Action video games normalise the phonemic awareness in pre-readers at risk for developmental dyslexia

The study investigates whether action video game (AVG) play can remediate early language-based predictors of reading in pre-readers at risk for developmental dyslexia (DD). Prior work shows AVGs enhance attentional control and perceptual learning, with far-transfer to reading and language skills in some populations. DD is linked to phonological processing deficits and potentially to sluggish attentional shifting and inefficient accumulation of sensory evidence. The authors hypothesized that AVG training, by engaging reward systems and attentional control and accelerating multisensory evidence accumulation, would improve phonemic awareness (phoneme discrimination), phonological working memory, and rapid automatized naming (RAN) in at-risk pre-readers, with the strongest effects expected for phonemic awareness. Demonstrating such effects at a pre-reading stage could inform preventive interventions that reduce future risk of DD.

The paper reviews evidence that AVGs induce neuroplasticity in visual attention (spatial, temporal, object-based) and executive functions, with transfer to reading and academic outcomes. Cross-sectional and longitudinal studies indicate video gaming is associated with faster inhibition, working memory, and improved intelligence/reading development after controlling for genetics and SES. AVGs may enhance auditory noise exclusion and accelerate auditory attentional shifting, linked to phonological memory and reading. Meta-analyses report AVG training benefits for visual attention (g≈0.72), reading speed (g≈0.44), and phonological processing (g≈0.45). In DD, deficits in attentional control and evidence accumulation have been proposed alongside phonological deficits, implicating right-lateralized multimodal networks (e.g., right inferior parietal ‘when’ pathway, ventral fronto-parietal attention, alerting systems, right anterior insula). Prior intervention studies show AVG-related improvements in reading fluency and phonological working memory in children with DD, though findings on RAN and broader phonological outcomes are mixed. These literatures motivate testing whether AVG training produces far transfer to phonological predictors in pre-readers at risk for DD.

Design: Prevention study with pre- and post-training assessments and a 6-month follow-up for the AVG group. At-risk pre-readers were randomized with unequal allocation to four groups: AVG, Serious Non-Action Video Game (SNAVG), Speech therapy (SPEECH, treatment-as-usual), or Waiting list (WAIT). Not-at-risk peers served as a comparison group to compute z-scores and evaluate catch-up. Participants: 120 Italian pre-readers in the last year of kindergarten (57 females, 63 males; mean age = 5.58 years, SD = 0.43), typical IQ (WPPSI Block Design mean = 12.08, SD = 3.47), no neurological/sensory/psychiatric disorders. Risk classification: at-risk if below −1.00 SD in at least one of: phonemic awareness, phonological working memory, or RAN; 79 at-risk and 41 not-at-risk. Group allocation (at-risk): AVG n=43; SNAVG n=11; SPEECH n=11; WAIT n=14. Not-at-risk children (n=41) completed two assessments over a similar interval (1.5–2.5 months) to control for spontaneous development and test–retest effects. Blinding: Different experimenters administered training vs assessments. Assessors were blind to group assignments. Assessments and outcome measures: - Phonemic awareness: Phonemic Discrimination (PD) task with 15 bisyllabic pseudoword pairs differing by one phoneme; accuracy used as the outcome. - Phonological working memory: Pseudoword Repetition (PWR) with 40 pseudowords varying in length/complexity; accuracy used. - Rapid Automatized Naming (RAN): serial naming of colored circles; speed (seconds) used (accuracy at ceiling). Timeline: T0 (pre), T1 (post). AVG participants also had T2 (6-month follow-up). Interventions: - AVG: Commercial action game "Space Invaders Extreme 2" on Nintendo DS (high-speed, peripheral stimuli, spatiotemporal unpredictability, adaptive difficulty). Small groups (3 children), supervised by a neuropsychologist. At least 20 sessions of 45 min (mean ≈ 26 sessions, SD ≈ 4.93), four times/week over ~1.5 months. - SNAVG: Set of serious mini-games training grapheme-phoneme mapping, rapid auditory discrimination, and visuospatial attention but without action-game mechanics; adaptive difficulty; small supervised groups. At least 25 sessions of 45 min (mean ≈ 26.5, SD ≈ 0.40), four times/week over ~1.5 months. - SPEECH: Individual speech therapy focusing on phonological awareness (phoneme deletion, counting, blending, syllable segmentation, rhymes) and grapheme–phoneme mapping. At least 21 sessions of 45 min (mean ≈ 29.47, SD ≈ 8.67) over ~3.7 months (typically twice/week). - WAIT: No training during the interval; reassessed after comparable time to active groups. Statistical analysis: For each outcome, z-scores were computed relative to the not-at-risk group. Individual gain Δ(T1−T0) was the primary outcome. ANCOVAs assessed group effects on Δ with T0 as covariate; post hoc comparisons via 1,000 bootstrap resamples. Combined active control (SNAVG+SPEECH) analyses were also conducted. Additional analyses: comparison of AVG T1 PD vs not-at-risk PD to assess catch-up; proportion of AVG participants exceeding the combined control mean gain; follow-up paired tests (T1 vs T2; T0 vs T1/T2). Power analysis (G*Power): repeated measures ANOVA, 4 groups, 2 measurements, medium effect, α=0.05, 80% power → minimum N=48. Baseline checks: Groups did not differ in sex, age, IQ, baseline PD/PWR/RAN, or hours of treatment across active groups.

