and those with typical development (TD). The resting-state functional connectivity
(RSFC) between the primary visual cortex (V1, BA17) and the entire brain was analyzed. In addition, how RSFC
maps are associated with TDT performance and reading ability in the DD and TD groups was examined. The results
demonstrated that the strength of the RSFC between V1 and the left middle frontal gyrus (LMFG, BA9/BA46)
was significantly correlated with both the threshold (SOA) of the TDT and reading fluency in TD children but
not in DD children. Moreover, LMFG-V1 resting-state connectivity played a mediating role in the association of
visual texture discrimination and reading fluency, but not in character reading, in TD children. In contrast, this
mediation was absent in DD children, albeit their strengths of RSFC between V1 and the left middle frontal gyrus
(LMFG) were comparable to those for the TD group. These findings indicate that typically developing children
use the linkage of the RSFC between the V1 and LMFG for visual perception skills, which in turn promote fluent
reading; in contrast, children with dyslexia, who had higher TDT thresholds than TD children, could not take
advantage of their frontal–occipital connectivity to improve reading fluency abilities. These findings suggest that
visual perception plays an important role in reading skills and that children with developmental dyslexia lack
the ability to use their frontal–occipital connectivity to link visual perception with reading fluency.

1. Introduction in developmental dyslexia (Eden et al., 1996;(Facoetti et al., 2010)

Lobier et al., 2012; (Vidyasagar and Pammer, 2010)). Reading re-
Developmental dyslexia (DD) is a failure to acquire reading skills quires visual resources to rapidly select the sublexical units through
that affects approximately 5–10% of children, despite otherwise av- the mapping of visual orthography with phonology codes (Farmer and
erage/above average intelligence and adequate education experience Klein, 1995) Ferretti et al., 2008; Franceschini et al., 2012;(Hari and
(Gori and Facoetti, 2014; McBride, 2015). Emerging studies have Renvall, 2001) Ho et al., 2000; (Plaza and Cohen, 2007)(Walker et al.,
demonstrated that visual perceptual processing deficits may result 2006). Some studies have revealed that visual perceptual interventions

✩
This work was supported by the National Natural Science Foundation of China (31971039, 81371206, 31871099, and 32000757), the National Defense Basic Sci-
entific Research Program of China (2018110B011), and the Beijing Brain Initiative of Beijing Municipal Science and Technology Commission (Z181100001518003).
We sincerely thank the children, parents, and schools for their participation and cooperation in our study.
∗
Corresponding authors.
E-mail addresses: mengxzh@pku.edu.cn (X. Meng), songyan@bnu.edu.cn (Y. Song).
1
Xiujie Yang and Jia Zhang contributed equally to this manuscript.

https://doi.org/10.1016/j.neuroimage.2021.117911
Received 1 July 2020; Received in revised form 22 February 2021; Accepted 23 February 2021
Available online 10 March 2021
1053-8119/© 2021 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/)
X. Yang, J. Zhang, Y. Lv et al. NeuroImage 233 (2021) 117911

promote reading performance in children with dyslexia (Chouake et al., that training on the visual texture discrimination task (TDT) induces
2012; Franceschini et al., 2013; Meng et al., 2014; Zorzi et al., 2012). long-lasting behavioral improvement restricted to the trained eye and
Chinese characters are logographs with a number of strokes packed trained visual field quadrant that did not transfer to other quadrants
into a square shape, which is in contrast to English words, which exhibit (Karni and Sagi, 1991). In most perceptual learning models, this speci-
different lengths (Ho and Bryant, 1999; (Tan et al., 2001a)). The com- ficity is attributed to neural plasticity in the early visual cortex, such as
plexity of a character requires children to concentrate more on the subtle the primary visual cortex (V1), where neurons are highly selective for
differences between strokes (McBride, 2016). Second, more than 39% stimulus location and orientation. Significant TDT learning effects on
of Chinese characters are compound characters with a phonetic radical V1 activity have been observed in many functional magnetic resonance
and a semantic radical (Zhou et al., 1978). Children have to make use imaging (fMRI) studies in humans (Schwartz et al., 2002; Walker et al.,
of visual perception to distinguish the phonetic radical to pronounce 2005; Yotsumoto et al., 2009). For instance, activation of the primary
it and distinguish the semantic radical to understand the meaning of visual cortex (V1) increases in the early encoding phase of TDT training
this character (McBride-Chang et al., 2005(Yang et al., 2019)). Further- but returns to baseline levels in the later retention phase (Yotsumoto
more, Chinese characters are logographs with less regularity in sound- et al., 2008). Moreover, the functional connectivity between V1 and the
print mapping (Meng et al., 2014). Children have to address phonol- lateral/medial temporal cortex is also increased after weeks of training
ogy in Chinese scripts and focus on the specific visual processes as- (Kang et al., 2018). On the other hand, (Wang et al., 2016a) demon-
sociated with orthographic analyses of Chinese characters (Yang and strated that the earliest ERP C1 component, known to reflect V1 activ-
Meng, 2020; Wang et al., 2012). Altogether, reading Chinese characters ity driven by feedforward inputs, was not modulated by visual texture
necessitates a great interactivity between phonology and orthography discrimination training. In contrast, one anterior ERP component over
and requires a heavy cognitive demand for visual perception. the prefrontal cortex was progressively modified each day. This ERP
Children with dyslexia are characterized by dysfunction of the left result suggests that changes in V1 activity induced by TDT learning pri-
middle frontal gyrus (LMFG, BA9/BA46). Several neuroimaging stud- marily originate from relatively late, higher-order cortical processing
ies have proposed that the involvement of the LMFG during visuospa- (Ding et al., 2014). Overall, these results suggest that visual texture dis-
tial analysis (Chen et al., 2002; Liu et al., 2008(Tan et al., 2001b)) and crimination should primarily involve the primary visual cortex (V1) as
through the rhyme judgment task (Ip et al., 2019; Kovelman et al., 2012; well as the functional connectivity between V1 and the higher-order
Siok et al., 2008; Tan et al., 2005; (Siok et al., 2003)) possibly con- brain cortex.
tributes to identifying reading performance across languages. Siok et al. Therefore, the present study focused on visual perception and its
(2004a, 2004b,; 2008; (Siok et al., 2009)) repeatedly found that chil- neural relationship with children’s reading performance. Specifically,
dren briefly retained “graphic” radicals in visual form while retriev- we selected V1 as the seed point and focused on how the FCs between
ing the pronunciation of characters when the LMFG was activated. This V1 and other cortical regions influence the connection between visual
gyrus might explain the functional and structural abnormalities in map- perception and reading abilities. This focus will provide important evi-
ping a character’s visual orthography to pronunciation in children with dence to improve the present understanding of the fundamental mecha-
dyslexia (Perfetti et al., 2006). nisms of dyslexia and will aid in the development of improved methods
Considering the intensive visuospatial analysis and the pronunci- to help children who are in need.
ation of various strokes demanded in Chinese reading, it is critical The texture discrimination task (TDT) has been widely used to indi-
to specifically determine how visual perception correlates to reading cate visual perception (Wang et al., 2012; Schwartz et al., 2002), and
and the neural mechanisms underlying their relationships in these chil- behavioral work has shown that reading performance can be enhanced
dren. Some behavioral studies have shown that visual perception con- by visual TDT training (Meng et al., 2014). In recent decades, studies
tributes to Chinese character reading through its role in acquiring visual- on neural mechanisms of the TDT have found that changes in the pri-
orthographic knowledge (e.g., Huang and Hanley, 1995, 1997; (Ho and mary visual cortex (V1) might underpin perceptual learning (Karni and
Bryant, 1997)1999; Siok and Fletcher, 2001). However, precisely how Sagi, 1991; 4). Meanwhile, top-down influences exerted on V1 from later
visual perception and reading processes are linked in neural mechanisms visual processing stages (V2 to V4) or from frontoparietal attention net-
remains unclear. Resting-state functional connectivity (RSFC) provides works have also been modulated with perceptual learning (Ahissar and
an approach to uncover the temporal synchronization of functionally Hochstein, 1996; Merigan, 2000; Wang et al., 2013, 2016a; Wang et al.,
related brain regions and to examine the functional relevance of intrin- 2016b; Yotsumoto et al., 2013; Zhang et al., 2015; Wang et al., 2016b).
sic brain activity patterns with behavioral performance. Previous studies We therefore made use of this task to assess children’s visual percep-
have proposed that correlating the strength of the RSFC with behavioral tion. Moreover, we examined whether the strengths of the RSFCs were
performance provides more direct evidence about the neural bases of a associated with visual perception and reading performance, as well as
wide range of cognitive functions (Wang et al., 2012). how the RSFCs influenced the relationships between TDT and Chinese
There have been a small number of fMRI studies investigating the un- reading skills.
derlying neural linkage of Chinese word processing with RSFC (Li et al., We now know that the brain areas supporting reading are influenced
2013; Wang et al., 2012; Zhou et al., 2015). Of note, many of these by culture (Bolger et al., 2005; Pugh et al., 2005). In the case of Chi-
studies have confirmed the importance of the function(s) of the LMFG nese individuals, deficits in visual-orthographic and semantic processes
and its connections with other brain regions in the resting state among might be characterized by dysfunction of the LMFG and its connection
Chinese readers (Li et al., 2013; Wang et al., 2012). However, only high- with other brain regions (Perfetti et al., 2006; Xu et al., 2019). Therefore,
level neural areas were calculated as seed regions to indicate the func- it was anticipated that TDT performance might correlate with RSFCs be-
tional connectivity in these studies. A more fundamental sensory pro- tween the V1 and LMFG, and the RSFC might be positively associated
cessing area, the primary visual cortex (V1, BA17), has not been set as with children’s reading fluency performance, which involves semantic
a seed region to explore its neural linkage with children’s reading per- processes.
formance. Importantly, abnormal lateralization of the primary visual
cortex (V1) in the dyslexic population (Jenner et al., 1999) and abnor- 2. Methods
mal visual magnocellular neurons defined in V1 (Evans et al., 1994;
Johnston et al., 2017; Stein, 2019) may account for the known visual 2.1. Participants
perception deficits measured in dyslexic individuals. For instance, asym-
metry in the size of neurons in the primary visual cortex (V1) in dyslexic For this study, fifty-two Chinese-speaking children in grades four,
individuals was shown to be associated with decreased visual perception five, and six were recruited. This study was approved by the Ethics
ability (Jenner et al., 1999). Previous behavioral evidence has illustrated Committee of the University. All parents of participating children pro-

