Enhanced Functional Connectivity of Executive Functions and Attention Networks During Reading Versus Narrative Comprehension in Dyslexia

Narrative and reading comprehension: Definition

Narrative comprehension refers to the capacity to form a meaningful understanding of spoken language. Several more basic forms of language processing are engaged while processing narratives, including speech perception, word recognition, and syntactic processing (Vannest et al., 2009). The initial stages of narrative comprehension, such as speech perception, auditory word recognition, and syntactic processing, involve both linguistic and cognitive abilities (Farah and Horowitz-Kraus, 2019). It is important to note, though, that this intuitive skill also relies on cognitive processes known as executive functions (EF) (Farah and Horowitz-Kraus, 2019).

The ability to comprehend written materials, i.e., reading comprehension, plays a significant role in achieving academic success, attaining college achievements, and enhancing career prospects (Cutting et al., 2009). It is undeniable that broad narrative comprehension skills significantly contribute to children’s reading abilities, with a particular emphasis on text-level reading comprehension skills (Whitehurst and Lonigan, 1998). Storch and Whitehurst (2002) have found that the impact of early oral language skills on reading becomes more significant during the later years of primary school, particularly when children start reading to gain new knowledge. Several theoretical models have been proposed to describe this multifaceted reading comprehension process. First and foremost, to comprehend written materials, there is a need to recognize words automatically, as per the traditional LaBerge and Samuels (1974) model for attention allocation. Once technical reading skills are acquired, reading comprehension involves the construction and the integration phases (Kintsch, 1988). The construction phase includes the mental representation of the text’s meaning, and the integration phase relies on higher-order cognitive abilities to connect and make sense of the information. Both models describe how the ability to comprehend written materials is not only dependent on comprehending language per se but also on more basic and higher-level cognitive processes. The relationship between narrative and reading comprehension was previously described in the Simple View of Reading (SVR) model (Hoover and Gough, 1990). The SVR model suggests that reading comprehension is a cognitive process that requires the utilization of both language comprehension and cognitive skills such as EF.

The simple view of reading model

An integration of narrative and reading comprehension is described in the SVR model (Hoover and Gough, 1990). According to this model, intact decoding and linguistic processing are needed to achieve proficient reading comprehension. Decoding refers to an efficient recognition of words, specifically the ability to access the suitable mental lexicon representation from a sequence of printed graphemes. The other component of the model, linguistic processing, refers to the ability to comprehend and interpret language and extract relevant information from complex linguistic stimuli. Language processing also involves semantic and syntactic processing of the spoken language, including narrative comprehension. Support for the SVR model has grown substantially over the past 20 years (Aaron et al., 1999; Abo‐elhija et al., 2022; Catts et al., 2003; Kim, 2020).

It was recently suggested that EF play a role in the model (Abo‐elhija et al., 2022; Taboada Barber et al., 2021) by supporting not only word reading (Christopher et al., 2012) but also language processing (including narrative comprehension) (Fisher et al., 2019) as well as reading comprehension (Butterfuss and Kendeou, 2018; Spencer et al., 2020).

The involvement of executive functions in reading and narrative comprehension

EF are part of the Attention Network model and include several cognitive abilities supporting learning and controlling behavior (Miller and Cohen, 2001). EF are composed of three main sub-components: (1) inhibition, (2) working memory (WM), and (3) switching (Miyake and Friedman, 2012). These sub-components are the foundation for more complicated functions, including planning, problem-solving, and logical thinking (Diamond, 2013). The Attention Network model also includes more basic attention abilities, including alerting and orienting attention (Callejas et al., 2004; Posner and Petersen, 1990). The literature supported the important role of EF in both narrative and reading comprehension.

EF play an essential role in narrative comprehension processes, in addition to basic linguistic abilities, such as phonological and semantic processing (Marini et al., 2020). Kim (2016) suggested that listeners first orient their attention to an orally presented sound and then remain alert to the speaker. WM is also necessary, as listeners must keep the verbal information in the phonological loop (Kim, 2016). The WM storage is being updated with new verbal information while comprehending the sentences and recalling previous episodes or content to ensure an accurate story organization (Marini et al., 2020). Switching (or shifting), on the other hand, plays a role in constructing coherent narratives by facilitating the creation of complete episodes and selecting informative words (Marini et al., 2020).

