CoChat
Sign Up
Cited by 132Cognitive Abilities and TestingCognitive and developmental aspects of mathematical skillsDevelopmental and Educational Neuropsychology

Working memory training in typically developing children: A multilevel meta-analysis

Open access
Sala, Giovanni, Gobet, Fernand·2020·Psychonomic Bulletin & Review
DOI 10.3758/s13423-019-01681-y
Open this paper in CoChat — free accountOpen PDF

Abstract

Working memory (WM) training in typically developing (TD) children aims to enhance not only performance in memory tasks but also other domain-general cognitive skills, such as fluid intelligence. These benefits are then believed to positively affect academic achievement. Despite the numerous studies carried out, researchers still disagree over the real benefits of WM training. With this meta-analysis (m = 41, k = 393, N = 2,375), we intended to resolve the discrepancies by focusing on the potential sources of within-study and between-study true heterogeneity. Small to medium effects were observed in memory tasks (i.e., near transfer). The size of these effects was proportional to the similarity between the training task and the outcome measure. By contrast, far-transfer measures of cognitive ability (e.g., intelligence) and academic achievement (mathematics and language ability) were essentially unaffected by the training programs, especially when the studies implemented active controls (g¯\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \overline{g} $$\end{document} = 0.001, SE = 0.055, p = .982, τ2 = 0.000). Crucially, all the models exhibited a null or low amount of true heterogeneity, which was wholly explained by the type of controls (nonactive vs. active) and by statistical artifacts, in contrast to the claim that this field has produced mixed results. Since the empirical evidence shows the absence of both generalized effects and true heterogeneity, we conclude that there is no reason to keep investing resources in WM training research with TD children.

AI snapshot

Written from the paper’s full text.

TL;DR

This multilevel meta-analysis of 41 studies (393 effect sizes, N = 2,375) finds that working memory training in typically developing children produces near-transfer memory benefits but no reliable far-transfer effects on intelligence or academic achievement. With active controls, far-transfer effects are essentially zero, and true heterogeneity is null or low.

Key findings

  • Near-transfer effects were small to medium: overall \( \bar{g} = 0.389 \), SE = 0.056, p < .001; nearest-transfer effects were larger (\( \bar{g} = 0.468 \)) than less-near-transfer effects (\( \bar{g} = 0.261 \)).
  • Far-transfer effects were small overall (\( \bar{g} = 0.092 \), SE = 0.033, p = .015), but became null with active controls (\( \bar{g} = 0.001 \), SE = 0.055, p = .982, \( \tau^2 = 0.000 \)).
  • Far-transfer follow-up effects were null (\( \bar{g} = 0.006 \), SE = 0.022, p = .809), while near-transfer follow-up effects were \( \bar{g} = 0.239 \), SE = 0.103, p = .059.
  • True heterogeneity was null or low across models; when present, it was explained by control type, baseline differences, transfer distance, and a few influential cases, not by age or training-task type.

Methods

  • Systematic search of MEDLINE, PsycINFO, Science Direct, and ProQuest Dissertation & Theses, plus reference lists, retrieved 3,080 records; 41 studies from 2007–2019 met inclusion criteria.
  • Included typically developing children (maximum mean age = 16 years), a WM training condition, at least one control group, objective cognitive/academic tasks, pre-post design, non-self-selected participants, and sufficient data for effect sizes.
  • Used robust variance estimation with hierarchical weights to model nested effect sizes and estimate within-cluster (\( \omega^2 \)) and between-cluster (\( \tau^2 \)) true heterogeneity; moderators included baseline difference, control group, age, training-task type, outcome measure, and near-transfer type.

Limitations

  • The paper states that most primary studies examined preschool and primary school children, with only a fraction including adolescents, making findings less generalizable to typically middle/high school students (e.g., 12–16 years).
  • The paper states that follow-up models included limited numbers of studies, so no further analysis was conducted for those models.
  • Evident from the text: effect-size calculations assumed a pre-post-test correlation of \( r_{xx} = r = .700 \) because primary studies rarely reported these values, though the authors replicated analyses using correlations from .500 to .800 and observed no significant differences.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.3758/s13423-019-01681-y

Cites

60 of 60 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

29 of 50 cited sources checked · 14 supported · 3 flagged · accuracy not yet measured

  • Premisecitation

    Multiple meta-analyses have reported that true heterogeneity in working memory training, both within and between studies, is null or low.

