Longer screen vs. reading time is related to greater functional connections between the salience network and executive functions regions in children with reading difficulties vs typical readers

Abstract

An adverse relationship between screen exposure time and brain functional/structural connectivity was reported in typically developing children, specifically related to neurobiological correlates of reading ability. As children with reading difficulties (RD) suffer from impairments in reading and executive functions (EF), we sought to determine the association between the ratio of screen time duration to reading time duration and functional connectivity of EF networks to the entire brain in children with RD compared to typical readers (TRs) using resting state data. Screen/reading time ratio was related to reduced reading and EF abilities. A larger screen/reading time ratio was correlated with increased functional connectivity between the salience network and frontal-EF regions in children with RD compared to TRs. We suggest that whereas greater screen/reading time ratio is related to excessive stimulation of the visual processing system in TRs, it may be related to decreased efficiency of the cognitive control system in RDs.

Introduction

Screen exposure and the reading network – a “competition” for EF?

A growing number of reports suggest that children around the world are spending an excessive amount of time in front of digital screens[1], with national reports of an average screen time per day of 3.20 hours (SD = 2.40 hours)[2]. Hence, various recommendations have been advanced [1] about the appropriate amount of time children should spend viewing digital content. This media overload has been suggested to be associated with mental illness[3], chronic physical conditions (e.g. diabetes)[4], sleep disorders[5] and more. One suggestion is that increased stimulation by screens may be related to restless behavior and inattentive symptoms[6] but the mechanism underlying this suggestion is not known. Only a few neuroimaging studies have attempted to reveal the association between screen exposure and brain activation, functional connectivity, or structural connectivity in healthy, typically developing children. In a group of young children (3–5 years old) undergoing diffusion tensor imaging (DTI), decreased structural connectivity and fractional anisotropy within white matter tracts associated with language (arcuate and inferior longitudinal fasciculi), executive functions (uncinate) and visual processing (the occipital end of the arcuate and the inferior longitudinal fasciculi) were found to be associated with increased screen exposure time[7]. As these white matter tracts connect the gray matter regions related to reading ability, the authors suggested that increased screen time might be related to decreased organization of white matter tracts originally used for literacy and future reading[7]. They concluded that this may pose a risk to future reading skills in those exposed to excessive screen time.

The Adolescence Brain Cognitive Development (ABCD) study found decreasing grey matter integrity in visual processing regions in nine-year-old children with increasing screen exposure, suggesting negative consequences of intense visual processing[8]. Prior studies found that increased electronics use/screen time and decreased reading time resulted in poorer reading abilities[9][10]. Augmented utilization of visual processing regions associated with screen exposure was also tested during a resting state condition in 8–12 year old children[11]. With increased screen time[11], typically developing children showed decreased functional connectivity between the fusiform gyrus (also referred to as the visual word form area (VWFA), part of the associative visual processing network[12]) and regions related to executive functions (EF) and language processing. Reading time, on the other hand, was associated with increased functional connections between the VWFA and similar EF regions. The authors suggested that screen and reading time may “compete” for influence on neural circuits supporting reading as these activities may modulate similar regions but in opposing ways[11]. The visual networks discerned in the resting state, can be divided into the associative visual network (composed of the VWFA and Brodman’s Area (BA) 19) and the primary visual processing network (which includes BA 17/18), both of which are part of the lower-level sensory processing network[12]. An assessment of directional connections between resting-state networks suggested that the associative visual processing network modulates the functional connections of the primary visual processing network, the auditory network (all in the lower-level sensory network) as well as the ventral attention network (part of the higher-order cognitive control level)[12]. The primary visual network, on the other hand, was found to directly modulate the functional connections of the default mode network[12]. The primary visual network is, in turn, modulated by inputs from the auditory network and the ventral attention network, which makes this network the only lower-level sensory network directly modulated by higher-order cognitive networks[12]. Hence, the work described in [11], only outlines the relationship of reading and screen exposure to functional connections between the associative visual processing and the EF-related regions in TRs. However, the functional connections between the primary visual processing network and higher-order cognitive networks in the context of screen and reading time ratio is still unknown.

Critically, due to the rapid changes in stimulus presentations when using digital media and the need to respond quickly in an adaptive way (for a review, see [13]), neural circuits related to attention switching, orienting, executive processing and, most of all, navigating between these networks, may be affected. It has been suggested that the salience network, one of the cognitive control networks, may be related to allocating attention from one source to the other, from internal to external processing[14]. Moreover, greater functional connections within the salience network were found in typical readers (TRs) versus children with reading difficulties (RD) while processing stories[15]. This finding is not surprising as the latter population also suffers from a general EF challenge[16].