2
X. Yang, J. Zhang, Y. Lv et al. NeuroImage 233 (2021) 117911

Table 1
Demographic characteristics (mean ± SD) and reading scores of the TD and DD groups.

TD group DD group T value P-value

N 25 18 - -
Age 10.4 ± 1.0 10.2 ± 1.0 0.31 n.s.
Sex(male/female) 12/13 12/6 – –
Nonverbal IQ 83.2 ± 13.1 80.8 ± 13.7 0.71 n.s.
Chinese character recognition (%) 89.1 ± 8.6 17.5 ± 15.4 18.96 p < 0.001
Reading fluency 49.8 ± 7.0 37.0 ± 3.7 6.88 p < 0.001

Note. TD= Typically Developing Children DD=Children with Developmental Dyslexia.

vided written informed consent according to the Declaration of Helsinki. low the norm in the character recognition test and lower than the mean
All participants were right-handed and had normal or corrected-to- scores for their grades in the reading fluency test or (2) at least 1 SD
normal vision. We excluded children with left handedness because below in the character recognition test and 0.5 SDs below in the read-
of the evidence that they might show more variability in language ing fluency test. This group criteria have been widely used in screening
dominance and, arguably, in the lateralization of other brain areas Chinese-speaking children with dyslexia (e.g., Meng et al., 2014).
(McManus, 1985a, 1985b; Pujol et al., 1999, Cai et al., 2010(Van der The texture discrimination task (TDT) has been used in numerous
Haegen et al., 2012)). visual perceptual learning studies, including in normal adults (Karni and
No participants had a history of neurological disease or other psy- Sagi, 1991. Schwartz et al., 2002. Yotsumoto et al., 2008; Wang et al.,
chiatric disorders. We assessed all participants with the cancelation test, 2016a), children with dyslexia (Meng et al., 2014; Wang et al., 2014)
which was adopted from the Wechsler Preschool and Primary Scale of and children with autism(Harris et al., 2015) (Meng et al., 2007).
Intelligence-Fourth Edition, WPPSI (Wechsler, 2012), to assess the chil- We revised the parameters based on these studies. The stimuli, white
dren’s general attention ability. Afterwards, for children who were sus- (54 cd/m2) on a uniform black background, were displayed on a 21-
pected of having attention problems, the administrators who were blind inch gamma linearized CRT monitor (1024 × 768 pixels at 85 Hz) at a
to the results of the evaluation further performed the ADHD assessment 110 cm viewing distance in the behavioral pretest.
from the Diagnostic and Statistical Manual of Mental Disorders-Fourth Edi- The visual stimulus was a texture display made of 19 × 19 high-
tion (DSM-IV) (American Psychiatric Association, 1994) (American Psy- contrast horizontal line segments positioned in the central visual field,
chiatric Association 1994) to characterize whether he/she should be di- covering an area of a 14° × 14° visual angle (Fig. 1). The lines were
agnosed with ADHD. 0.46° × 0.04° in size and spaced 0.76° apart. The position of each line
For the resting-state fMRI session, data from 9 children (TD: n = 3, segment was jittered randomly by 0° ~ 0.1° A randomly rotated letter
DD: n = 6) were excluded due to excessive head motion (> 3 mm max- ‘‘T’’ or ‘‘L’’ was presented at the center of the texture stimuli. A target
imum translation or 3° rotation). As a result, 25 TD children (mean was generated by tilting three adjacent bars in the texture stimuli from
age=10.4 ± 1.0 years old) and 18 DD children (mean age= 10.2 ± 1.0 horizontal to 135°, forming either a horizontal or a vertical configura-
years old) were included in the final resting-state fMRI analysis. Age tion (Fig. 1). The target array was embedded in the upper-left visual field
and IQ were matched between the two groups. Demographic data and at a location 2.5°~5° away from the fixation. A mask of the same size as
scores of the reading tests are reported in Table 1. the stimuli was made of 19 × 19 randomly oriented V-shaped patterns
except at the fixation, where a superimposed “T” and “L” were used to
2.2. Reading measures mask the “T” or “L” in the stimulus pattern (Fig. 1). The stimuli were
generated by a MATLAB program, and the experiment was performed
The Standardized Chinese Character Recognition Test was used to with E-prime (Fig. 2).
assess character reading. It consists of 210 characters and is divided Children’s performance was measured using an adapted procedure
into ten groups based on reading difficulty (Wang and Tao, 1993) . The (staircase) with a starting SOA value of 600 ms. The staircase fol-
participants were asked to write down a compound word based on a lowed a 2-down, 1-up rule, which resulted in a 70.7% convergence
constituent morpheme provided on the sheet. Their performance was rate (Leek, 2001). Each staircase (approximately 60 trials) consisted of
measured by the total number of correct characters (morphemes) that four preliminary reversals and six experimental reversals. The geometric
they could utilize in word compositions. The correct characters each mean of the experimental reversals was taken as the threshold for each
participant completed corresponded to a specific normal reading grade staircase. Each participant completed three staircases. The step size was
(Wang and Tao, 1993). 36 ms until the fourth reversal and then changed to 24 ms. All partici-
The Reading Fluency Test was utilized to examine children’s reading pants practiced the task (40 trials) before the formal experiment.
fluency ability. It was composed of 95 sentences (You et al., 2011). Each The procedure for TDT in the task-based fMRI scan was as follows:
sentence was paired with five multiple-choice pictures. The participants (1) the presentation time of the first fixation was 400 ms, (2) a fixed SOA
were asked to read each sentence and select, from five pictures, the one time of 564 ms was applied, and (3) the response time window was fixed
that best illustrated the meaning of the sentence. All of the sentences in at 2 s; thus, participants were told that they needed to react as soon as
this test were composed of high-frequency characters and words. The possible while guaranteeing accuracy. Beginning with a fixation block,
children were encouraged to complete as many paragraphs as possible the fMRI sequence that subjects received was comprised of three 26.4 s
within seven minutes. The total number of sentences that the partici- experimental blocks (8 trials), each of which was followed by a 24 s
pants could understand determined the performance score. fixation block.

2.3. Grouping criteria 2.4. Imaging data acquisition

As assessed by Raven’s Standard Progressive Matrices, all 52 children In the scanner, children viewed the stimuli through a mirror located
had typically developed nonverbal IQ scores (Zhang and Wang, 1985). above their eyes, and the stimuli were back-projected by a video projec-
The control group of typically developing readers all obtained above tor (1024 × 768 pixels at 60 Hz) onto a translucent screen placed inside
average scores in the Chinese character recognition test. Children were the scanner bore. The viewing distance was 60 cm, which made the TDT
placed in the dyslexia group if they scored either (1) at least 1.5 SDs be- stimuli have the same visual angle as that in the behavioral pretest.

3
X. Yang, J. Zhang, Y. Lv et al. NeuroImage 233 (2021) 117911

Fig. 1. The experimental stimuli display. The

texture stimulus (left) was composed of three
adjacent diagonal bars (vertically or horizon-
tally aligned) embedded within a background
of horizontal bars, with the letter “T” or “L” as
the fixation. The mask (right) was made of ran-
domly oriented V”, with superimposed T and L
at the fixation position.

Fig. 2. Trial sequences of texture discrimina-

tion task (TDT).