    “Interestingly, several meta-analyses have estimated within-and between-study true heterogeneity in WM training to be null or low, both with near-transfer and far-transfer effects.”
    Quote in the paperSubstantial claim2 of 3 sources checked
    • cites Sala et al., 2019b source
    • cites Aksayli et al., 2019 source
    • cites Melby-Lervåg et al., 2016 source

      The Results section, “Effects of Working Memory Training on Near Transfer Measures” (page unavailable), explicitly says near-transfer findings vary across studies and that true heterogeneity is present. The Results also report true heterogeneity between studies for intermediate-transfer measures. Thus, this paper contradicts the claim that heterogeneity is null or low across transfer outcomes; a near-zero far-transfer effect is not the same as low heterogeneity. (LLM, conf=0.88)

      Checked by a reasoning model — may be wrong.

      Experimental — accuracy not yet measured
  • Interpretationreasoning

    Positive far-transfer effects from working memory training are unlikely to occur in specific subgroups, such as baseline underachievers.

    “Moreover, it is unlikely that WM training exerts positive far-transfer effects on subgroups of individuals (e.g., underachievers at baseline assessment; Jaeggi et al., 2011).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Jaeggi et al., 2011 source

      The Results (page unavailable) report positive far transfer to fluid intelligence in the subgroup of children with large working-memory training gains, contrary to the claim that positive effects are unlikely in subgroups. The Discussion (page unavailable) finds no significant difference in transfer between children with high and low initial fluid-intelligence performance; it does not identify baseline underachievers as unlikely to benefit. (LLM, conf=0.88)

      Checked by a reasoning model — may be wrong.

      Experimental — accuracy not yet measured
  • Interpretationcitation

    Contrary to earlier hypotheses, working memory and fluid intelligence do not rely on the same neural mechanisms.

    “Second and most important, our findings support recent empirical evidence showing that WM and fluid intelligence do not share the same neural mechanisms as previously hypothesized (e.g., Halford et al., 2007;Jaeggi et al., 2008;Strobach & Karbach, 2016;Taatgen, 2013Taatgen, , 2016)).”
    Quote in the paperSubstantial claim1 of 3 sources checked
    • cites Halford et al., 2007 source
    • cites Jaeggi et al., 2008 source

      The Abstract states the opposite: working memory and fluid intelligence seem to rely on similar neural networks, particularly in lateral prefrontal and parietal cortices. Its qualification that they are not identical constructs does not establish that they rely on different neural mechanisms. (LLM, conf=0.90)

      Checked by a reasoning model — may be wrong.

      Experimental — accuracy not yet measured
    • cites Strobach & Karbach, 2016 source
  • Premisecitation

    Prior meta-analyses found that working memory training has a medium effect on near-transfer memory tasks.

    “Both meta-analyses reach the conclusion that while WM training exerts a medium effect on memory-task performance (near transfer), no other cognitive or academic skills (far transfer) seem to be affected, regardless of the population examined; in particular, no effects are observed when active controls are implemented to rule out placebo effects”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites Melby-Lervåg et al., 2016 source

      The Results section reports significant, moderate effects on verbal and visuo-spatial working-memory measures, supporting the claim that training benefits memory-task performance. But it classifies those measures as intermediate transfer; its separate near-transfer section describes large effects on tasks similar or identical to those trained. The claimed effect size therefore does not match the paper’s near-transfer category. Page numbers were not available in our copy. (LLM, conf=0.88)

      Experimental — accuracy not yet measured
    • cites Sala et al., 2019a source
  • Premisecitation

    Sala and Gobet (2017) found a modest far-transfer effect (g=0.12) in typically developing children.

    “With TD children (aged 3 to 16), Sala and Gobet (2017) found a medium effect (𝑔̅ = 0.46) with near transfer and a modest effect with far transfer (𝑔̅ = 0.12), with the qualification that the better the quality of the design (use of an active control group), the smaller the effect sizes.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Sala & Gobet, 2017 source
  • May combine two claims. 32
  • Classified as probably not a substantial claim. 14
  • The quote only partly states this claim. 5

  • May combine two claims.

    Premisecitation

    Prior meta-analyses found no far-transfer effects on other cognitive or academic skills, especially when active controls are used.