Children with RD and EF

Reading difficulties (or dyslexia) are defined as continuous challenges in acquiring the alphabetical principle, and slow and inaccurate reading despite appropriate exposure and intact intelligence[17]. In addition to their specific difficulty in reading, readers with RD also demonstrate altered EF abilities[16, 18]. Several studies have suggested that error monitoring among readers with RD is reduced both while making reading errors[19] and during non-linguistic tasks[16], and that this population also demonstrates decreased functional connections within networks related to EF (e.g. the cingulo-opercular network; during rest[20]) and reading[21]). Alterations in functional connectivity of other EF networks, such as between seeds of the fronto-parietal network, were also associated with comprehension abilities in children with RD during stories listening [22] and during a specific EF task such as the Stroop task[18]. Recently, the literature related to the involvement of other cognitive control networks in reading has debated the level of functional connections of the salience network during cognitive tasks in children with RD. It has been shown that readers with RD have decreased functional connections within the salience network while listening to stories, suggesting altered modulation of the top-down and bottom-up processes by the salience network during this task (i.e. EF and attention networks, respectively) [23]. However, increased overall functional connectivity between the combined EF and cognitive control (salience) networks and seeds related to reading was observed during an EF task[18]. These outcomes align with the findings that greater challenges in performing a task are related to increased functional connections between EF and cognitive control networks and other task-related brain regions[24]. As the EF challenges were also observed in children at-risk for RD[25], it is assumed that challenges in subcomponents of EF are part of the disorder[16]. As outlined above, screen exposure seems to recruit similar neural networks as those used for reading[11], which may mean that the longer time the readers with RD spend using screens, the more they are straining their already overtaxed neural networks.

Hence, the goal of the current study is to define the neurobiological correlates of cognitive control and EF and visual processing modulation by the screen/reading time ratio in children with RD vs. TRs. Based on the reports showing the relationship between decreased reading and increased usage of electronics and future RD[9, 10], we hypothesized that children with RD will have greater screen vs. reading time compared to TRs, and that a higher ratio will be associated with reduced reading and EF ability. We also postulated that children with RD will demonstrate increased functional connectivity between the cognitive control (salience) network and regions associated with EF is relations to screen vs reading time ratio compared to TRs. On the other hand, we suggest that TRs will demonstrate an increased positive association between the screen vs. reading time ratio and functional connections between the salience network and visual processing regions, including regions in the primary visual processing network (see the previously reported association of the primary visual network with both higher-order and lower-level networks[12]), compared to children with RD, due to the overstimulation resulting from screens.

Methods

Participants

Children with RD (N = 29; mean age = 9.77 years, SD = 1.35 years) and typical readers (N = 28; mean age = 10.0 years, SD = 1.33 years) participated in the current study between the years 2017–2019. There was no significant age difference between the two groups [t(55) = 0.665, p = ns]. All participants were monolingual native English speakers. The same behavioral and neuroimaging metrics were used for both groups. Participants had no history of neurological or psychiatric impairments, nor any visual/auditory processing deficits. Following the criteria described in [26], participants were included in the RD group if they had a standard score of −1 or below in at least two of the administered reading tests (for the list of reading tests, see reading measures section below). Informed consent and assent were given by the parents and participants. The study was reviewed and approved by the Cincinnati Children’s Hospital Institutional Review Board.

Behavioral Measures

Executive functioning measures.

Executive functioning was assessed using several subtests. 1) inhibition: walk/don’t walk task; 2) auditory attention: the score! subset; 3) visual attention: the sky search DT subtest from the Test of Everyday Attention for Children (TEA-CH)[27]; 5) switching/inhibition: the Stroop subtests[28]; 6) fluency: the fluency subtests for letters and categories[28]; 7) working memory: the digits span subtest[29]; 8) nonverbal processing speed: the coding and symbol search subtests[29]; 9) verbal processing speed: the number naming and letter naming subtests from the CTOPP battery[30]; and 10) overall EF: the general cognitive score from the Behavior Rating Inventory of Executive Function (BRIEF)[31].

Reading measures.