Images were acquired using a Siemens TRIO 3T scanner at the Imag- However, due to children’s emotions and circumstances, especially
ing Center for Brain Research, Beijing Normal University. The partici- with respect to dyslexia, only some of the children completed the TDT
pants laid in the scanner with their heads fixed with straps and foam task-based fMRI scanning sessions, including 24 TD children and 19 DD
pads to minimize head movement, holding an optical response box children. Among them, 15 children (TD: n = 6, DD: n = 9) were excluded
in their hands. The resting-state functional images were acquired be- due to excessive head movement (> 3 mm maximum translation or 3°
fore the anatomical imaging scan and task fMRI experiment. Data com- rotation) or behavioral accuracy values that were too low (< 25%) dur-
prised 180 continuous echo-planar imaging (EPI) functional volumes ing the scan, leaving 18 TD children and 10 DD children included for
[40 slices, repetition time (TR) = 2400 ms, echo time (TE) = 30 ms, flip analysis.
angle = 81°, field of view (FOV) = 192 mm, matrix size = 64 × 64, voxel For the first-level analysis, data from each participant were entered
size = 3.0 × 3.0 × 3.0 mm3 , and slice orientation: transversal]. During into a general linear model (GLM) to construct a multiple regression
the scan, participants were instructed to relax with their eyes closed, design matrix. The six movement parameters of the realignment (body
not to think about anything systematically and not to fall asleep. translation in the x-, y- and z-planes and rotation around the x-, y- and
For functional images of the task fMRI experiment, an EPI se- z-axes) were also added to the design as multiple regressors. Through
quence was used to measure blood oxygenation level-dependent (BOLD) model estimation, the contrast of the TDT for the experimental condition
changes with the same parameters as those used in the resting state. In versus that at baseline (fixation block) was applied to compute simple
addition, the following high-resolution T1-weighted anatomical images main effects and produce individual beta-weight images. One-sample
were also acquired using a 3D MPRAGE sequence: 176 slices, repeti- t-tests were performed on beta-weighted images for experimental con-
tion time (TR) = 2300 ms, echo time (TE) = 4.18 ms, flip angle = 9°, ditions to compare participants in the TD and DD groups. Since the key
field of view (FOV) = 256 mm, matrix size = 256 × 256, and voxel issue in the study was V1 activation, we conducted statistical analysis
size = 1.0 × 1.0 × 1.0 mm3 . using the mask of BA17 (V1 generally refers to the BA17). The thresh-
old of the activation maps was set at voxel level p < 0.001 and cluster
2.5. Task-based fMRI data preprocessing and analysis level FWE-corrected p < 0.05. Then, the overlap region of the activation
maps for each group was made, which was used as the region of interest
Task fMRI data were preprocessed and analyzed using Statistical (ROI) for further resting-state data analysis.
Parametric Mapping (SPM12, http://www.fil.ion.ucl.ac.uk/spm). The
first 2 volumes were discarded to allow for the magnetic saturation effect 2.6. Resting-state fMRI data preprocessing
and adaptation of the participants to the scanning noise. The remain-
ing images were corrected for the acquisition time delay between dif- Preprocessing of resting-state functional images was performed us-
ferent slices and corrected for head motion, followed by normalization ing the Data Processing Assistant for Resting-State fMRI (Yan and
to Montreal Neurological Institute (MNI) space using T1-image unified Zang, 2010). The first 10 volumes were discarded. In addition to the
segmentation and smoothed for decreasing spatial noise (using a 6-mm same procedure used for the task fMRI images, the remaining 170 im-
full-width at half maximum Gaussian kernel). The high-pass filter was ages were preprocessed as follows: (1) after spatial smoothing, the linear
set to 128 s. trend of the time course was removed; (2) temporal bandpass filtering

4
X. Yang, J. Zhang, Y. Lv et al. NeuroImage 233 (2021) 117911

(0.01–0.08 Hz) was performed to reduce the effects of low-frequency

drift and high-frequency noise (Fox et al., 2005); and (3) 24 head mo-
tion parameters and two repressors, including the white matter and cere-
brospinal fluid signals, were removed as nuisance covariates to control
physiological effects and head motion in children. The global signals re-
mained. To examine the potential influence of excessive head motion
in the DD group, we calculated the mean frame-wise displacement (FD)
for each remaining participant. All individualized mean FD values fell
within a reasonable range (i.e., no more than 0.2) suggested by previ-
ous studies (Yan et al., 2013) and were tested for group differences us-
ing a two-sample t-test. Furthermore, as suggested by previous research
(Guo et al., 2019; 2020), the mean FD was taken as a covariate in lin-
ear regression models in the following analysis to parse out the effect of
head motion.
Fig. 3. The behavioral result of SOA for two groups. ∗ ∗ ∗ : p < 0.001.
2.7. Resting-state functional connectivity (RSFC) analysis
approach, in which the RSFC of all voxels within a brain region was
The resting-state functional connectivity (RSFC) analysis made use
averaged and entered into the analyses. In each permutation, the divi-
of the overlap region of the activation maps based on the TDT-based
sion of participants into the two groups was randomized, and the sample
fMRI results for each group as the seed ROI. Then, the mean time series
size for each group remained unchanged. Then, Pearson’s correlation of
of this V1 seed ROI was extracted by averaging the time series of all
reading fluency and RSFC was computed and transformed to Fisher’s
voxels within it. An RSFC map for each participant was obtained after
Z-score for each of the two randomized groups. The Z-scores were sub-
correlation analysis between the mean time series of V1 and other brain
tracted between groups. This procedure was repeated 10,000 times, and
regions in a voxel-wise manner. To improve the normality of the correla-
we obtained a permutation distribution of group differences in the cor-
tion coefficients, we used z–FC maps instead of r–FC maps in subsequent
relation Z-score. The null hypothesis was that there was no group dif-
analyses by using Fisher r–to–z transformation. A one-sample t-test was
ference in the correlation between reading fluency and V1-ROI RSFC.
computed on an individual z–FC map to generate the group z–FC map
for each group separately, which was used as a mask for further analysis. 3. Results

2.8. Statistical analyses at the group level 3.1. Behavioral TDT results

To identify the cortical regions whose RSFC with the V1 seed were As illustrated in Fig. 3, the threshold of TDT (SOA) for the DD
associated with the observed group difference in TDT performance, we group was significantly higher than that of the TD group (TD vs. DD:
conducted a linear regression analysis in which the V1-associated z–FC mean = 563.15 vs. 854.62 ms; t (50) = −2.948, p < 0.006). Next, we
images across two groups were entered as a dependent variable, with the performed correlation analyses between the TDT and reading abilities.
variables of TDT performance, group, and their interaction as three re- The original threshold SOA of TDT was log-transformed to minimize
gressors of interest. This voxel-based analysis was conducted across the the influence of skewness. As illustrated in Fig. 4, the log-transformed
entire brain. Sex, age and mean FD were entered as covariates of nonin- threshold of TDT was significantly associated with the performance on
terest. We used “1″ and “−1″ to code the nominal variable of the group, the standardized Chinese character recognition test (rcharacter = −0.316,
“1″ to represent the TD group, and “−1″ to represent the DD group. p < 0.039), as well as the performance on the reading fluency test
The regressor denoting the interaction between TDT performance and (rfluency = −0.326, p < 0.033). After controlling for age and sex, the
group was generated by multiplying their values. Of note, for this encod- two correlations remained significant (rcharacter = 0.310, p < 0.049;
ing approach of variables, a negative interaction effect indicates that the rfluency = −0.319, p < 0.042). We also calculated the correlations be-
slope of the TDT performance for the TD group is less positive (or more tween TDT and reading for the TD and DD groups. It turned out
negative) than that for the DD group. After identifying brain regions that only in the TD group was TDT associated with reading fluency
exhibiting significant interaction effects of TDT performance × group, (rfluency = −0.486, p < 0.015), while other correlations were not sig-
we further conducted simple-effect analyses for each region by calculat- nificant in either group.
ing the correlation between TDT performance and V1-associated RSFC
within each group. This statistical analysis method has been widely used 3.2. Task-based fMRI results
to examine different effects of group differences on various brain regions
(e.g., de Chastelaine et al., 2017; Hsu et al., 2018; Li et al., 2017). The Using task-based functional MRI data acquired from 18 TD children
AAL template (Tzourio-Mazoyer et al., 2002) was used as an explicit and 10 children with dyslexia, we estimated the BOLD activities in V1
mask with the exception of the cerebellum, and multiple comparison using the BA17 mask. As shown in Fig. 5, voxels in V1 were maximally
correction was applied at p < 0.05 using AlphaSim. and broadly activated during the TDT task for both the TD and DD
groups (TD: 3, −84, −3, MNI coordinate; t = 14.42, cluster size = 267;
2.9. Correlation between TDT-associated RSFC and reading fluency DD: 12, −87, 0, MNI coordinate; t = 25.8, cluster size = 164, voxel level
p < 0.001, cluster level FWE-corrected p < 0.05). Then, we made the
Regions exhibiting significant interaction effects in the above lin- overlap of the activation maps of the TD and DD groups (Fig. 5C) and
ear model were considered ROIs in our following analyses. We first ex- extracted the averaged BOLD signals in this overlap region. We found
tracted the z–FC values between these ROIs and V1 (hereafter “V1-ROIs that children with DD exhibited significantly higher V1 activation than
RSFC”). Then, we calculated the correlation between the V1-ROI RSFC TD children (t (26) = 5.419, p < 0.001, two-sample t-test; see Fig. 5D).
and reading fluency for each group.
To verify the observed differences in correlation between TD and 3.3. Resting-state fMRI results
DD, we further used a nonparametric permutation test to examine the
significance of group differences in the correlation of reading fluency Analyses demonstrated that the DD group exhibited increased head
and V1-ROI RSFC. We conducted the permutation tests in a ROI-wise motion values compared to the TD group during resting-state fMRI (t

5
X. Yang, J. Zhang, Y. Lv et al. NeuroImage 233 (2021) 117911

Fig. 4. Correlation plots between SOA of TDT and reading tests. (A) Scatter plots depicting the relationship between the SOA of TDT and the percentage of Chinese
character recognition. (B) Scatter plots depicting the relationship between the SOA of TDT and reading fluency.