    “Both meta-analyses reach the conclusion that while WM training exerts a medium effect on memory-task performance (near transfer), no other cognitive or academic skills (far transfer) seem to be affected, regardless of the population examined; in particular, no effects are observed when active controls are implemented to rule out placebo effects”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites Melby-Lervåg et al., 2016 source

      The Results (Immediate Effects of Working Memory Training on Far Transfer Measures; page unavailable) find near-zero, nonsignificant effects for several cognitive and academic outcomes with treated controls, supporting the main far-transfer conclusion. But they also report a small, significant reading-comprehension effect before excluding problematic comparisons, so “no effects” is too absolute. The Results (Effects of Working Memory Training on Near Transfer Measures; page unavailable) describe near-transfer effects as large, rather than medium. (LLM, conf=0.86)

      Experimental — accuracy not yet measured
    • cites Sala et al., 2019a source

  • May combine two claims.

    Premisecitation

    When true heterogeneity is present, it is explained by control group type, statistical artifacts such as regression to the mean, and a few extreme effect sizes.

    “When present at all, true heterogeneity is accounted for by the type of control group used (active or non-active), statistical artifacts such as pre-post-test regression to the mean due to baseline differences between the experimental and control groups, and, to a lesser extent, a few extreme effect sizes.”
    Quote in the paperSubstantial claim2 of 3 sources checked
    • cites Sala et al., 2019b source
    • cites Aksayli et al., 2019 source

      The Results (section 4.1.1) attribute observed far-transfer heterogeneity to regression to the mean, and sections 3.7 and 4.2.2 describe excluding extreme effect sizes. But section 3.4 only identifies active versus non-active control groups as a moderator considered; the reported near-transfer heterogeneity in section 4.2.2 is mainly explained by baseline differences and the type of memory task, not control-group type. The claim therefore overstates what the paper says explains heterogeneity. (LLM, conf=0.85)

      Experimental — accuracy not yet measured
    • cites Melby-Lervåg et al., 2016 source

      The Results (Immediate Effects on Far Transfer Measures; page unavailable) report differences by control-group type and show that unusually large effects or problematic pretest-to-posttest changes affect some findings. The paper does not establish that regression to the mean from baseline group differences explains true heterogeneity. Its Results on near and intermediate transfer also identify other moderators, so the claim that the listed factors account for heterogeneity whenever it occurs is too broad. (LLM, conf=0.82)

      Experimental — accuracy not yet measured

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Working memory capacity grows from infancy through childhood into adolescence, driven by maturation and knowledge gains.

    “WM capacitythat is, the maximum amount of information that WM can store and manipulate -steadily increases throughout infancy and childhood up to adolescence (Cowan, 2016;Gathercole, Pickering, Ambridge, & Wearing, 2004), both due to maturation and increase in knowledge (Cowan, 2016;Jones, Gobet, & Pine, 2007).”
    Quote in the paperNot a substantial claim2 of 3 sources checked
    • cites Cowan, 2016 source

      In “Development of Working Memory Capacity?” (page unavailable), Cowan reports steady improvement in working-memory test performance from ages 4 to 15; the infant-studies section also describes apparent gains during infancy. But “The Working Memory Capacity Growth Hypothesis” says that a simple increase in capacity across ages remains unproven, while “The Factor of Increasing Knowledge” treats knowledge as a contributor to performance rather than establishing that it increases basic capacity. The claim overstates a steady capacity increase from infancy through adolescence and its proposed causes. (LLM, conf=0.86)

      Experimental — accuracy not yet measured
    • cites Gathercole et al., 2004 source
    • cites Jones et al., 2007 source

      In “Interaction of phonological working memory with LTM knowledge,” the paper explains that accumulated knowledge can increase how much information children hold in working memory. It does not establish a steady increase from infancy through adolescence or show that maturation drives it; instead, it says developmental changes in capacity may not be necessary to explain the performance changes studied. (LLM, conf=0.87)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Interpretationcitation

    The degree of transfer depends on how similar the training task is to the target task, not on the specific features of the training program.