Reading abilities were assessed via several subtests evaluating several reading subdomains: 1) phonological processing: using the Elision subtest from the Comprehensive Test of Phonological Processing (CTOPP)[30]; 2) timed decoding: using the phonetic decoding efficiency [PDE] subtests from the Test of Word Reading Efficiency (TOWRE)[32]; 3) nontimed decoding: using the word-attack subtest from the Woodcock Johnson (WJ-III) battery[33]; 4) timed orthographical abilities: using the sight word efficiency [SWE], from the Test of Word Reading Efficiency (TOWRE)[32]; 5) nontimed orthographical abilities: using the letter–word subtest from the Woodcock Johnson (WJ-III) battery[33]; and 6) timed reading comprehension: using the TOSREC[34] and nontimed reading comprehension: using the WJ-III batteries[33].

Screen/reading time ratio

Parents were asked the average hours per week, in the past 6 months, their children spent reading hard-copy materials such as books, newspapers, or any other paper-based reading materials, including during homework. They were also asked to report the average hours per week, in the past 6 months, their children were using or were exposed to screens, including the use of computer, television, and smart devices. Both passive and active screen use were quantified (i.e. watching television, playing games on the computer, or using devices for homework). This information was collected via questionnaires administered to parents with explicit instructions to consider time spent on any of the above reading or screen exposure modalities per week. A measure of screen time divided by reading time was calculated for each participant, and compared between the two groups.

Neuroimaging data

Resting state data acquisition

All images were acquired using a Philips Ingenia 3T MRI scanner (Philips Medical Systems, Best, The Netherlands). A T2*-weighted, gradient-echo, echo planar imaging (EPI) sequence was used for resting-state fMRI with parameters: TR/TE = 700/30 msec, slice thickness = 3 mm, resulting in a voxel size of 2.5 × 2.5 × 3 mm³.

Two five-minute resting-state scans were acquired, resulting in a total of 430 whole-brain volumes, acquired in a total imaging time of 10 minutes. In addition, a high-resolution T1-weighted 3D anatomical scan was acquired using an inversion recovery (IR)-prepared turbo gradient-echo acquisition protocol with a spatial resolution of 1 × 1 × 1 mm³. Participants were acclimated and desensitized to the scanner to condition them for comfort during imaging[35].

Functional Connectivity Analysis

Data were preprocessed using SPM12(http://www.fil.ion.ucl.ac.uk/spm/)[36] implemented in CONN toolbox[37]. Normalized bias-corrected T1 images were generated in SPM and segmented into gray matter, white matter, and cerebral spinal fluid (CSF). The first 16 eigenvariates of the BOLD time-courses from white matter and CSF, as well as six motion-correction parameters, were included as regressors of no interest. The functional data were then band-pass filtered between 0.008 and 0.2 Hz (as recommended in [38]).

Independent component analysis (ICA).

The post-processed images were submitted to a subject-wise group ICA implemented in the MATLAB-scripted (MathWorks, Natick, MA; https://www.mathworks.com/products/matlab.html) functional connectivity (CONN) toolbox, Version 17f in [38]. CONN includes a predefined network atlas, with the salience network comprised of the following seven MNI regions-of-interest (ROIs): 1) anterior cingulate cortex (0, 22, 35); 2) left anterior insula (−44, 13, 1); 3) right anterior insula (47, 14, 0); 4) left rostral prefrontal cortex (−32, 45, 27); 5) right rostral prefrontal cortex (32, 46, 27);6) left superior marginal gyrus (−60, −39, 31); and 7) the right superior marginal gyrus (62, −35, 32).

The group ICA was performed with 20 components and a dimensionality reduction of 64. Out of the 20 ICs, only one IC corresponded to the salience network and was selected for further analyses. The IC corresponding to the salience network was determined via correlational spatial match-to-template using the predefined networks provided by the CONN toolbox[37]. See Figure 1 for the network.

Figure 1. Definition of the Salience network (axial axis).

Z coordinates are listed below each slice.

Statistical analysis of the behavioral data

To find differences between children with RD and TRs in EF and reading measures, independent two-sample t-tests were performed on the behavioral data comparing the two groups, corrected for multiple comparisons using a Bonferroni correction.

Correlation between reading and EF behavioral measures and screen/reading time ratio

To find the relations between reading, EF and screen/reading time ratio, Pearson’s correlations were performed between these measures across both groups. Bonferroni correction was applied.