Fig. 5. V1 activation maps, overlap region in the activation maps and difference between the TD and DD groups. (A) V1 activation map in the TD group. (B) V1
activation map in the DD group. (C) The overlap region within the TD and DD groups. Red represents activation in the DD group. Blue represents activation in the
TD group. Pink represents the overlap region between the TD and DD groups. (D) V1 activation difference between the two groups.

(41)= 2.434, p = 0.019). However, none of the participants met the cri- observe significant positive interaction effects. Specifically, in the TD
terion of mean FD > 0.2 (Yan et al., 2013). For each group of children, group, there was a significant negative correlation between TDT per-
the resting-state time course was extracted for the overlap activated V1 formance and V1-associated RSFC in the left SFG (r (23) = −0.624, p <
region obtained during the task-based fMRI TDT task. For each partic- 0.001) and the left MFG (r (23) = −0.692, p < 0.001). In the DD group,
ipant, we obtained the RSFC map via correlation analysis between the there was a trend for a positive correlation between TDT performance
mean time series of this overlap and the whole brain in a voxel-wise and V1-associated RSFC in the left SFG (r (16) = 0.440, p = 0.068) and
manner. The RSFC maps showed that there were wide areas that had the left MFG (r (16) = 0.336, p = 0.173).
significant FCs with V1 for both the TD and DD groups (voxel level p
< 0.001, cluster level FWE-corrected p < 0.05, see Fig. 6), suggesting
3.5. RSFC correlates with reading fluency
the important role of V1, even in the resting-state brain. Next, our focus
was to identify brain regions in which the RSFC with V1 significantly
In the following analysis, we aimed to (1) examine whether the RSFC
correlated with TDT performance and reading skills and whether such
between V1 and SFG/MFG that we observed above was related to read-
associations were distinct across the DD group and the TD group. There-
ing ability and (2) to determine whether these correlations differ be-
fore, we performed the following linear regression analyses in the TD
tween the two groups. Therefore, we performed correlation analyses
and DD groups.
between RSFC and reading fluency in the TD and DD groups. Specifi-
cally, we extracted and averaged the z-FC values of the voxels within
3.4. Group differences in the relationship between V1-associated RSFC and SFG/MFG in the V1 z-FC maps and performed Pearson correlation anal-
TDT performance ysis with the SOA of TDT and reading fluency. As shown in Fig. 8, in
the TD group, we found that the RSFC between V1 and MFG predicted
We first performed linear regression analyses to explore the children’s reading fluency well (r (23) = 0.564, p = 0.003). Moreover,
brain regions showing group differences in the relationship between this effect remained significant after controlling for age, sex and mean
V1-associated RSFC and children’s TDT performance. The results FD (r (20) = 0.662, p = 0.001). We did not observe a significant corre-
demonstrated significant negative interaction effects of TDT perfor- lation between the V1-SFG RSFC and TD children’s reading fluency (r
mance × group on V1-associated RSFC in two brain regions, including (23) = 0.317, p = 0.123).
the left superior frontal gyrus (SFG) extended to the left MFG and the left For the DD group, no significant results were observed for either the
MFG (p < 0.005, AlphaSim corrected to p < 0.05, Fig. 7). We did not left MFG (r (16) = −0.101, p = 0.691) or SFG (r (16) = 0.197, p = 0.434).

6
X. Yang, J. Zhang, Y. Lv et al. NeuroImage 233 (2021) 117911

Fig. 6. Z-FC maps of the V1 in TD and DD groups. (A) Maps show voxels having significant FC with V1 in the TD group. (B) Maps show voxels having significant FC
with V1 in the DD group. The threshold was set at voxel level p < 0.001 and cluster level FWE-corrected p < 0.05.

Fig. 7. Group differences of the relationship

between the V1 Z-FC map and TDT perfor-
mance within the left SFG and the left MFG.
∗
Significance of negative correlation between
V1-SFG/MFG RSFC and TDT performance only
in the TD group.

Fig. 8. (A) Scatter plots depicting the relationship between FC strength in the V1-MFG and reading fluency in the TD and DD groups. (B) Scatter plots depicting the
relationship between FC strength in the V1-SFG and reading fluency in the TD and DD groups.

7
X. Yang, J. Zhang, Y. Lv et al. NeuroImage 233 (2021) 117911

Fig. 9. Two-wise correlations among FC of the V1-LMFG, SOA of the TDT and reading fluency in TD children. The r value (−0.542) was the correlation between the
SOA of the TDT and reading fluency after controlling for age, sex, and head motion, while the r value (−0.158) was the correlation between the SOA of the TDT and
reading fluency after controlling for age, sex, head motion, and V1-LMFG connection. ∗ ∗ ∗ p<0.001, ∗ ∗ p<0.01, ∗ p<0.05.

To address the possibility that the strength of related FC in children with the TD group. Please refer to the Supplementary Material (Figure S6)
dyslexia was weaker than that in TD children, we extracted and aver- for details.
aged the z−FC values between the aforementioned MFG region and the
voxels of V1 in the TD and DD groups. The results showed no signif-
icant difference in the strength of LMFG-V1 resting-state connectivity 4. Discussion
between the two groups (t (41) = −0.429, p = 0.67, two-sample t-test),
suggesting that the strength of related FC in children with dyslexia was The present study investigated the brain mechanism of the relation-
not weaker than that observed in TD children. ship between visual perception, particularly visual texture discrimina-
Moreover, we used a nonparametric permutation test to examine the tion, and reading in Chinese DD and TD children. The results showed
significance of group differences in the correlation of reading fluency that the participants’ performance in the TDT task was significantly as-
and the V1-MFG RSFC. We found that the TD group showed a signifi- sociated with character reading accuracy and reading fluency, and the
cantly more positive correlation of reading fluency and V1-MFG RSFC threshold of TDT (SOA) in the DD group was significantly higher in
than the DD group (p < 0.05, permutation 10,000 times). These tests the TD group, suggesting a deficit in basic visual texture discrimina-
further confirmed the findings of significant group differences observed tion in DD children. As mentioned earlier, the Chinese writing system
in the correlation analyses. is logographic in nature, and Chinese characters are visually complex.
LMFG-V1 Mediates the Relationship between TDT and Reading Fluency. Effective use of visual perception, which is thought to be the core cogni-
Finally, we conducted a mediation analysis to examine how the cor- tive processing in TDT (Censor et al., 2009; Schubo et al., 2001), allows
relation between the TDT and reading fluency was influenced by the readers to concentrate on character features and the lengths of sentences
LMFG-V1 resting-state connectivity in TD children. Age, sex, and head (Zhou et al., 2015).
motion were considered as confounding factors. The mediation analy- Moreover, TDT task-based fMRI indicated that children with DD ex-
sis revealed that V1-LMFG connectivity fully mediated the association hibited significantly stronger V1 activation than TD children. Previous
of TDT and reading fluency, indirect estimate = 0.35, p = 0.013. More- neuroimaging studies utilizing TDT have demonstrated that changes
over, after controlling for the effect of the LMFG-V1 resting-state con- in the primary visual cortex (V1) might underpin individuals’ visual
nectivity and other covariates, the correlation between TDT and reading perception (Karni and Sagi, 1991; Schwartz et al., 2002; Yotsumoto
fluency was no longer significant (r = −0.158, p = 0.494). We summa- et al., 2008). Moreover, over the course of completing TDT, the ERP
rized that there were pairwise correlations among FC of the V1-LMFG, components over the occipital cortex and the prefrontal cortex were
SOA of the TDT and reading fluency (see Fig. 9). Taken together, this re- progressively modified, suggesting that experience-dependent changes
sult indicates that the V1-LMFG connection might function as the neural may take place in the higher-level visual cortex (Wang et al., 2016a).
substrate linking TDT and reading fluency in TD children. Therefore, stronger V1 activation in children with DD might, on the one
As the groups were quite imbalanced, we also performed the above- hand, indicate that those children probably require more V1 neuronal
mentioned analysis by randomly selecting 10 children from the TD resources when discriminating the visual texture (Pourtois et al., 2008),
group and reconducting the analyses together with 10 DD children (Sup- given that hyperactivation might help them compensate for their ineffi-
plementary Material, Figure S1-S5). Specifically, we first performed ac- ciency in completing this task. In contrast, it is also possible that children
tivation analysis during the TDT task in these two groups. Then, we with DD lack regulated resources from high-level cognitive processing,
identified the overlapping active region in V1 in the two groups. In the such as visual-spatial attention (Franceschini et al., 2012), and there-
following RSFC analysis, this overlapping region of V1 was regarded as fore might need to activate V1 as much as possible to compensate for
the seed region. The results showed that the V1-MFG RSFC pattern was the lack of regulations from higher-level cognitive processing.
only correlated with reading fluency in the TD group, which was quite The RSFC between V1 and LMFG was found to mediate the relation-
similar to the main analysis in which 18 TD and 10 DD children were ship between TDT and reading fluency in TD children. When children
included. In addition, we performed the analysis again by regressing read printed texts, their visual systems confront several challenges, in-
out the global signal. The V1-LMFG connection effect was reproducible cluding the competing visual constraint of congregating different words
in the TD group, which was consistent with the main analysis. Finally, into a sentence in a certain sequence (Johnston et al., 2017). V1 is fun-
to make it more convincing, we additionally selected the second visual damental for high-level cognitive processing, such as word recognition
cortex (V2) as the seed, created its resting-state functional connectivity (Jenner et al., 1999). There were findings showing abnormal neuronal
with the whole brain regions, and performed a similar linear regression activations in the primary visual cortex (V1) in dyslexic individuals who
analysis. In contrast with previous results with V1 as the seed, we did were associated with their decreased visual perceptual ability (Rosen &
not find such frontal–occipital connectivity linked to reading fluency in Galaburda, 1999). Of note, abnormal visual magnocellular neurons de-