    “This outcome is in line with the findings ofMelby-Lervåg et al. (2016) and corroborate the idea that transfer is function of distance between training task and target task rather than the features of the training program per se (e.g.,Byrne et al., 2019;Pergher et al., 2019).”
    Quote in the paperSubstantial claim2 of 3 sources checked
    • cites Melby-Lervåg et al., 2016 source

      The Method defines near transfer by similarity to the trained task, and the Results report large effects on similar or identical tasks. The Discussion finds virtually no evidence that training type or dose moderates far transfer. But it also reports moderators of near and intermediate transfer, so it does not establish that transfer depends on task similarity rather than training-program features in general. The supplied passages have section labels but no page numbers. (LLM, conf=0.83)

      Experimental — accuracy not yet measured
    • cites Byrne et al., 2019 source

      The Discussion (pp. 12–13) examines overlap and differences between backward-recall and n-back tasks, not transfer after training. A search for training transfer found references to other studies but no passage in this paper establishing that transfer depends on task similarity rather than features of the training program. (LLM, conf=0.76)

      Experimental — accuracy not yet measured
    • cites Pergher et al., 2019

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Premisecitation

    This pattern of results also appears in adults, older adults, and children with learning disabilities.

    “Furthermore, as the same pattern of results has been found in adults, older adults, and children with learning disabilities (Aksayli et al., 2019;Melby-Lervåg et al., 2016;Sala et al., 2019a), the most parsimonious and plausible conclusion is that WM training does not lead to far transfer.”
    Quote in the paperSubstantial claim3 of 3 sources checked
    • cites Aksayli et al., 2019 source
    • cites Melby-Lervåg et al., 2016 source
    • cites Sala et al., 2019a source

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationcitation

    The convergence of experimental, correlational, and neural evidence is a successful case of triangulation.

    “Such convergent findings at different levels of empirical evidence (experimental, correlational, and neural) is a successful example of triangulation that does not leave much room for further debate (Campbell & Fiske, 1959;Munafò & Smith, 2018).”
    Quote in the paperNot a substantial claim1 of 2 sources checked
    • cites Campbell & Fiske, 1959 source
    • cites Munafò & Smith, 2018 source

      In “Repeating experiments is not enough” (page unavailable), the paper defines triangulation as using multiple approaches to one question and says agreement across methodologies makes artefacts less likely. It does not assess the citing sentence’s particular experimental, correlational, and neural findings, or say that their convergence leaves little room for further debate. (LLM, conf=0.84)

      Experimental — accuracy not yet measured

  • The quote only partly states this claim.

    Findingown data

    After excluding the influential effect and merging dependent effects, the between-study true heterogeneity for less-near transfer became non-significant (τ2 = 0.010, p = .234).

    “Excluding the influential effect and merging the statistically dependent effects lowered the between-study true heterogeneity to a non-significant amount (τ 2 = 0.010, p = .234).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Near-transfer immediate post-test overall effect was medium and significant (g=0.389, p<.001).

    “The overall effect size was 𝑔̅ = 0.389, SE = 0.056, 95% CI [0.271; 0.507], m = 29, k = 123, df = 18.8, p < .001,ω 2 = 0.006, τ 2 = 0.059.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    All models exhibited null or low true heterogeneity, explained by control type and statistical artifacts.

    “Crucially, all the models exhibited a null or low amount of true heterogeneity, wholly explained by the type of controls (non-active vs. active) and statistical artifacts, in contrast to the claim that this field has produced mixed results.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Far-transfer immediate post-test overall effect was small and statistically significant (g=0.092, p=.015).

    “The overall effect size of the RVE intercept model was 𝑔̅ = 0.092, SE = 0.033, 95% CI [0.021; 0.163], m = 34, k = 146, df = 14.8, p = .015,ω 2 = 0.000, τ 2 = 0.000.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The near-transfer models exhibit somewhat greater between-study true heterogeneity, which is partly attributable to the type of near transfer (nearest vs. less near).

    “The near-transfer models show slightly higher between-study true heterogeneity, which is partly explained by the type of near transfer (nearest vs. less near).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Control group type and type of training were not significant moderators of near-transfer effects (p=.845 and ps≥.477).

    “The type of control group and Type of training were not significant moderators either (p = .845and ps ≥ .477,respectively).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The meta-analytic models show high internal consistency, with no appreciable within-study true heterogeneity (ω2 ≈ 0.000 in all models).