EF and attention also play an important role in reading comprehension: evidence for the relationship between WM and reading comprehension was proposed (Cain et al., 2004; Lee Swanson et al., 2006). WM capacity is essential for facilitating comprehension, as it contains several cognitive resources that are simultaneously active during the reading process, such as decoding of new, unfamiliar words, retrieving semantic knowledge from familiar words, and keeping the previously read text in mind (Sesma et al., 2009). In addition, inhibition has been linked to reading comprehension, although relatively few studies investigated these relations in depth (Savage et al., 2006).

The neurobiological correlates for the Attention Network model suggested that EF are associated with the functional connections of the fronto-parietal (FP) and cingulo-opercular (CO) networks (Barbosa et al., 2019) and the dorsal and ventral attention functional brain networks (DAN and VAN, respectively), related to orienting/alerting attention (Posner, 2012). The involvement of these EF and attention networks in both reading and linguistic tasks was previously demonstrated. Farah and Horowitz-Kraus (2019) have demonstrated that these networks were gradually involved in narrative comprehension from 18 months to 9 years, with an increase within CO network’s functional connectivity with age (Farah and Horowitz-Kraus, 2019), suggesting a positive relationship between EF networks and language development from early infancy. Their findings emphasize the crucial role of EF as early as infancy, with a gradual involvement along development as the neural networks support EF maturation later in life. During sentence comprehension, Church (2023) showed that the FP and CO networks are engaged (Church, 2023), whereas more basic attention networks (DAN) are involved in word reading (Vogel et al., 2012) and other resting-state studies showed correlations between word reading improvement and functional connections within the CO (Horowitz-Kraus et al., 2015a) and between language and memory and attention networks related regions after reading intervention (Horowitz-Kraus et al., 2019). We have recently demonstrated how a gradual engagement of EF networks while processing narratives from age 5 to 18 years predicts reading abilities and comprehension at the age of 18 years (Horowitz-Kraus et al., 2024). Taken together, these findings highlight the tight relations between narrative and reading comprehension and the involvement of EF in both processes. Notably, differences in EF involvement have been observed among individuals with reading difficulties. A population suffering from both reading and narrative comprehension (Georgiou et al., 2022), is dyslexia.

Dyslexia: A specific reading difficulty

Dyslexia is a neurobiological-based specific learning disability that impacts word recognition (Habib and Giraud, 2013). Dyslexia is characterized by poor spelling and difficulties in word decoding, i.e., recognizing printed words (Shaywitz and Shaywitz, 2008). Additionally, individuals with dyslexia also face notable challenges in reading comprehension, although their intelligence is intact (Elkind et al., 1993; Snowling et al., 2020). Whereas the primary deficit in dyslexia is phonemic awareness (Bruck, 1992), the performance level of individuals with dyslexia is often exacerbated by various secondary deficits, including deficits in EF (Barbosa et al., 2019; Cutting et al., 2009) and persist into adulthood (Altemeier et al., 2008; Gabrieli, 2009; Horowitz-Kraus, 2014).

Neuroimaging studies examining the role of EF during reading in children with dyslexia demonstrated alteration in several neural regions associated with EF (Asbjørnsen and Bryden, 1998; Dosenbach et al., 2008). These readers showed a decreased activation in the anterior cingulate cortex, a core region within the CO network (Dosenbach et al., 2008). Meri and Horowitz-Kraus (2019) showed that children with dyslexia equally engage EF and attention networks during sentence and word reading. In contrast, the same study showed how different engagements were found in TR, demonstrating their challenge in processing both words and sentences. Children with dyslexia also showed alterations in narrative comprehension. Lower EF abilities and greater activation in brain regions related to the FP network were observed during a narrative comprehending functional MRI (fMRI) task contrary to TR (Horowitz-Kraus et al., 2016).

The goal of the current study is to determine the utilization of EF and attention networks during reading versus narrative comprehension in children with dyslexia group compared to TR. We hypothesized that children with dyslexia would demonstrate a smaller difference between the functional connectivity of EF and attention networks during reading versus narrative comprehension compared to TR. Our hypothesis was based on the essential role of these networks during reading comprehension and the need for higher engagement within and between them. Pinpointing specific neural circuits related to EF and attention, as well as the interaction between these networks, will provide a more in-depth understanding of the specific cognitive processes that lead to difficulties in reading and narrative comprehension in children with dyslexia. Revealing these mechanisms will potentially enable a targeted intervention that may improve both narrative and reading comprehension in these individuals.