- Baseline differences (at-risk vs not-at-risk): Significant group effect across reading-related variables at T0 controlling for sex (MANCOVA F(3,115)=51.108, p<0.001, ηp2=0.571). Univariate: PD F(1,117)=44.510, p<0.001, ηp2=0.276; PWR F(1,117)=73.458, p<0.001, ηp2=0.386; RAN F(1,117)=21.766, p<0.001, ηp2=0.157. - Training group equivalence: No differences at T0 across at-risk training groups (MANOVA F(9,219)=1.769, p=0.075); no sociodemographic differences (sex p=0.180; age F(3,75)=0.411, p=0.746); IQ (Block Design) F(3,75)=1.072, p=0.366; hours of treatment F(2,62)=1.878, p=0.161. - Primary outcome (PD gain ΔT1−T0): Significant group effect (F(3,74)=5.175, p=0.003, ηp2=0.173). Mean gains (95% CI): AVG 1.660 (1.248–2.072); WAIT 0.160 (−0.590–0.910); SNAVG 0.645 (0.192–1.482); SPEECH 0.842 (−0.024–1.659). Post hoc (bootstrap, 1,000 resamples): AVG>WAIT mean diff 1.500 (0.391–2.785), p=0.014; AVG>SNAVG diff 1.015 (0.283–1.738), p=0.009; AVG>SPEECH diff 0.818 (0.109–1.608), p=0.039. Paired tests T0 vs T1: Significant PD improvement only in AVG (t(42)=−6.622, mean diff −1.671, 95% CI −2.152/−1.189, p=0.001); not significant in WAIT, SNAVG, SPEECH. - Combined active control (SNAVG+SPEECH, n=22): PD gain effect maintained (ANCOVA F(1,62)=7.987, p=0.006, ηp2=0.114); combined control mean gain 0.631 (0.096–1.162); no significant within-group PD improvement. - Catch-up: At T1, AVG group PD scores did not differ from not-at-risk children (t(82)=1.608, mean diff 0.417, 95% CI −0.077–0.908, p=0.126), indicating normalization. - Clinical relevance: >80% of AVG participants showed PD gain above the mean gain of the combined control group. - Durability: In AVG participants with follow-up (n=33), PD improvement maintained from T1 to T2 (6 months): t(32)=−0.494, mean diff −0.118, 95% CI −0.626–0.294, p=0.626; significant improvements from T0 to T1 (t(32)=−5.889, mean diff −1.706, 95% CI −2.265/−1.109, p=0.001) and T0 to T2 (t(32)=−6.336, mean diff −1.824, 95% CI −2.372/−1.221, p=0.001). - Secondary outcomes (PWR, RAN): No significant group differences in gains; AVG did not outperform controls on phonological working memory or RAN.

The findings support a causal effect of action video game training on phonemic awareness in pre-readers at risk for developmental dyslexia. Despite no explicit phonological content in the AVG, far-transfer occurred selectively for phonemic awareness, not for phonological working memory or RAN. The specificity is consistent with a mechanism whereby AVGs enhance attentional control and accelerate multisensory accumulation of sensory evidence, thereby reducing sluggish attentional shifting and improving phoneme-level processing. The observed normalization relative to not-at-risk peers, high proportion of responders (>80%), and maintenance over 6 months underscore clinical relevance and feasibility of a fun, engaging preventive approach. These results align with prior reports of AVG benefits on auditory-phonological functions and extend them to a pre-reading, at-risk population using both passive and active controls, strengthening claims of specificity and causality. The authors propose that right-lateralized multimodal attention networks (e.g., ventral attention, alerting systems, right insula) may mediate these benefits by tuning phonological representations via improved temporal sampling and evidence accumulation. The lack of effects on PWR and RAN may reflect the greater developmental complexity and different neural substrates of these functions at age ~5, suggesting AVGs primarily modulate networks crucial for phonemic discrimination during this developmental window.

This prevention study demonstrates that a brief action video game intervention can selectively and durably enhance phonemic awareness in pre-readers at risk for developmental dyslexia, normalizing performance relative to not-at-risk peers and benefiting the majority of participants. The results highlight AVGs as a promising, child-friendly tool to target early attentional and multisensory processing mechanisms that support phonological development. Future research should: (1) replicate in larger samples with balanced group sizes; (2) include longer-term follow-ups tracking reading acquisition in primary school; (3) incorporate neuroimaging to elucidate neural mechanisms; (4) compare intervention intensities and dosing across programs; and (5) refine participant selection criteria and evaluate moderators (e.g., cognitive profiles, genetics) to optimize personalized preventive strategies.

- Small sample sizes in the SPEECH, SNAVG, and WAIT groups; although effects persisted when combining active controls, larger samples are needed. - Liberal at-risk inclusion criterion (≤ −1.00 SD in any one task); chosen to mitigate regression to the mean, but may include milder impairments. - Follow-up limited to 6 months post-training and did not assess subsequent reading outcomes in primary school. - No neurobiological measures; proposed right-lateralized attentional/evidence-accumulation mechanisms remain to be tested. - Different intervention intensities/durations across programs (SPEECH longer timeframe), allowing potential spontaneous development effects; future work should equate or systematically vary dose. - Need for longer-term and larger controlled prevention studies to delineate generalizability and impact on reading development.