    “The examination of true heterogeneity reveals that the meta-analytic models exhibit high internal consistency.No appreciable within-study true heterogeneity is observed (ω 2 ≈ 0.000 in all the models).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    With non-active controls, far-transfer immediate post-test effect was small and significant (g=0.139, p=.015).

    “When non-active controls were used, the overall effect size was 𝑔̅ = 0.139, SE = 0.045, 95% CI [0.034; 0.243], m = 21, k = 75, df = 8.2, p = .015,ω 2 = 0.000, τ 2 = 0.005.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    For far transfer, the between-study true heterogeneity is very low and is explained by control group type, baseline differences, and a few influential cases.

    “Regarding far transfer, this heterogeneity is very low and is accounted for by the type of control group, baseline differences, and a few influential cases.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Neither baseline nor age was a significant moderator of near-transfer effects (p=.154 and p=.914).

    “The meta-regression showed that neither Baseline nor Age were a significant moderator (p = .154and p = .914,respectively).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Far-transfer effects are small with non-active controls and null with active controls.

    “The findings were clear: whereas WM training fosters the performance on memory tasks, small (with non-active controls) to null (with active controls) far-transfer effects are observed.”
    • Also stated in: § Conclusions · plain
      “By contrast, small to null effects are found on far-transfer measures (i.e., fluid intelligence, attention, language, and mathematics).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The present meta-analysis included 41 studies, 393 effect sizes, and 2,375 participants.

    “In total, we found 41 studies, conducted from 2007 to 2019, that met all the inclusion criteria (Appendix A in the Supplemental materials).These studies included 393 effect sizes and a total of 2,375 participants.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisenone given

    The distribution of working memory skills in typically developing children has a larger range and is more homogeneous across studies than in learning-disabled children.

    “Moreover, the distribution of WM skills in TD children encompasses a larger range (which reduces biases related to range restriction), and it is more homogeneous across studies.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning

    If training could improve working memory skills, the benefits would extend to many other cognitive and real-life skills.

    “It is thus reasonable to believe that if WM skills could be improved by training, the benefits would spread across many other cognitive and real-life skills.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    Working memory capacity correlates positively with fluid intelligence and attentional processes.

    “WM capacity is positively correlated with essential cognitive functions such as fluid intelligence and attentional processes (Engle, 2018;Kane, Hambrick, & Conway, 2005;Süß, Oberauer, Wittmann, Wilhelm, & Schulze, 2002).”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites Engle, 2018 source
    • cites Kane et al., 2005 source
    • cites Süß et al., 2002

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Interpretationreasoning

    Typically developing children have the greatest likelihood of benefiting from cognitive-training programs, since their cognitive skills are still developing and brain plasticity is at its peak.

    “Finally, and probably most importantly, TD children represent that population in which cognitive skills are still developing, and brain plasticity is at its peak.In other words, TD children are the most likely to benefit from cognitive-training interventions.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Premisenone given

    The population of typically developing children is larger than that of children with learning disabilities.

    “Most obvious, the population of TD children is larger than the population of children with learning disabilities, who suffer from different disorders (e.g., ADHD, dyslexia, and language impairment).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationreasoning

    A null result in typically developing children would cast serious doubt on obtaining generalized effects in other populations.

    “Therefore, a null result would cast serious doubts on the possibility to obtain generalized effects in other populations as well (e.g., healthy adults).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    General cognitive ability strongly predicts academic achievement and job performance.

    “It is widely acknowledged that general cognitive ability is a major predictor of academic achievement and job performance (Detterman, 2014;Gobet, 2016;Schmidt, 2017;Wai, Brown, & Chabris, 2018).”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites Detterman, 2014 source
    • cites Gobet, 2016 source
    • cites Schmidt, 2017 source

  • The quote only partly states this claim.

    Study design and conductown data

    The selection model estimated some publication bias for less-near transfer (g = 0.196).

    “Finally, the selection model estimated some publication bias (𝑔̅ = 0.196).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductown data

    Inter-rater agreement for control group classification was 98%.

    “The inter-rater agreement was 98%; here and elsewhere, the two raters resolved every discrepancy by discussion;”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data

    Due to the limited number of studies in the follow-up model, no further analysis was conducted.

    “Due to the limited number of studies included in this model, no further analysis was conducted.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductown data

    The majority of primary studies examined working memory training in preschool and primary school typically developing children, with only a minority including adolescents, which limits generalizability to middle/high school students (12-16 years).