Participants

One hundred and 20 children, ages 8–12 years, participated in the current study. All participants were monolingual native English speakers and were divided into two groups: 55 TR (mean age = 9.95 years, standard deviation [SD] = 1.42, 24 females), and 65 children with dyslexia (mean age = 9.51 years, SD = 1.36, 35 females, t(120) = −1.722, p > 0.05). Participants with dyslexia were diagnosed with dyslexia by professionals before joining the study. To verify their inclusion in the dyslexia group, participants scored −1 SD or lower than the mean of at least two behavioral tests assessing reading and phonological abilities, approving their reading impairments. Neurological and psychiatric disorders were exclusionary in this study, and attention difficulties were excluded using the Conners parent report (Conners, 1989). There were no significant differences between the groups regarding average household income (Dyslexia and TR = $75,000–$100,000, t(112) = −0.28, p > 0.05) or maternal level of education (Dyslexia: mean = 17.64 years, SD = 2.37, TR: mean = 18.19 years, SD = 1.97, t(113) = −1.33, p > 0.05) as measured using a demographic questionnaire (following (Romeo et al., 2018)).

Study procedure

The following inclusion criteria were applied: (1) ages 8–12 years, (2) intact vision and hearing, (3) no history of psychiatric or neurological delays, (4) intact nonverbal IQ (>85), (4) born term (>37 weeks gestational age), and (5) no contradiction to MR scanning: claustrophobia, any metal object in the body, including cardiac pacemaker, dental braces, and cochlear implants. After verifying that all participants were eligible to undergo MRI scanning, they underwent a comprehensive behavioral assessment that appraised their EF, reading, and language abilities. Children and their parents signed informed consents and assents before participating. The protocol was approved by the Institutional Review Board, and the data were collected in Cincinnati Children’s Hospital Medical Center, Ohio, USA. Participants were compensated for their participation with $75.

Behavioral measures

A behavioral battery that assessed both language and EF abilities was administered. First, to clarify intact nonverbal IQ, children completed the Test of Nonverbal Intelligence (Brown et al., 2010).

Then, several tests evaluating EF abilities were performed, including (1) WM, using the Digit Span subtest from the Wechsler Intelligence Scale for Children [WISC, (Wechsler, 1999)]; (2) switching and inhibition, using the Stroop Color-Word Interference subtest [Delis–Kaplan Executive Function System, D-KEFS, (Delis et al., 2001)]; and (3) general EF abilities, using the Self-Report Behavior Rating Inventory of Executive Function (Gioia et al., 2000). Speed of processing, an ability related to EF was measured using the Symbol Search subtest [WISC, (Wechsler, 1999)].

To evaluate language abilities, the vocabulary test from the Peabody Picture Vocabulary Test was used, PPVT (Dunn and Dunn, 2007). Reading and reading-related abilities were administered, including (1) rapid letter naming using the Comprehensive Test of Phonological Processing (Wagner et al., 1999); (2) orthographic abilities were assessed using the Test of Word Reading Efficiency [TOWRE-sight word efficiency (SWE), (Torgesen et al., 1999)]; (3) deecoding abilities were measured using the Word Attack subtest from Woodcock–Johnson test (Woodcock and Johnson, 1989); and (4) reading fluency and comprehension abilities were measured using the Gray Reading Oral Test(Wiederholt and Bryant, 2012).

Neuroimaging measures

Neuroimaging tasks

Narrative comprehension task

Children also underwent the stories listening Task inside the scanner following Horowitz-Kraus et al. (2013). The Task included five stories presented without disturbances by an adult female voice in a block design. Each story was presented for 30 sec, followed by a control period of pure tones that lasted for another 30 sec. The stories consisted of 9–11 sentences with differing linguistic structures. On average, 87 words were presented in each story. Children were informed that they would be asked several questions regarding the stories after the scan and, hence, were instructed to be attentive to the stories. Only the stories blocks were used for the analysis (see Fig. 1).

OPEN IN VIEWER

FIG. 1.

An overview of the narrative comprehension task. The Task contains the story listening and the background speech (i.e., control period) both are presented in blocks of 30 sec each. Participants were asked to answer ten multiple-choice questions at the end of the scan.