8
X. Yang, J. Zhang, Y. Lv et al. NeuroImage 233 (2021) 117911

fined in V1 successfully explained the visual perception deficits in Chi- Collectively, compared to recent fMRI studies that reported Chinese
nese dyslexic individuals (Qian et al., 2015). dyslexic dysfunction in isolated dorsal regions, such as the IPS/SPL (Siok
In contrast to alphabetic reading systems, Chinese reading involves et al., 2009), the V5/MT (Qian and Bi, 2015; Qian et al., 2015) and the
characters occupying two-dimensional space both in height and width, LMFG (Siok et al. 2004a, 2004b), our results extend these findings by
and their structures are much more visually complex (McBride, 2015; establishing altered synchronization in the linkage of the primary vi-
Wang et al., 2012). Some previous studies have established that the sual cortex and the dorsal area (i.e., LMFG) in Chinese children with
LMFG is significantly activated when completing visual perception tasks DD. However, this study has some limitations. The sample size for task-
( (Belger et al., 1998) ; (Carlson et al., 1998) Zhou et al., 2015). Strong based fMRI data was relatively small due to excessive head motion in
activation of the LMFG in reading might be associated with visual prints some children. Task-based fMRI between TD children and children with
across languages, such as the unique square configuration of Chinese lo- DD still needs further investigation. Moreover, because visual percep-
gographs. In line with this, neuroimaging studies have established that tion involves spatial attention (Wang et al., 2013; 2016b) and TDT might
LMFG is activated in visuo-orthographic processing of Chinese reading correlate with reading fluency performance via children’s spatial atten-
(Chen et al., 2002; Kuo et al., 2004; Siok et al. 2004a, 2004b), which was tion ability, further study is still required to provide direct evidence for
also observed in English reading (Kovelman et al., 2012; Ip et al., 2019). why the RSFCs among children with DD fail to mediate the associations
Comparing French and Chinese reading, another study has revealed that between TDT and reading fluency. Last, apart from V1, connections of
the LMFG is not specific to Chinese writing but includes representations the LMFG to other brain areas has been found to be important for fluent
of handwriting gestures that are engaged in both alphabetical and non- Chinese reading (Zhou et al., 2015). More RSFCs between the LMFG and
alphabetical languages(Feng et al., 2020) . Taken together, the func- other brain areas with the LMFG as a seed region might be required to
tional connectivity between V1 and the LMFG is reasonable supposed determine their linkages with Chinese reading.
to be the supportive neural basis underlying the connection of visual In conclusion, the current study mapped out the V1-LMFG resting-
perception and fluent reading, particularly Chinese reading. state connectivity that correlated with visual texture discrimination and
In contrast, in children with DD, even though the strength of the FC Chinese reading fluency and demonstrated the mediating role of the V1-
between the V1 and LMFG was comparable to that of TD children, the LMFG resting-state connectivity in the associations between TDT and
LMFG-V1 resting-state FC did not correlate with either visual percep- reading fluency among the TD group. In contrast, there was no such me-
tion or reading fluency. These findings indicate that Chinese children diation effect in DD children. These findings revealed that the functional
with DD cannot take advantage of the FC between the V1 and LMFG connectivity of the V1 and LMFG might be the common underlying neu-
to link visual perception and reading fluency. This might be because ral mechanisms supporting the relationship between complicated visual
children with DD may lack visual spatial attentional resources to co- perception and fluent reading, whereas in DD children, frontal–occipital
ordinate visual perception and reading processes. Indeed, the V1 sys- connectivity might be dissociated from a precise spatial sampling of vi-
tem is fundamental in processing basic visual information, while the sual input and was not involved in their development of reading skills.
LMFG was found to be important in supporting visual spatial atten-
tion during Chinese reading (Siok et al. 2004a, 2004b). Furthermore, Author statement
Perfetti et al. (2006) proposed that the LMFG is recruited to coordinate
visual-orthographic and semantic information during Chinese reading. Xiujie Yang: conceptualization, methodology, formal analysis, writ-
Taken together, these findings suggest that in children with DD, frontal– ing – original draft, writing – review & editing, visualization.
occipital connectivity might be dissociated from a precise spatial sam- Jia Zhang: methodology, formal analysis, validation, writing – re-
pling of visual input, and they performed a substandard coordination of view & editing, visualization.
visual perception and reading fluency. Yaping Lv: methodology, validation, formal analysis.
The role of frontal–occipital connectivity in character reading was Fang Wang: methodology, validation.
not established, possibly because children required fewer visual per- Guosheng Ding: software, methodology, validation, writing- re-
ceptual resources during character reading than sentence reading. In- viewing.
deed, in the reading fluency task, sentences consisting of multiple char- Manli Zhang: writing- reviewing and editing, validation.
acters in a spatial sequence were presented. Due to the constraint of Xiangzhi Meng: conceptualization, writing – original draft, writing-
humans’ attention resources, children need efficient attention shift- reviewing, validation, supervision.
ing/allocating across successive fixations and to disengage in irrelevant Yan Song: conceptualization, writing- reviewing, validation, super-
noise (Moores et al., 2011) . In the character recognition task, the target vision.
characters were presented one by one and without time pressure. Corre-
spondingly, Liu et al. (2015) found that visual perception still predicted Data and code availability statements
reading comprehension after controlling for Chinese character reading,
suggesting that visual perception is more essential for reading multiple All of the data used in the current study were collected and shared
sentences than for reading single characters. by Song’s lab, Meng’s lab (corresponding authors’ labs), and Ding’s lab.
Apart from visual-orthographic processing in Chinese, the different This experiment includes fifty-two Chinese-speaking children. Written
correlations of resting-state functional connectivity with reading fluency consent was obtained from each participant and his/her guardians be-
between typically developing children and children with dyslexia might fore the experiment. All of the participants were selected and recruited
also be related to phonological processing (Ip et al., 2019), which might from several primary schools in Beijing, China.
not be limited in Chinese. For example, Kovelman et al. (2012) em- We acknowledged that sharing data is a very good idea and practice.
ployed fMRI to identify the neural correlates of phonological awareness However, we had promised our participants that we would keep all of
using an auditory word-rhyming task in English-speaking children who their information as well as the data a secret and not share with others
were typical readers or who had dyslexia and found that typically de- unless we get their permission. Given the project is still ongoing, cur-
veloping children, rather than children with dyslexia, recruited the left rently we prefer to keep the participants’ information and the data that
dorsolateral prefrontal cortex (DLPFC) when making explicit phonolog- we collected in privacy before the project is completely done. Neverthe-
ical judgments. Notably, the DLPFC in the Kovelman et al. (2012) study less, the data that support the findings of this study are available from
nicely overlaps with the LMFG (BA9) in the current study. Taken to- the corresponding author upon reasonable request with signing a data
gether, the association of LMFG-V1 resting-state FC and reading fluency sharing agreement.
might not only require the processes of visual orthography but may also In the data-sharing agreement, the receiver should state as specifi-
reflect the rapid retrieval and maintenance of phonological information. cally as possible how the receiver will use the data. For the period of