Statistical analysis of the neuroimaging data

To address the current study’s questions related to the association between screen/reading time ratio and functional connectivity between the salience network and the whole brain, a seed-to-voxel analysis (where the seed is defined as the salience network represented by the IC identified above) using the screen/reading time ratio as a covariate of interest was conducted for the following contrasts: 1) children with RD only, 2) TR only, 3) children with RD vs TRs and 4) TRs vs children with RD.

Results

Behavioral data

Reading and EF measures:

Children with RD showed significantly lower reading ability in all examined measures compared to TRs. They also demonstrated decreased EF in several domains (Naming, visual attention, inhibition, fluency, inhibition/switching, working memory). No significant differences between the groups were observed in reading comprehension. See Tables 1 and 2.

Table 1.

Reading ability, screen and reading time in children with RD and TRs

		Typical readers (N=28)	Children with RD (N=29)
Category	Measure	Mean ± SD	Mean ± SD	T(P)
Demographics	Age (in years)	10.0±1.336	9.77±1.357	.665
	Socio-economic status (average income per household)	7.84±1.748	7.76±1.964	.152
	Socio-economic status (maternal education (Years)	17.84±2.23	17.42±2.969	.565
General abilities	Nonverbal ability, TONI (Scaled score)	106.66±19.01	102.07±14.229	1.052
	General verbal abilities, PPVT (Scaled score)	119.46±17.015	110.2±14.864	2.212
Attention	Conners (Parent, T Score)	52.63±11.92	58.32±16.817	−1.452
Reading fluency	TOSREC (scaled score)	100.86±11.94	80.5±15.433	5.655***
Phonological processing	CTOPP, Ellison, (Percentile)	63.0±21.044	31.47±27.378	4.679***
Naming	CTOPP, Number naming (Scaled Score)	9.79±2.691	6.6±2.699	4.544***
Orthographical, timed	TOWRE, SWE (Scaled Score)	103.03±9.466	79.3±14.307	7.548***
	TOWRE, SWE (number of errors)	.69+.930	2.29±2.432	−3.251**
Decoding, timed	TOWRE, PDE (Number of errors)	4.24±3.28	7.75±5.961	−2.74**
	TOWRE, PDE (Scaled Score)	101.52±9.23	78.47±11.866	8.344***
	TOWRE Efficiency Index (Scaled Score)	103.517±11.636	78.6±27.758	4.523***
Orthographical, non-timed	Woodcock Johnson, Letter Word (Number of errors)	10.55±3.077	11.25±4.835	−.648, p=.52
	Woodcock Johnson, Letter Word (Standard Score)	113.03±11. 378	84.17±17.92	7.358***
Comprehension	Woodcock Johnson Passage Comprehension Time, (Seconds)	662.89±217.172	747.7±303.915	−1.187, p=.241
	Woodcock Johnson Passage, Comprehension, (Number of errors)	11.0±2.841	9.0±2.854	2.561**
	Woodcock Johnson Passage Comprehension (Scaled Score)	104.76±12.417	84.4±16.122	5.421***
Decoding, non-timed	Woodcock Johnson, Word Attack, Time (Seconds)	119.41±36.169	149.77±56.196	−2.352**
	Woodcock Johnson, Word Attack (Number of errors)	5.17±2.916	8.82±1.945	−5.575***
	Woodcock Johnson, Word Attack (Standard Score)	116.24±15.64	87.4±16.404	6.908***
Screen vs reading	Ratio between hours per week spent on screens vs. reading	3.79±5.826	5.255±5.827	−.940
	Number of books the child has	59.34±55.786	37.97±27.865	1.846 p = .07
	Number of hours the child reads during the week	3.46±2.415	3.36±3.49	.128
	Number of hours the child spends watching TV and playing on the computer	9.76±11.398	10.38±7.975	−.240

TONI: Test of Nonverbal Intelligence; PPVT: Peabody Picture Vocabulary Test; TOSREC: The Test of Silent Reading Efficiency and Comprehension; CTOPP: Comprehensive Test of Phonological Processing; TOWRE: Test of Word Reading Efficiency; SWE: Sight Word Efficiency; PDE: Pseudowords Decoding Efficiency

^*p<.05;

^**p<.001;

^***p<.001

Table 2.