Reading comprehension task

During the reading comprehension task, children were presented with a block design reading comprehension task and were asked to read texts presented on the screen silently. This Task included three main conditions: (1) reading comprehension, a short text that appears on the screen (5 stories, 44 sec each, average of 84 words per story); (2) a reading fluency condition, a short text that appeared on the screen and was deleted letter by letter (5 stories, 44 sec each, average of 91 words per story) and; and (3) resting state condition, a fixation cross appeared on the screen for 52 sec (5 times). Immediately following each text (the still and deleted conditions), children had to answer a yes/no question that appeared on the screen for 6 sec, and following the fixation condition, children were instructed to push any button for the same amount of time (6 sec). After they responded, a 2-sec Inter Stimulus Interval (ISI) fixation cross appeared before the following trial. The overall duration of the Task was 13 min. Since the current study focused on reading comprehension, we only focused on the still text blocks for our analysis (see Fig. 2 for the description of the Task).

OPEN IN VIEWER

FIG. 2.

An overview of the reading comprehension task. Three task conditions were presented for 5 times each: still or deleted text (44 sec each), followed by a 6 sec yes/no question, and a control condition (52 sec).

Neuroimaging data acquisition

Participants were desensitized before the scan in order to introduce them to the scanner environment in a friendly way while practicing lying on the scanner’s bed in a still position to avoid motion during the scan. To control their head motions, elastic bands affixed to the head of the coil device were used. All scans were acquired using a 3T Philips Ingenia MRI scanner. An MRI-compatible audio/visual system (Avotec, SS3150/SS7100) was used to present the stimuli. A 3D high-resolution, isotropic, T1-weighted turbo field echo anatomical imaging sequence was performed using 8.1 msec repetition time (TR), 3.7 msec echo time (TE), contiguous slices with a 1 mm thickness, and 1 × 1 × 1-mm³ voxel size. A gradient echo planar sequence with a multi-band of 4 was used for a T2*-weighted BOLD fMRI scan with the following parameters: TR/TE = 1000/30 msec, BW = 125 kHz, field of view (FOV) = 20 × 20 × 14.4 cm, matrix = 80 × 80, and slice thickness = 3 mm, voxel size 2.5 × 2.5 × 3 mm³.

Behavioral data analyses

Several analyses were conducted to answer the different research questions as listed in Table 1 and detailed in the following section.

To compare the behavioral measures between the dyslexia and TR groups, t-tests were used. Bonferroni correction was applied to all t-tests by the number of comparisons (n = 12), in order to minimize the probability of Type I error. To examine the association between EF abilities and narrative and reading comprehension abilities between dyslexia and TR, several Pearson correlations were conducted between several behavioral EF abilities and behavioral reading comprehension abilities.

Relations between behavioral EF and neuroimaging narrative and reading comprehension abilities

Pearson’s correlations were calculated between the relevant behavioral measures of EF and the behavioral measures from the neuroimaging narrative and reading comprehension fMRI tasks.

To assess the relationship between EF and narrative and reading comprehension abilities, path analysis using PROCESS (Preacher and Hayes, 2004) was conducted in SPSS to determine if EF moderates the effect of narrative comprehension on reading comprehension. The percentage of correct answers in the narrative and reading comprehension neuroimaging tasks were used as the predictor and the outcome, respectively. A hierarchical linear regression model was used to investigate which EF contributes the most to the variance in reading comprehension. The most meaningful EF component received by the model was used as a moderator in the path analysis.

Neuroimaging data analyses

Functional MRI data preprocessing

Data were preprocessed using SPM12 (Welcome Department of Cognitive Neurology, London, UK). Preprocessing of both tasks included a realignment to the first image of the session to control motion correction using three rotational and three translational parameters, slice time correction, and coregistration of the anatomical image to the mean aligned functional image. In addition, normalization of all images to the Montreal Neurological Institute-152 template version ICBM152 nonlinear 2009c suited for children age five and above was conducted, and smoothing using an 8-mm full width at half minimum spatial smoothing kernel.