    “That being said, it is worth noting that most of the primary studies have investigated the effects of WM training in preschool and primary school TD children (see Descriptive Statistics section).Only a fraction of the primary studies include adolescent samples, which makes our findings somewhat less generalizable with typically middle/high school students (e.g., 12-16 years of age).”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S1 (asserted), S3 (load-bearing), S4 (based). The quote only partly states this claim.

    Study design and conductown data

    No missing study was retrieved by the trim-and-fill L0 or R0 estimator for less-near transfer.

    “No missing study was retrieved with either the L0 or R0 estimator.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Interpretationreasoning

    Were the heterogeneity not a statistical artifact, between-study true heterogeneity would still appear after averaging effect sizes within each study.

    “Otherwise, between-study true heterogeneity would occur even after averaging the effect sizes within the same study.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductown data

    Cohen's kappa for type of training task classification was 1.00.

    “Cohen's kappa was κ = 1.00;”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductnone given

    The two authors independently coded each effect size and reached 100% agreement.

    “The two authors coded each effect size independently and reached 100% agreement.”
    Quote in the paperNot a substantial claim

  • The quote only partly states this claim.

    Interpretationown data

    The follow-up overall effects are consistent with the pattern of small-to-null far-transfer effects.

    “The follow-up overall effects are consistent with this pattern of results.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductown data

    Inter-rater agreement for type of near transfer classification was 97%.

    “The inter-rater agreement was 97%.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductown data

    Cohen's kappa for outcome measure classification was .98.

    “Cohen's kappa was κ = .98;”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    The collected evidence indicates no generalized effects and no true heterogeneity.

    “Since the empirical evidence shows an absence of generalized effects and true heterogeneity, we conclude that there is no reason to keep investing resources in WM training research with TD children.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    The memory tests measuring near transfer in the primary studies are intercorrelated and can be averaged within each study to obtain more precise measures.

    “In addition, this result suggests that the memory tests (near transfer) used in the primary studies are correlated to each other and can be averaged by study to get more precise measures.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Neuroimaging findings indicate that working memory performance correlates with greater network segregation, while fluid intelligence tasks show the reverse pattern.

    “Brain imaging data suggest that WM performance is associated with increased network segregation, whereas the opposite pattern occurs when participants are asked to solve fluid intelligence tasks (Lebedev, Nilsson, & Lövdén, 2018).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Lebedev, Nilsson, & Lövdén, 2018 source

  • May combine two claims.

    Interpretationreasoning

    Working memory and fluid intelligence appear neither isomorphic nor causally linked.

    “Thus, WM and fluid intelligence do not appear isomorphic or even causally related, which would explain why WM training fails to induce any far-transfer effect despite the wellknown correlation between measures of WM capacity, fluid intelligence, and academic achievement.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    This pattern-transfer depending on training-outcome similarity-has been replicated across many domains and seems to be a constant in human cognition.

    “This pattern of results has been replicated in many different domains and appears to be a constant in human cognition (for a review, see Sala & Gobet, 2019).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Sala & Gobet, 2019

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Interpretationreasoning

    Nearest transfer effects are not strong evidence of cognitive enhancement; they reflect children's improved ability to perform the training task or a similar variant.

    “First, nearest transfer effects do not constitute any robust evidence for cognitive enhancement.Rather, they are clearly a measure of children's boosted ability to perform the training task or one of its variants.”
    Quote in the paperSubstantial claim

  • May combine two claims. The quote only partly states this claim.

    Interpretationreasoning

    Neither theory nor empirical evidence suggests that working memory training harms cognition or has negative effects on children's cognitive skills or academic achievement.

    “However, there is no theoretical reason nor any empirical evidence to believe that WM training exerts a detrimental effect on one's cognition.”
    • Also stated in: § Practical and Theoretical Implications · hedged
      “Although there seems to be no reason to believe that WM training negatively affects children's cognitive skills or academic achievement, implementing such programs would represent a waste of financial and time resources.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Certain less-near-transfer memory tasks, such as the odd-one-out task, are not included in the training programs yet still overlap with some training tasks, like simple-span tasks.