9
X. Yang, J. Zhang, Y. Lv et al. NeuroImage 233 (2021) 117911

agreement, we clearly define that the provider will give the data to the Fox, M.D., Snyder, A.Z., Vincent, J.L., Corbetta, M., Vanessen, D.C., Raichle, M.E., 2005.
receiver within three working days upon request, and the receiver will The human brain is intrinsically organized into dynamic, anticorrelated functional
networks. Proc. Natl. Acad. Sci. 102, 9673–9678. doi:10.1073/pnas.0504136102.
be able to use the data within ten working days. To ensure that data re- Franceschini, S., Gori, S., Ruffino, M., Viola, S., Molteni, M., Facoetti, A., 2013. Ac-
main confidential, the data should be returned to the provider or should tion video games make dyslexic children read better. Curr. Biol. 23 (6), 462–466.
be deleted from their hard drives after usage. The receiver cannot share, doi:10.1016/j.cub.2013.01.044.
Franceschini, S., Gori, S., Ruffino, M., Pedrolli, K., Facoetti, A., 2012. A causal link be-
sell or distribute data findings or any part of the database to another tween visual spatial attention and reading acquisition. Curr. Biol. 22 (9), 814–819.
agency. Because some data contain information that can be linked to doi:10.1016/j.cub.2012.03.013.
individuals, personal data should remain confidential and should not be Gori, S., Facoetti, A., 2014. Perceptual learning as a possible new approach for
remediation and prevention of developmental dyslexia. Vision Res. 99, 78–87.
disclosed verbally or in writing to an unauthorised third party, by acci-
doi:10.1016/j.visres.2013.11.011.
dent or otherwise. The receiver should not report information that iden- Guo, X., Duan, X., Suckling, J., Chen, H., Liao, W., Cui, Q., Chen, H., 2019. Partially
tifies individuals. Data will be transferred from the provider to the re- impaired functional connectivity states between right anterior insula and default
mode network in autism spectrum disorder. Hum. Brain Mapp. 40 (4), 1264–1275.
ceiver electronically. For data security, electronic copies of data should
doi:10.1002/hbm.24447.
be password protected or kept on a secure disk. Everyone at the receiver Guo, X., Duan, X., Chen, H., He, C., Xiao, J., Han, S., ..., Chen, H., 2020. Altered inter-and
agency have the same level of access to data. If the receiver generates intrahemispheric functional connectivity dynamics in autistic children. Hum. Brain
a report based on the data, the report belongs to the receiver and the Mapp. 41 (2), 419–428. doi:10.1002/hbm.24812.
Ho, C.S.H., Law, T.P.S., Ng, P.M, 2000. The phonological deficit hypothesis in Chinese
provider. The monetary costs of sharing the data should be paid by the developmental dyslexia. Read Writ. 13 (1–2), 57–79. doi:10.1023/A:1008040922662.
receiver if needed. Hari, R., Renvall, H., 2001. Impaired processing of rapid stimulus sequences in dyslexia.
Trends in cognitive sciences 5 (12), 525–532. doi:10.1016/s1364-6613(00)01801-5.
Harris, H., Israeli, D., Minshew, N., Bonneh, Y., Heeger, D. J., Behrmann, M., Sagi, D.,
Supplementary materials 2015. Perceptual learning in autism: over-specificity and possible remedies. Nature
neuroscience 18 (11), 1574–1576.doi. doi:10.1038/nn.4129.
Supplementary material associated with this article can be found, in Ho, C. S. H., Bryant, P., 1997. Phonological skills are important in learning to read Chinese.
Developmental psychology 33 (6), 946–951. doi:10.1037/0012-1649.33.6.946.
the online version, at doi:10.1016/j.neuroimage.2021.117911. Ho, C.S., Bryant, P., 1999. Different visual skills are important in learning to read English
and Chinese. Educ. Child Psychol. 16 (4), 4–14.
References Hsu, C.T., Sims, T., Chakrabarti, B., 2018. How mimicry influences the neu-
ral correlates of reward: an fMRI study. Neuropsychologia 116, 61–67.
Ahissar, M., Hochstein, S., 1996. Learning pop-out detection: Specificities to stimulus char- doi:10.1016/j.neuropsychologia.2017.08.018.
acteristics. Vision research 36 (21), 3487–3500. doi:10.1016/0042-6989(96)00036-3. Huang, H.S., Hanley, J.R., 1995. Phonological awareness and visual skills
American Psychiatric Association, 1994. Diagnostic and Statistical Manual of Mental Dis- in learning to read Chinese and English. Cognition 54 (1), 73–98.
orders, 4th edition American Psychiatric Association, Washington DC. doi:10.1016/0010-0277(94)00641-W.
Belger, A., Puce, A., Krystal, J. H., Gore, J. C., Goldman‐Rakic, P., & Mc- Huang, H.S., Hanley, J.R., 1997. A longitudinal study of phonological awareness, visual
Carthy, G. (1998). Dissociation of mnemonic and perceptual processes dur- skills, and Chinese reading acquisition among first-graders in Taiwan. Int. J. Behav.
ing spatial and nonspatial working memory using fMRI. Human brain map- Dev. 20, 249–268. doi:10.1080/016502597385324.
ping, 6(1), 14-32. doi:10.1002/(SICI)1097-0193(1998)6:1&#x0003c;14::AID- Ip, K.I., Marks, R.A., Hsu, L.S.J., Desai, N., Kuan, J.L., Tardif, T, 2019. Morphologi-
HBM2&#x0003e;3.0.CO;2-O cal processing in Chinese engages left temporal regions. Brain Lang. 199, 104696.
Bolger, D.J., Perfetti, C.A., Schneider, W., 2005. Cross-cultural effect on the brain revisited: doi:10.1016/j.bandl.2019.104696.
universal structures plus writing system variation. Hum. Brain Mapp. 25 (1), 92–104. Jenner, A.R., Rosen, G.D., Galaburda, A.M., 1999. Neuronal asymmetries in primary
doi:10.1002/hbm. visual cortex of dyslexic and nondyslexic brains. Ann. Neurol. 46 (2), 189–196
Cai, Q., Paulignan, Y., Brysbaert, M., Ibarrola, D., Nazir, T.A., 2010. The left ventral https://doi.org/10.1002/1531-8249(199908)46:2<189::aid−ana8>3.0.co;2-n.
occipito-temporal response to words depends on language lateralization but not on Johnston, R., Pitchford, N.J., Roach, N.W., Ledgeway, T., 2017. Encoding of rapid time-