Following the preprocessing, task-based data were fed into CONN (Whitfield-Gabrieli and Nieto-Castanon, 2012), a functional connectivity toolbox for MATLAB (Version R2018b, The MathWorks, Natick, MA). To account for potential outliers in the fMRI time series, we used Artifact Detection Tools within CONN. Outlier volumes were identified based on two criteria: (1) framewise displacement (FD) exceeding 0.5 mm, and (2) global signal intensity changes exceeding a z-score of 3. These outlier volumes were included as nuisance regressors in the denoising step to minimize their impact on functional connectivity estimates. Additionally, we applied CompCor denoising to regress out physiological noise from white matter and cerebrospinal fluid, along with six motion parameters and their first-order derivatives. Following denoising, the quality of data was assessed by examining mean FD and the number of scrubbed volumes per participant. Next, regions of interest (ROIs) within and between networks corresponding to EF and attention regions were defined: the FP, CO, VAN, and DAN were defined based on coordinates reported by Power et al. (2011) (see Fig. 3 for the spatial representation of the networks and Supplementary Table S1 for the Supplementary Data for the networks’ coordinates). To assess the impact of motion during scanning, an average calculation of FD was performed in both tasks across all participants and revealed an average of 0.18 for the narrative comprehension task, and 0.19 for the reading comprehension task. Independent-sample t-tests revealed no significant difference between the groups [narrative comprehension: children with dyslexia (mean = 0.189, SD = 0.04), TR (mean = 0.183, SD = 0.042) (t(120) = 0.807, p = 0.42), reading comprehension: children with dyslexia (mean = 0.193, SD = 0.04), TR (mean = 0.192, SD = 0.04) (t(120) = 0.122, p = 0.90)].

OPEN IN VIEWER

FIG. 3.

A graphical representation of the ROIs composing the EF and attention networks based on Power’s atlas. The EF networks include the fronto-parietal network (in blue), and the cingulo-opercular network (in green). The attention networks include the ventral attention network (in pink), and the dorsal attention network (in purple) are presented in lateral, medial, and dorsal views. EF, Executive functions; ROIs, regions of interest.

Moderated mediation analysis

After confirming that behavioral EF abilities affect the relationship between narrative and reading comprehension, a subsequent Path analysis using PROCESS procedure was conducted to examine whether these cognitive processes are moderated by EF and attention networks. The same narrative and reading measures used in the previous path analysis were used in the current analysis. To determine which network contributed the most to the variance in reading comprehension, a hierarchical linear regression model was conducted. The functional connectivity values of within and between networks during both narrative and reading comprehension were used. The functional connectivity of the most significant network was used in the path analysis.

Later, a mediation-moderation analysis combining narrative, reading, behavioral EF, and neuroimaging EF was conducted to assess the relationship between these measures.

Behavioral data

The two reading groups demonstrated equal general intelligence and attention abilities. As expected, TR demonstrated higher receptive vocabulary scores (PPVT). In addition, children with dyslexia exhibited significantly lower scores in the behavioral oral reading fluency and comprehension test (comprehension). Moreover, children with dyslexia also exhibited lower general EF abilities and lower switching and inhibition abilities (see Table 2).

Neuroimaging data

Moderation path analysis

Prior to the moderation path analysis, a hierarchical linear regression was conducted. At each level in the model, a different EF measure was added. The results revealed that only WM significantly contributed to the model and accounted for 9% of the variation in reading comprehension (R ² = 0.09, p = 0.035). Therefore, the EF component used in the path analysis was WM. The results of the moderation model suggested that narrative comprehension abilities (measured via the percentage of correct answers in the fMRI narrative comprehension task) significantly predicted reading comprehension abilities (measured via the percentage of correct answers in the fMRI reading comprehension task) (β = 6.65, p = 0.000, SE (β) = 1.49), and this effect was significantly moderated by WM (β = 1.56, p = 0.001, SE (β) = 0.435) (see Fig. 4).

OPEN IN VIEWER

FIG. 4.

A graphical representation of the behavioral path analysis for the relations between narrative comprehension and reading comprehension moderated by EF. The tested path is represented with a dashed line.

The consecutive path analysis revealed that narrative comprehension significantly predicted reading comprehension abilities (β = 0.406, p = 0.012, SE (β) = 0.159), and this relationship was moderated by the functional connectivity within the DAN network (β = 2.73, p = 0.019, SE (β) = 1.15). The hierarchical linear regression model revealed that the DAN network during narrative comprehension contributed the most to the variance in reading comprehension (R ² = 0.26, p = 0.000) (see Fig. 5).

OPEN IN VIEWER

FIG. 5.

A graphical representation of the neuroimaging path analysis for the relations between narrative comprehension and reading comprehension moderated by an EF network. The tested path is represented with a dashed line.