Executive functions in children with RD and typical readers

		Typical readers (N=28)	Children with RD (N=29)
Ability	Measure	Mean ± SD	Mean ± SD	T(P)
Working memory	Digit Span, WISC (Standard score)	10.93±2.549	9.34±2.991	2.174*
Speed of processing	Coding, WISC, (Standard score)	9.48 ±2.309	8.24±3.055	1.746 p = .086
Speed of processing	Symbol Search, WISC (Standard score)	11.31±2.727	10.31±2.436	1.473 p = .146
Switching / inhibition, time	Stroop, Color naming, DKEFS, Time (Standard score)	12.36±9.015	7.79±4.411	2.441*
	Stroop, Word naming, Time, DKEFS (Standard score)	12.25±4.452	7.9 ± 3.098	4.298***
	Stroop, Color Word inhibition, DKEFS, Time (Standard score)	13.07±13.757	8.24±3.757	1.822 p = .074
	Stroop, switching, DKEFS, Time (Standard score)	13.29±12.884	9.0 ± 2.659	1.754 p = .08
Switching / inhibition, accuracy	Color Naming and Reading, DKEFS (Composite score)	11.21±2.149	8.24±4.059	3.438**
Fluency	semantics, DKEFS, Total Correct (Standard score)	11.28±2.477	8.97±2.897	3.264**
	Accuracy Switching, DKEFS (Standard Score)	11.66±2.092	11.17±7.843	.32 p = .75
	Phonemic, Summary, DKEFS, first interval, total (Standard score)	11.76 ±3.69	9.55 ±3.46	2.349*
	Phonemic Summary, DKEFS, second interval, total (Standard score)	10.07±2.789	8.62±3.144	1.856 p = .06
	Phonemic Summary, DKEFS, forth interval, total (Standard score)	10.21±2.731	8.62±3.087	2.073*
	Phonemic, Summary, DKEFS, percent category switching (Scaled score)	11.52±1.214	9.29±3.242	3.419**
Inhibition	Trail Making, DKEFS (Scaled score)	11.24±2.355	9.38±3.736	2.271*
Spatial attention	Design Fluency, DKEFS, total set loss (Scaled score)	8.0±3.665	5.79 ±3.83	2.242*
	Design Fluency, DKEFS, total repeated words (Scaled score)	11.31±2.436	10.97±2.692	.511 p = .611
	Design Fluency, DKEFS, total attempted (Scaled score)	12.0±3.251	11.72±3.104	.33 p = .74
	Design Fluency, DKEFS, percent accuracy (Scaled score)	7.34±3.801	5.10±3.255	2.412*
Visual attention	Sky Search, time per target (Scaled score)	8.07 ±3.012	6.96±2.769	1.423 p = .16
	Sky Search, TEACH, Attention (Scaled score)	8.46 ±2.983	6.93±2.775	1.957*
Auditory attention	Score, TEACH (Standard score)	8.88 ±2.819	7.61±2.859	1.652 p = .105
Naming	Object, RAS (Standard score)	94.793±12.3906	87.037±11.2711	2.444*
	Colors, RAS (Standard Score	95.0± 15.8272	83.889±13.127	2.848**
	Numbers, RAS (Standard score)	100.0 ± 14.7455	88.111±12.042	3.29**
	Letters, RAS (Standard score)	94.552±11.456	86.815±11.7048	2.499**
	Multiple alternating stimuli, RAS (Standard score)	102.207±20.013	87.259±13.7829	3.232**
Sustained attention	Detectability, CPT (T Score)	48.82±16.947	55.14±17.19	−1.305
	Omissions, CPT (T Score)	51.9± 20.913	61.1 ± 23.215	−1.461
	Commission, CPT (T Score)	44.73±15.407	50.03±15.463	−1.210
	Preservation, CPT (T Score)	49.47±19.408	54.92±20.403	−.962
	Variability, CPT (T Score)	50.12±18.179	58.21±20.732	−1.455
General Executive functions	Parents report, General cognitive, BRIEF (T Score)	48.07±8.185	52.52±10.563	−1.769 p = .08

WISC: Wechsler Intelligence Scale for Children; TEACH: Test of Everyday Attention; DKEFS: Delis–Kaplan Executive Function System; CPT: Continues Performance Test; RAS: Rapid Automatized Naming and Rapid Alternating Stimulus Tests; BRIEF: Behavior Rating Inventory of Executive Function

^*p<.05;

^**p<.01;

^***p<.001

Screen vs. reading time ratio:

Paired t-tests demonstrated a significantly greater screen exposure time compared to reading time in TRs (t = −2.736, p < .01) as well as in children with RD (t = −4.151, p < .001). No significant difference in the screen time vs. reading time ratio was found between the groups; see Table 1.