visual familiarity. Cerebr. Cortex 20, 1153–1163. doi:10.1093/cercor/bhp175. varying information is impaired in poor readers. J. Vis. 17 (5). doi:10.1167/17.5.1,
Carlson, S., Martinkauppi, S., Rämä, P., Salli, E., Korvenoja, A., Aronen, H.J., 1998. Distri- 1–1.
bution of cortical activation during visuospatial n-back tasks as revealed by functional ... Kang, D.W., Kim, D., Chang, L.H., Kim, Y.H., Takahashi, E., Cain, M.S., Watanbe, T.,
magnetic resonance imaging. Cerebral cortex 8 (8), 743–752.doi. doi:10.1093/cer- Sasaki, Y., 2018. Structural and functional connectivity changes beyond visual
cor/8.8.743. cortex in a later phase of visual perceptual learning. Sci. Rep. 8 (1), 1–9.
Censor, N., Bonneh, Y., Arieli, A., Sagi, D., 2009. Early-vision brain responses doi:10.1038/s41598-018-23487-z.
which predict human visual segmentation and learning. J. Vis. 9 (4), 12 Karni, A., Sagi, D., 1991. Where practice makes perfect in texture discrimination: ev-
https://doi.org/10.1167/9.4.12. idence for primary visual cortex plasticity. Proc. Natl. Acad. Sci. 88, 4966–4970.
Chen, Y., Fu, S., Iversen, S.D., Smith, S.M., Matthews, P.M., 2002. Testing for doi:10.1073/pnas.88.11.4966.
dual brain processing routes in reading: a direct contrast of Chinese charac- ... Kovelman, I., Norton, E. S., Christodoulou, J. A., Gaab, N., Lieberman, D. A., Triantafyl-
ter and pinyin reading using fMRI. J. Cogn. Neurosci. 14 (7), 1088–1098. lou, C., Wolf, M, Whitfield-Gabrieli, S, Gabrieli, J. D, 2012. Brain basis of phonological
doi:10.1162/089892902320474535. awareness for spoken language in children and its disruption in dyslexia. Cerebr. Cor-
Chouake, T., Levy, T., Javitt, D.C., Lavidor, M., 2012. Magnocellular training improves vi- tex 22 (4), 754–764. doi:10.1093/cercor/bhr094.
sual wordrecognition. Front. Hum. Neurosci. 6 (14). doi:10.3389/fnhum.2012.00014. Kuo, W.J., Yeh, T.C., Lee, J.R., Chen, L.F., Lee, P.L., Chen, S.S., Ho, L.T.,
de Chastelaine, M., Mattson, J.T., Wang, T.H., Donley, B.E., Rugg, M.D., 2017. In- Hung, D.L., Tzeng, O.J.L., Hsieh, J.C, 2004. Orthographic and phonological pro-
dependent contributions of fMRI familiarity and novelty effects to recognition cessing of Chinese characters: an fMRI study. Neuroimage 21 (4), 1721–1731.
memory and their stability across the adult lifespan. Neuroimage 156, 340–351. doi:10.1016/j.neuroimage.2003.12.007.
doi:10.1016/j.neuroimage.2017.05.039. Leek, M.R., 2001. Adaptive procedures in psychophysical research. Attent., Percept. Psy-
Ding, Y., Martinez, A., Qu, Z., Hillyard, S.A., 2014. Earliest stages of visual cortical pro- chophys. 63 (8), 1279–1292. doi:10.3758/BF03194543.
cessing are not modified by attentional load. Hum. Brain Mapp. 35 (7), 3008–3024. Li, L., Liu, J., Chen, F., Feng, L., Li, H., Tian, J., et al., 2013. Resting-state neural networks
doi:10.1002/hbm.22381. for visual Chinese word processing in Chinese adults and children. Neuropsychologia
Eden, G.F., VanMeter, J.W., Rumsey, J.M., Maisog, J.M., Woods, R.P., Zeffiro, T.A., 1996. 51, 1571–1583. doi:10.1016/j.neuropsychologia.2013.05.011.
Abnormal processing of visual motion in dyslexia revealed by functional brain imag- ... Li, L., Abutalebi, J., Emmorey, K., Gong, G., Yan, X., Feng, X., Zou, L, Ding, G., 2017.
ing. Nature 382 (6586), 66–69. doi:10.1038/382066a0. How bilingualism protects the brain from aging: insights from bimodal bilinguals.
Evans, B.J., Drasdo, N., Richards, I.L., 1994. An investigation of some sen- Hum. Brain Mapp. 38 (8), 4109–4124. doi:10.1002/hbm.23652.
sory and refractive visual factors in dyslexia. Vis. Res. 34 (14), 1913–1926. Liu, C., Zhang, W.T., Tang, Y.Y., Mai, X.Q., Chen, H.C., Tardif, T., Luo, Y.J., 2008. The
doi:10.1016/0042-6989(94)90315-8. visual word form area: evidence from an fMRI study of implicit processing of Chinese
Facoetti, A., Corradi, N., Ruffino, M., Gori, S., Zorzi, M., 2010. Visual spatial attention and characters. Neuroimage 40 (3), 1350–1361. doi:10.1016/j.neuroimage.2007.10.014.
speech segmentation are both impaired in preschoolers at familial risk for develop- Liu, D., Chen, X., Chung, K.K., 2015. Performance in a visual search task uniquely predicts
mental dyslexia. Dyslexia 16 (3), 226–239. https://doi.org/10.1002/dys.413. reading abilities in third-grade Hong Kong Chinese children. Sci. Stud. Read. 19 (4),
Farmer, M.E., Klein, R.M., 1995. The evidence for a temporal process- 307–324. doi:10.1080/10888438.2015.1030749.
ing deficit linked to dyslexia: A review. Psychon Bull Rev 2, 460–493. McManus, I.C., 1985a. Handedness, language dominance and aphasia: a genetic model.
https://doi.org/10.3758/BF03210983. Psychol. Med. – Monogr. Suppl., 8, 1–40.
Feng, X., Altarelli, I., Monzalvo, K., Ding, G., Ramus, F., Shu, H., Dehaene, S., Meng, X., Lobier, M., Zoubrinetzky, R., Valdois, S., 2012. The visual attention span deficit in dyslexia
Dehaene-Lambertz, G., 2020. A universal reading network and its modulation by is visual and not verbal. Cortex 48 (6), 768–773. doi:10.1016/j.cortex.2011.09.003.
writing system and reading ability in French and Chinese children. Elife 9, e54591. McBride, C. (2015). Children’s literacy development: A cross-cultural perspective on learning
doi:10.7554/eLife.54591. to read and write. Routledge.
Ferretti, G., Mazzotti, S., Brizzolara, D. A. N. I. E. L. A., 2008. Visual scanning and read- McBride-Chang, C., Chow, B.W.-Y., Zhong, Y., Burgess, S., Hayward, W.G, 2005. Chinese
ing ability in normal and dyslexic children. Behavioural Neurology 19 (1–2), 87–92. character acquisition and visual skills in two Chinese scripts. Read Writ. 18, 99–128.
doi:10.1155/2008/564561. doi:10.1007/s11145-004-7343-5.