Finally, the mediation-moderation analysis including the neuroimaging component revealed significant results for the moderated mediation (β = 5.74, 95% confidence interval [0.80, 11.46]). The relationship between behavioral EF (i.e., WM) and narrative comprehension was significantly moderated by the EF network (i.e., DAN, β = 2.73, p = 0.033). There was a significant direct effect of narrative comprehension on reading comprehension (β = 0.51, p = 0.004). No significant direct effect was found between behavioral EF and reading comprehension (β = −1.58, p = 0.083). The model explained 6% of the variance in reading comprehension abilities (R ² = 0.06) suggesting that the relationship between narrative and reading comprehension is moderated by functional connectivity within the EF network (see Fig. 6).

OPEN IN VIEWER

FIG. 6.

A graphical representation of the mediation-moderation analysis for the relations between narrative comprehension, reading comprehension, and behavioral executive functions moderated by an EF network. The tested paths are represented with dashed lines.

Functional connectivity results

Differences in functional connectivity within and between networks in children with dyslexia versus TR

An independent-sample t-test revealed that children with dyslexia showed significantly higher functional connectivity between VAN-FP networks in the reading comprehension task than TR (see Supplementary Table S2 and Fig. 7).

OPEN IN VIEWER

FIG. 7.

Correlation coefficient matrices for each Task per reading group (A, B, C, D) and T statistics map for between-group comparison (E, F) per Task and within-group comparison (G, H) in different tasks.

Greater utilization of EF and attention networks in reading versus narrative comprehension in children with dyslexia

The findings in the present study indicate that children with dyslexia notably exhibited higher levels of functional connectivity between DAN-FP, VAN-DAN, CO-FP, and CO-DAN during reading versus narrative comprehension tasks, compared to TR. These neuroimaging results imply that children with dyslexia involve all EF and attention networks during the reading comprehension task, highlighting their need for greater involvement of higher-order cognitive networks during reading comprehension.

The CO network is a key component in higher-order EF networks, playing a crucial role in multiple EF-related processes, including error detection (Horowitz-Kraus et al., 2015b), task maintenance (Marek and Dosenbach, 2019), and WM (Wallis et al., 2015). This network showed higher functional connectivity with the DAN and FP networks during reading comprehension. The FP network is associated with cognitive flexibility and WM, and the DAN is linked to alerting attention. The aforementioned abilities and their synchronization hold significant importance in the process of reading comprehension. Collectively, these findings indicate that the integration of multiple cognitive processes is essential for achieving complex abilities like reading comprehension, which in turn rely on distinct large-scale neural networks.

The results of the current study also revealed that children with dyslexia significantly exhibited higher levels of functional connectivity between the VAN and FP networks during reading comprehension than TR. The VAN network is involved in the automatic orienting of attention towards a relevant stimulus. Crucially, these fMRI results provide further evidence that EF and attention networks play a significant role in reading comprehension processes in individuals with dyslexia, which aligns with previous literature in the field, for example, Butterfuss and Kendeou (2018) suggested that EF play an important role in integrative processes associated with reading comprehension. Meri and Horowitz-Kraus (2019), children with dyslexia equally utilize EF and attention networks in both contextual and isolated word reading, highlighting their need for the engagement of these networks.

The behavioral results found in the current investigation strengthen the fMRI results, as children with dyslexia exhibited significantly lower scores in the oral reading fluency and comprehension tests. This baseline result may be an indication of their need for higher functional connectivity in EF-related networks during the reading comprehension process. Furthermore, they showed lower accuracy rates in the reading comprehension fMRI task compared to TR. This result highlights their impaired reading comprehension abilities and their potential requirement for increased involvement in EF-related networks.