Correlation between screen vs reading time ratio and reading ability across both groups

A significantly positive correlation between the screen vs. reading time ratio and reading deficiency (number of errors for the TOWRE site word reading task;r = .247, p < .05) and a trend towards correlations with reading time (letter–word from the WJ battery; r = .189, p = .08) was found. A larger ratio between screen and reading time was related to more reading errors and longer reading time across both groups.

Correlation between screen and reading time ratio and EF abilities across both groups

Significantly negative correlations between the screen and reading time ratio with fluency (phonemic fluency, DKEF, r = −.224, p < .05; semantic fluency, DKEF r = −.223, p < .05), working memory (digit span, WISC r = −.244, p < .05), inhibition/switching (trail making, DKEF r = −.262, p < .05), switching (color–word, Stroop DKEF r=−.238, p<.05), speed of processing (number–letter, CTOPPr = −.228, p < .05; symbol search, WISC r = −.238, p < .05), and positive correlations with attention abilities (CPT, preservative scores r = .232, p < .05; block change r = .325, p < .05) and parents’ reports for general EF abilities (BRIEF parents reports r = .376, p < .01) were found across both groups. Children with a higher screen/reading time ratio showed decreased abilities in all the above measures.

Neuroimaging data results

Association between screen vs. reading time ratio and functional connectivity between the salience network and the whole brain (seed-to-voxel)

A significant association between the screen vs. reading time ratio and the degree of functional connectivity of the salience network to voxel clusters across the whole brain (seed-to-voxel) was found for the whole cohort [T(57) = 3.48, p < .05 FWE corrected].

Children with RD:

This group showed positive correlations between screen vs. reading time ratio and functional correlations between the salience network and bilateral cognitive control regions, associative visual processing regions, and negative functional connections with sub-cortical cognitive control, language and visual regions (primary visual cortex).

TRs:

These readers showed positive correlations between screen vs reading time ratio and functional connectivity between the salience network and right language regions, regions in the left associative and primary visual networks and motor-related regions, whereas negative correlations were found with right and left language regions, bilateral cerebellum, and right subcortical regions (i.e. thalamus).

Children with RD vs TRs:

When comparing the correlations between the two groups, children with RD, showed greater positive correlations between screen vs reading time ratio and functional connectivity between the salience network and frontal-EF regions compared to TRs.

TRs vs children with RD:

TRs showed greater positive correlations between screen vs time ratio and functional connections between the salience network and the occipital regions specifically in the primary visual cortex compared to children with RD; see Table 3, Figure 2 and supplementary figures 1–2 for these details.

Table 3.

Correlations between screen vs reading time ratio and functional connectivity between the salience network and the whole brain (seed-to-voxel) in children with RD, TRs, and the comparison between the two groups (RD>TR, TR>RD).

Group	Condition	Anatomical region	X	Y	Z	Number of Voxels
Children with RD	Positive correlation	Right frontal pole (BA 8)	−6	50	40	18120
		Cingulate gyrus (post division, BA 24)	2	−18	40	614
		Right superior frontal gyrus (BA 6)	16	18	58	318
		Left Middle temporal gyrus (fusiform gyrus, BA 19)	−30	−56	14	237
		Right cerebellum	36	−52	−38	221
		Left postcentral gyrus (BA 2)	−54	−26	50	220
		Left superior frontal gyrus (BA 6)	−6	22	62	141
		Left precuneus (19)	−12	−54	2	131
	Negative correlation	Right lateral occipital cortex (angular gyrus, BA 39)	36	−60	36	1642
		Right putamen	22	−14	6	560
		Left thalamus	−18	−10	−2	465
		Right middle frontal gyrus (BA 46)	56	32	28	370
		Left middle temporal gyrus (BA 37)	−52	−46	−6	248
		Left occipital pole ( BA 18)	−2	−96	10	216
		Left frontal pole (BA 46)	−52	40	12	194
Typical readers	Positive correlations	Right and left frontal pole (BA 13)	34	12	−14	28698
		Right and left precuneus (BA 30)	−8	−56	8	594
		Left postcentral gyrus (BA 5)	−28	−46	68	390
		Right planum temporale (central opercular cortex, Heschl’s gyri, BA 22)	48	−14	−4	356
		Left occipital fusiform gyrus (BA 19)	−50	−68	−14	296
		Left occipital pole (BA 18)	−14	−100	4	117
		Right cerebellum	46	−62	−46	274
	Negative correlations	Left middle temporal gyrus (BA 40)	−44	−32	28	1874
		Right thalamus and angular gyrus (BA 22)	36	−50	12	1631
		Right cerebellum	−2	−78	−26	927
		Left inferior frontal gyrus (BA 46)	−52	40	2	738
		Right parietal lobe (precuneus, BA 39)	38	−70	34	544
		Left thalamus	30	−30	42	488
		Right parietal lobe (precuneus, BA 7)	10	−62	52	476
		Brain stem	6	−36	−44	465
		Right middle frontal gyrus (BA 6)	30	8	48	304
		Left temporal pole	−30	10	−48	180
		Right temporal pole (BA 38)	46	26	−30	172
		Left precentral gyrus (BA 6)	−62	−6	38	168
		Brain stem	−10	−34	−38	162
		Right temporal pole (parahippocampal gyrus, BA 38)	18	4	−36	162
		Right inferior frontal gyrus (BA 46)	48	18	22	160
		Left cerebellum	−8	−68	−10	150
		Right cerebellum	14	−54	−16	142
TR>RD		Left occipital pole (cuneus, BA 18)	−8	−82	14	570
RD>TR		Right superior frontal gyrus (BA 6)	22	6	64	196