    “As noted by Shipstead, Redick, and Engle (2012), even though some less-near-transfer memory tasks (e.g., odd-one-out task) are not part of the training programs, they still share some overlap with some training tasks (e.g., simple-span tasks).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Shipstead, Redick, and Engle, 2012 source

  • May combine two claims.

    Interpretationreasoning

    Participants in working memory training probably become somewhat better at certain memory tasks than controls, which accounts for the small less-near-transfer effects and the lack of far transfer.

    “Rather, they most likely acquire the ability to perform some memory tasks somewhat better than controls, which explains the small effect sizes reported in less-near-transfer measures and the absence of far transfer.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    This absence of isomorphism or causal relation would account for why working memory training produces no far transfer even though working memory capacity, fluid intelligence, and academic achievement are correlated.

    “Thus, WM and fluid intelligence do not appear isomorphic or even causally related, which would explain why WM training fails to induce any far-transfer effect despite the wellknown correlation between measures of WM capacity, fluid intelligence, and academic achievement.”
    Quote in the paperSubstantial claim
46 of 46

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Meta-analytic conventions have consistently classified backward and complex span within the span domain rather than with updating paradigms.

    “meta-analytic conventions have consistently classified them within the span domain rather than with updating paradigms (Sala & Gobet, 2020)”

    From The asymmetry of working memory training transfer: A systematic review and meta-analysis (2026) · cites this paper as Sala & Gobet, 2020

    The Moderators section (page unavailable) classifies reverse digit span and odd-one-out tasks as span tasks, separately from updating tasks such as n-back and running tasks. This supports the distinction for the examples given, but the paper does not establish that meta-analyses have consistently classified backward and complex span this way. (LLM, conf=0.82)

    Experimental — accuracy not yet measured

  • Working memory training does not enhance children's domain-general cognitive skills or academic achievement.

    “WM training does not enhance children’s domain-general cognitive skills or academic achievement (Aksayli, Sala, & Gobet, 2019; Melby-Lervåg et al., 2016; Sala & Gobet, 2020).”

    From Cognitive and academic benefits of music training with children: A multilevel meta-analysis (2020) · cites this paper as Sala & Gobet, 2020

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Within-study heterogeneity is not approximately zero in every model

    Minor

    The discussion says ω2 is approximately 0.000 in all models, but the nearest-transfer model reports ω2 = 0.011.

  • Moderator count does not match the numbered list

    Minor

    The methods introduce four potential moderators, but the numbered list contains six.

2 of 2

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Related

Authors
Sala, G., & Gobet, F.
Affiliations
  • Fujita Health University (JP)
  • London School of Economics and Political Science (GB)
Corresponding author
Sala, Giovanni
ORCID
Venue
Psychonomic Bulletin & Review
Publisher
Springer Science and Business Media LLC
Published
2020-01-14
Volume
27
Issue
3
Pages
423-434
ISSN
  • 1069-9384 (ISSN-L)
  • 1531-5320
DOI
10.3758/s13423-019-01681-y
OpenAlex
W2963815297
Semantic Scholar
b4b4c8bd24401110fbad4e671be8cb1ef29984f3
Cited by
131
References
58
Citations by year
2026: 19 · 2025: 17 · 2024: 31 · 2023: 19 · 2022: 18 · 2021: 17 · 2020: 9
FWCI
14.56
Type
journal article
Language
en
Open access
Licence
Locations
  • Psychonomic Bulletin & Review (publishedVersion)
  • PubMed (publishedVersion)
  • London School of Economics and Political Science Research Online (London School of Economics and Political Science) (submittedVersion)
  • London School of Economics and Political Science Theses Online (London School of Economics and Political Science) (submittedVersion)
Funders
Japan Society for the Promotion of Science (17F17313)
Topics
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Cognitive Abilities and Testing
  • Physical Sciences › Mathematics › Statistics and Probability › Cognitive and developmental aspects of mathematical skills
  • Social Sciences › Psychology › Neuropsychology and Physiological Psychology › Developmental and Educational Neuropsychology
Keywords
working memory training, typically developing children, near transfer, far transfer, fluid intelligence, academic performance, active control, statistical heterogeneity
MeSH terms
Academic Success, Child, Child Development, Cognition, Humans, Intelligence, Mathematics, Memory, Short-Term, Research Design

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-08.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-08.

Cited sources are checked by gpt-6-sol.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review