Greater utilization of EF and EF-related networks during narrative comprehension versus reading comprehension in TR

In support of previous evidence concerning the crucial role of narrative comprehension as a precursor to intact reading comprehension (Lynch et al., 2008), this study’s findings showed that TR significantly exhibited higher functional connectivity within VAN, and between VAN-FP in the narrative compared to the reading comprehension task. These findings suggest that TR engage all EF and attention neural networks when comprehending oral language. There are several plausible explanations for the fact that TR engaged more EF and attention networks during narrative comprehension than during reading comprehension. First, it could be that a proficient narrative comprehension process in these individuals may have established a foundation that contributes to a more effortless process of reading comprehension. This suggestion aligns with past literature investigating the contribution of narrative comprehension to reading comprehension skills. Kendeou et al. (2009) showed that advanced language skills, such as advanced narrative comprehension, play a crucial role in an individual’s overall language proficiency and can provide an additional and distinctive contribution to the process of reading comprehension. This was reinforced in another study by Catts et al. (2015), which showed that reading comprehension can be reliably predicted by assessing the precursors of word reading and language comprehension. Interestingly, the results of the current study also confirm previous findings showing that TR present higher baseline reading abilities compared to children with dyslexia, emphasizing their automatic reading processes. For example, Horowitz-Kraus and Holland (2015) demonstrated greater functional connectivity between reading and error-detection regions after reading intervention only in children with dyslexia, suggesting a higher starting point for TR.

Second, the VAN network is primarily active in the right hemisphere of the brain and is typically engaged when an unexpected event takes place (Farrant and Uddin, 2015). However, this engagement is hindered when there is a high workload on the individual’s WM (Hutton et al., 2019). WM is responsible for maintaining and processing information in complex cognitive tasks, including reading and reading comprehension. It is possible that the observed higher functional connectivity in the VAN during narrative versus reading comprehension can be attributed to the excessive cognitive load experienced during the reading comprehension process as the written information is available for the reader, whereas the narrative one does not.

Simple view of reading model, reading comprehension, narrative comprehension, and EF

The findings in the current study reinforced the original SVR model, emphasizing the pivotal role of narrative comprehension in reading comprehension processes. This statement is reflected in the results as TR showed higher functional connectivity during narrative compared to reading comprehension task. It is possible that their proficient narrative comprehension skills facilitated their reading comprehension process. Therefore, they show a lower functional connectivity in EF and attention networks.

However, the results of this study provided an additional important component to this model. The original SVR model has been challenged and recent studies have suggested that EF also play a critical role in the model (Cirino et al., 2019; Taboada Barber et al., 2021). The results of this investigation also reinforced this suggestion, as the path analysis results showed that the effect of narrative comprehension on reading comprehension was moderated by an EF component, WM. Crucially, these results were augmented and supported by the involvement of EF and attention networks during narrative and reading comprehension, as it was proven that the functional connectivity within DAN moderated the relationship between these two cognitive processes. Furthermore, a subsequent mediation-moderation model that included narrative and reading comprehension abilities, as well as both behavioral and neuroimaging EF components, highlighted the tight relationship between them. Both behavioral and neuroimaging EF components played a prominent role in the effect of narrative comprehension on reading comprehension, emphasizing its essential contribution to these language processes.

Previous evidence supporting the SVR model, which categorized children with dyslexia as children with deficits in word recognition but intact language comprehension (Catts et al., 2006), is supported by the results in the current study. Children with dyslexia exhibited higher functional connectivity in the reading comprehension task compared to the narrative comprehension one. This may explain the functional connectivity pattern found only in the reading comprehension task and the need for an overall EF and attention networks involvement only in this Task. Perhaps the fact that their narrative comprehension was intact has positively contributed to their lower functional connectivity observed in this Task. Another possible and reasonable explanation is the involvement of EF abilities during reading comprehension. It was also found that children with dyslexia showed significant positive correlations between EF and reading comprehension measures. This implies that individuals with high reading comprehension also require high EF abilities, highlighting the significant role of EF in the reading comprehension process. Consequently, this provides further evidence supporting the inclusion of EF in the updated SVR model (see Fig. 8).

OPEN IN VIEWER

FIG. 8.

A suggested updated SVR model, including EF abilities. SVR, Simple View of Reading.

Study limitations

There are several limitations in this study. First, the experimental fMRI reading comprehension task did not include only a still text condition. Future studies focusing exclusively on reading comprehension and including an ongoing, natural reading comprehension condition are warranted. Second, because only English-speaking individuals were recruited, it is unknown whether the findings of the current study can be projected to other languages. Third, it is uncertain whether the relationship between narrative comprehension, reading comprehension, and EF discovered in this study will persist in older ages. To deepen our understanding on this topic, longitudinal studies are required. Fourth, the narrative comprehension task had a shorter acquisition time compared to the reading comprehension task. To ensure enough data are collected for both tasks, future studies should incorporate tasks with longer acquisition times.