BA: Brodmann areas

Figure 2. Regions with connectivity to the salience network that is correlated with the screen/reading time ratio: contrasting the RD vs TR groups.

Contrasting the results of whole-brain seed-to-voxel analysis using the salience network as a seed where connectivity is correlated with screen/reading time ratio for typical readers vs. children with RD. A. A cluster with correlation greater for TRs vs. RD is noted by hot colors and a cluster with correlation greater for RD vs. TRs is noted by cool colors (−5.00 < T < 5.98). B. A glass brain outlines the clusters in several brain orientations; positive correlation is noted in red (TR>RD), negative in blue (RD>TR). C. Effect sizes for the positive (upper scale) and negative (lower scale) correlations.

Discussion

Reading is critical for academic success and everyday communication. According to previous reports, screen exposure may “compete” for neural circuits related to reading[11]. We sought to determine the relationship between screen vs. reading time ratio and functional connections with the neural network related to switching attention and mental resources, i.e., the salience network, in children with RD vs. TRs. Although both groups had more screen time compared to reading time, children with RD did not demonstrate higher screen vs reading time ratio than TRs, which was in contradiction of what we hypothesized. In line with our expectation, however, the screen vs. reading time ratio was associated with reduced reading and EF abilities across both groups. Neurobiologically, children with RD showed increased correlation between screen vs. reading time ratio and functional connections between the salience network and regions associated with Frontal-EF compared to TRs. TRs on the other hand, demonstrated increased correlation between screen to reading time ratio and functional connections between the salience network and visual processing regions specifically in the primary visual cortex, as the screen/reading time ratio increased compared to RDs. The groups did not differ in the correlation between the screen vs reading time ratio and functional connectivity between the salience network and regions within the associative visual processing network. Our results suggest that greater screen vs. reading time ratio may be related to different neurobiological pathways in children with RD and TRs.

Screen time and EF

Several studies have highlighted the allocation of attention resources during screen exposure[6, 13]. Recent studies have suggested that greater screen exposure is related to increased frequencies in the theta/beta ratio in 4–6 year old children[6] and increased negative functional connections between word form areas and frontal regions related to EF[11]. Altered functional connections in language and visual processing regions during a language processing task[39] and decreased fractional anisotropy in white matter tracts related to EF in 3–5 years old children[7] were also observed. Our findings add to this literature and suggest that the increased screen time relative to reading time may absorb attention abilities, speed of processing resources, the ability to switch between activities and memory abilities (as found in the current study). As EF mature relatively later in life, around adolescence[40], this negative association between screen time and EF may raise a concern, especially in light of the negative association between screen time and reading ability we observed. A future longitudinal study assessing the direct effect of screen exposure on EF and associated neural networks in these two groups of readers is warranted.