10
X. Yang, J. Zhang, Y. Lv et al. NeuroImage 233 (2021) 117911

McManus, I.C., 1985b. Handedness, language dominance and apha- Tzourio-Mazoyer, N., Landeau, B., Papathanassiou, D., Crivello, F., Etard, O., Delcroix, N.,
sia: a genetic model. Psychol. Med. – Monogr. Suppl. 8, 1–40 Mazoyer, B., Joliot, M., 2002. Automated anatomical labeling of activations in SPM
https://doi.org/10.1017/s0264180100001879. using a macroscopic anatomical parcellation of the MNI MRI single-subject brain.
McBride, C.A., 2016. Is Chinese special? Four aspects of Chinese literacy acquisition that Neuroimage 15 (1), 273–289. doi:10.1006/nimg.2001.0978.
might distinguish learning Chinese from learning alphabetic orthographies. Educ. Psy- Van der Haegen, L., Cai, Q., Brysbaert, M., 2012. Colateralization of Broca’s area and the
chol. Rev. 28 (3), 523–549. doi:10.1007/s10648-015-9318-2. visual word form area in left-handers: fMRI evidence. Brain Lang. 122 (3), 171–178.
Meng, X., Lin, O., Wang, F., Jiang, Y., Song, Y., 2014. Reading performance is enhanced doi:10.1016/j.bandl.2011.11.004.
by visual texture discrimination training in Chinese-speaking children with develop- Vidyasagar, T. R., Pammer, K., 2010. Dyslexia: a deficit in visuo-spatial attention,
mental dyslexia. PLoS One 9 (9), 108–274. doi:10.1371/journal.pone.0108274. not in phonological processing. Trends in Cognitive Sciences 14 (2), 57–63.
Meng, X., Tian, X., Jian, J., Zhou, X., 2007. Orthographic and phonological processing in https://doi.org/10.1016/j.tics.2009.12.003.
Chinese dyslexic children: an ERP study on sentence reading. Brain Res. 1179, 119– Walker, K., Hall, S., Klein, R., Phillips, D., 2006. Development of percep-
130. doi:10.1016/j.brainres.2007.08.046. tual correlates of readingperformance. Brain Res. 1124 (1), 126–141.
Merigan, W. H., 2000. Cortical area V4 is critical for certain texture discriminations, doi:10.1016/j.brainres.2006.09.080.
but this effect is not dependent on attention. Visual neuroscience 17 (6), 949–958. Walker, M.P., Stickgold, R., Jolesz, F.A., Yoo, S.S., 2005. The functional anatomy of sleep-
doi:10.1017/S095252380017614X. dependentvisual skill learning. Cerebr. Cortex 15 (11), 1666–1675. doi:10.1093/cer-
Moores, E., Cassim, R., Talcott, J. B., 2011. Adults with dyslexia exhibit large cor/bhi043.
effects of crowding, increased dependence on cues, and detrimental ef- Wang, F., Huang, J., Lv, Y., Ma, X., Yang, B., Wang, E., Du, B., Li, W., Song, Y., 2016a. Pre-
fects of distractors in visual search tasks. Neuropsychologia 49, 3881–3890. dicting perceptual learning from higher-order cortical processing. Neuroimage 124,
doi:10.1016/j.neuropsychologia.2011.10.005. 682–692. doi:10.1016/j.neuroimage.2015.09.024.
Perfetti, C.A., Tan, L.H., Siok, W.T., 2006. Brain-behavior relations in reading Wang, R., Wang, J., Zhang, J. Y., Xie, X. Y., Yang, Y. X., Luo, S. H., Yu, C., Li, W., 2016b.
and dyslexia: implications of Chinese results. Brain Lang. 98 (3), 344–346. Perceptual learning at a conceptual level. Journal of Neuroscience 36 (7), 2238–2246.
doi:10.1016/j.bandl.2006.04.010. doi:10.1523/JNEUROSCI.2732-15.2016.
Plaza, M., Cohen, H., 2007. The contribution of phonological awareness and visual atten- Wang, F., Zhong, X., Huang, J., Ding, Y., Song, Y., 2013. Comparison of perceptual learning
tion in early reading and spelling. Dyslexia 13 (1), 67–76. doi:10.1002/dys.330. of real and virtual line orientations: an event-related potential study. Vision Res. 93,
Pourtois, G., Rauss, K.S., Vuilleumier, P., Schwartz, S., 2008. Effects of perceptual 1–9. doi:10.1016/j.visres.2013.10.004.
learning on primary visual cortex activity in humans. Vision Res. 48 (1), 55–62. Wang, X., Han, Z., He, Y., Liu, L., Bi, Y., 2012. Resting-state functional connectiv-
doi:10.1016/j.visres.2007.10.027. ity patterns predict Chinese word reading competency. PLoS ONE 7 (9), e44848.
Pugh, K.R., Sandak, R., Frost, S.J., Moore, D., Mencl, W.E, 2005. Examining reading doi:10.1371/journal.pone.0044848.
development and reading disability in English language learners: potential con- Wang, X.L., Tao, B.P., 1993. Chinese Character Recognition Test Battery and Assessment
tributions from functional neuroimaging. Learn. Disab. Res. Pract. 20 (1), 24–30 Scale for Primary School Children. Shanghai Education Press, Shanghai.
https://doi.org/10.1111/j.1540-5826.2005.00117.x. Wang, Z., Cheng-Lai, A., Song, Y., Cutting, L., Jiang, Y., Lin, O., Meng, X., Zhou, X., 2014.
Pujol, J., Deus, J., Losilla, J.M., Capdevila, A., 1999. Cerebral lateralization of lan- A perceptual learning deficit in Chinese developmental dyslexia as revealed by visual
guage in normal left-handed people studied by functional MRI. Neurology 52 (5). texture discrimination training. Dyslexia 20 (3), 280–296. doi:10.1002/dys.1475.
doi:10.1212/wnl.52.5.1038, 1038–1038. Wechsler, D., 2012. Wechsler Preschool and Primary Scale of Intelligence-Fourth Edition.
Qian, Y., Bi, H.Y., 2015. The effect of magnocellular-based visual-motor interven- NCS Pearson, San Antonio, TX.
tion on Chinese children with developmental dyslexia. Front. Psychol. 6, 1529. Xu, M., Tan, L.H., Perfetti, C., 2019. Developmental dyslexia in Chinese. In: Developmental
doi:10.3389/fpsyg.2015.01529. Dyslexia across Languages and Writing Systems, 10, pp. 200–225.
Qian, Y., Deng, Y., Zhao, J., Bi, H.Y., 2015. Magnocellular-dorsal path- Yan, C.G., Cheung, B., Kelly, C., Colcombe, S., Craddock, R.C., Di Martino, A., Li, Q,
way function is associated with orthographic but not phonological skill: Zuo, X.N., Castellanos, F.X., Milham, M.P., 2013. A comprehensive assessment of re-
fMRI evidence from skilled Chinese readers. Neuropsychologia 71, 84–90. gional variation in the impact of head micromovements on functional connectomics.
doi:10.1016/j.neuropsychologia.2015.03.024. Neuroimage 76, 183–201. doi:10.1016/j.neuroimage.2013.03.004.
Schubö, A., Schlaghecken, F., Meinecke, C., 2001. Learning to ignore the mask in tex- Yan, C.G., Zang, Y.F., 2010. DPARSF: a MATLAB toolbox for “pipeline” data analysis of
ture segmentation tasks. J. Exp. Psychol.: Hum. Percept. Perform. 27 (4), 919. resting-state fMRI. Front. Syst. Neurosci. 4, 13. doi:10.3389/fnsys.2010.00013.
doi:10.1037/0096-1523.27.4.919. Yang, X., Meng, X., 2020. Visual processing matters in chinese reading acquisition and
Schwartz, S., Maquet, P., Frith, C., 2002. Neural correlates of perceptual learning: a func- early mathematics. Front. Psychol. 11 (462), 1–12. doi:10.3389/fpsyg.2020.00462.
tional MRI study of visual texture discrimination. Proc. Natl. Acad. Sci. 99 (26), Yang, X., Peng, P., Meng, X., 2019. Contributions of basic cognitive processing to Chinese
17137–17142. doi:10.1073/pnas.242414599. reading: the mediation effect of basic language processing. Front. Psychol. 9 (2670),
Siok, W.T., Perfetti, C.A., Jin, Z., Tan, L.H., 2004a. Biological abnormality of impaired 1–11. doi:10.3389/fpsyg.2018.02670.
reading is constrained by culture. Nature 43, 71–76. doi:10.1038/nature02865. Yotsumoto, Y., Chang, L.H., Ni, R., Pierce, R., Andersen, G.J., Watanabe, T., Sasaki, Y.,
Siok, W.T., Perfetti, C.A., Jin, Z., Tan, L.H., 2004b. Biological abnormality of impaired 2013. White matter in the older brain is more plastic than in the younger brain. Nat.
reading isconstrained by culture. Nature 43 (7004), 71–76. doi:10.1038/nature02865. Commun. 19 (5), 5504. doi:10.1038/ncomms6504.
Siok, W.T., Jin, Z., Fletcher, P., Tan, L.H., 2003. Distinct brain regions associated with Yotsumoto, Y., Sasaki, Y., Chan, P., Vasios, C.E., Bonmasssar, G., Ito, N., Nanez, J.E.,
syllable and phoneme. Hum. Brain Mapp. 18 (3), 201–207. doi:10.1002/hbm.10094. Shimojo, S., Watanabe, T., 2009. Location-specific cortical activation changes dur-
Siok, W.T., Niu, Z., Jin, Z., Perfetti, C.A., Tan, L.H., 2008. A structural–functional basis for ing sleep after training forperceptual learning. Curr. Biol. 19 (15), 1278–1282.
dyslexia in the cortex of Chinese readers. Proc. Natl. Acad. Sci. 105 (14), 5561–5566. doi:10.1016/j.cub.2009.06.011.
doi:10.1073/pnas.0801750105. Yotsumoto, Y., Watanabe, T., Sasaki, Y., 2008. Different dynamics of performance and
Siok, W.T., Fletcher, P., 2001. The role of phonological awareness and visual– brain activation in the time course of perceptual learning. Neuron 57 (6), 827–833.
orthographic skills in chinese reading acquisition. Dev. Psychol. 37, 886–899. doi:10.1016/j.neuron.2008.02.034.
doi:10.1037/0012-1649.37.6.886. You, H., Gaab, N., Wei, N., Cheng-Lai, A., Wang, Z., Jian, J., Song, M., Meng, X., Ding, G.,
Siok, W. T., Spinks, J. A., Jin, Z., Tan, L. H., 2009. Developmental dyslexia is characterized 2011. Neural deficits in second language reading: fMRI evidence from Chinese-
by the co-existence of visuospatial and phonological disorders in Chinese children. speaking children with English reading impairment. Neuroimage 57 (3), 760–770.
Current biology 19 (19), R890–R892. doi:10.1016/j.cub.2009.08.014. doi:10.1016/j.neuroimage.2010.12.003.
Stein, J., 2019. The current status of the magnocellular theory of developmental dyslexia. Zhang, G.L., Li, H., Song, Y., Yu, C., 2015. ERP C1 is top-down modulated by orientation
Neuropsychologia 130, 66–77.doi. doi:10.1016/j.neuropsychologia.2018.03.022. perceptual learning. J. Vis. 15 (10), 8. doi:10.1167/15.10.8.
Tan, L.H., Feng, C.M., Fox, P.T., Gao, J.H., 2001a. An fMRI study with written Chinese. Zhang, H.C., Wang, X.P., 1985. Raven Standard Progressive Matrices: Chinese City Revi-
Neuroreport 12 (1), 83–88. doi:10.1097/00001756-200101220-00024. sion. The National Revision Collaborative Group, Beijing.
Tan, L.H., Laird, A.R., Li, K., Fox, P.T., 2005. Neuroanatomical correlates of phonological Zhou, W., Xia, Z., Bi, Y., Shu, H., 2015. Altered connectivity of the dorsal and ventral
processing of Chinese characters and alphabetic words: a meta-analysis. Hum. Brain visual regions in dyslexic children: a resting-state fMRI study. Front. Hum. Neurosci.
Mapp. 25 (1), 83–91. doi:10.1002/hbm.20134. 9, 495. doi:10.3389/fnhum.2015.00495.
Tan, L.H., Liu, H.L., Perfetti, C.A., Spinks, J.A., Fox, P.T., Gao, J.H., 2001b. The neu- Zhou, Y.G., 1978. To what degree are the “phonetics” of present-day Chinese characters
ral system underlying Chinese logograph reading. Neuroimage 13 (5), 826–846. still phonetic. Zhong Guo Yu Wen 146 (3), 172–177.
doi:10.1006/nimg.2001.0749. Zorzi, M., Barbiero, C., Facoetti, A., Lonciari, I., Carrozzi, M., Montico, M., Bravar, L,
George, F, Pech-Georgel, C, Ziegler, J.C., 2012. Extra-large letter spacing improves
reading in dyslexia. Proc. Natl. Acad. Sci. 109 (28), 11455–11459.

11