Recruitment of salience and left primary visual cortex regions in relation to screen/reading ratio in TRs vs. children with RD

Our study points at increased functional connections between the salience network and the left occipital regions specifically the primary visual cortex in TRs vs. children with RD in relation to the screen/reading time ratio. These findings may suggest, in TRs, greater synchronization between changes in attention resources from internal to external stimuli (the role of the salience network) and the activation of the region related to visual processing associated with movement (primary visual cortex) with increased screen/reading ratio. These results extend and strengthen our previous results suggesting “competing” relations between screen exposure with regions and brain networks related to reading (such as the salience network[15, 41] and the left associative cortex in the occipital lobe related to word recognition [11, 42, 43]).

Another possible explanation for the increased correlations between screen vs reading time ratio and functional connectivity between the salience network and left primary visual cortex in the TR vs the RD group, may be a decreased responsiveness of the primary visual cortex in relation to screen time in children with TR. Previous findings suggest that in children who demonstrate over-responsiveness to sensory stimulation (especially visual), an increased functional connection between the salience and visual cortex is involved[44]. it may also be that in TRs, the excessive exposure to screens vs. reading materials results in decreased sensitivity, even desensitization, towards visual stimulation related to moving stimuli as in the case of digital media[44]. It would be intriguing to administer a sensory stimulation assessment (also described in Green et al [44]) to define the level of stimulation sensitivity in TRs and children with RDs in several senses in relation to screen exposure time.

Interestingly, the TR cohort in the current study showed an increased number of books in the household vs. children with RD, which may indicate increased exposure to literacy at home[45]. Alternatively, it may be that when children read well, such as in the case of TRs, they tend to collect more books. As increased literacy exposure was previously related to increased left occipital activation[46], the effect of screen time may be even more devastating in children with RDs with a low number of books in the household (such as in children with a low socioeconomic status). As the current study did not assess the number of children’s books vs an overall number of books in the household, an additional study should examine if these differences between TRs and RDs in the number of books exist also in the number of children’s books.

Increased recruitment of the salience and frontal regions in RDs

Children with RD showed an increased association between the connectivity of the salience network and right frontal cortices and the screen/reading time ratio compared to TRs. Previous studies suggest that the greater dissociation between the salience and fronto-parietal networks (the right superior frontal gyrus found in the current study is within the borders of this network) are related to better performance[14]. It seems that the greater screen/reading time ratio in children with RD is related to the opposite outcome: increased connectivity between these networks. This increase is also related to negative outcomes (i.e., low EF and low reading ability).

Another explanation for the positive correlation between connectivity of the salience network and right frontal regions and screen/reading time ratio may be that children with RD use EF regions to compensate for their reading challenges[47–50]. We assume that increasing screen time may be even more devastating for them, as it competes even with the compensatory mechanisms, i.e., with EF, hence, impairing reading even more. Future longitudinal research should look at the long-term effects on reading proficiency in these readers.

Study limitations

The following limitations should be taken into account when examining the results of the current study. First, we did not differentiate between the types of screen to which the children were exposed. Using social media is not ideal but does involve reading and may be related to reading exposure of a different type. Second, we did not assess these networks during an actual reading task. It would be interesting to examine the functional connections of the salience network in relation to the screen/reading ratio when an actual reading task is used. Third, as we noted above, we did not track these children longitudinally and we do not know how overall exposure starting from a young age affected their functional connections. A longitudinal study directly assessing these outcomes is warranted.

Conclusions

This is the first study to assess the relations between screen vs. reading time exposure and neural circuits related to EF in children with RD and primary visual processing regions in TRs (see supplemental Figure 3 adapted from[12] for the regions found in the current study and their related networks). Our findings of the different neural circuits engaged in TRs vs children with RD highlight the neurobiological differences between the groups, and strengthen the relations between EF and reading abilities previously suggested. It also suggests that technology, as fascinating and engaging it is, should be used cautiously, especially with children.

Summary

The goal of the current study was to determine the association between the ratio of screen time duration to reading time duration and functional connectivity of the cognitive control network (i.e. the salience network) to the entire brain in children with RD compared to typical readers (TRs). Significantly higher screen vs. reading time was observed in children regardless of whether they have RD or are TRs. Interestingly we found that increased screen to reading time ratio was related to reduced reading and executive function abilities. This Increased screen to reading time ratio was related to increased functional connectivity between the salience network and frontal regions part of the fronto-parietal network in children with RD. These findings may suggest a “competition” of screen time over reading time in children with RD over neural circuits related to reading. As executive functions are important for child development as well as academic abilities, a careful attention of monitoring screen time and content in children in general and of those with RD is needed.