CoChat
Sign Up
Cited by 727Cognitive Abilities and TestingCognitive and developmental aspects of mathematical skillsCognitive Functions and Memory

The DOI matches its Crossref record; 71 of the 73 references it cites were found in a registry or an index; 6 of the 28 sources its claims cite were checked for support.

Improving fluid intelligence with training on working memory: a meta-analysis

Open access
Au, Jacky, Sheehan, Ellen, Tsai, Nancy, Duncan, Greg J., Buschkuehl, Martin, Jaeggi, Susanne M.·2015·Psychonomic Bulletin & Review
DOI 10.3758/s13423-014-0699-x
Open this paper in CoChat — free accountOpen PDF

Abstract

Working memory (WM), the ability to store and manipulate information for short periods of time, is an important predictor of scholastic aptitude and a critical bottleneck underlying higher-order cognitive processes, including controlled attention and reasoning. Recent interventions targeting WM have suggested plasticity of the WM system by demonstrating improvements in both trained and untrained WM tasks. However, evidence on transfer of improved WM into more general cognitive domains such as fluid intelligence (Gf) has been more equivocal. Therefore, we conducted a meta-analysis focusing on one specific training program, n-back. We searched PubMed and Google Scholar for all n-back training studies with Gf outcome measures, a control group, and healthy participants between 18 and 50 years of age. In total, we included 20 studies in our analyses that met our criteria and found a small but significant positive effect of n-back training on improving Gf. Several factors that moderate this transfer are identified and discussed. We conclude that short-term cognitive training on the order of weeks can result in beneficial effects in important cognitive functions as measured by laboratory tests.

Overview

Integrity signals

Registry

Crossref · 10.3758/s13423-014-0699-x

Cites

71 of 71 references verified0 citation errors · 0 not checkable

Statistics

Verification

Claims the paper makes

6 of 28 cited sources checked · 4 supported · 0 flagged · accuracy not yet measured

  • Premisecitation

    Fluid intelligence and n-back performance rely on overlapping cognitive and neural processes, including shared working memory and interference-resolution demands.

    “On theoretical grounds, the observed improvements are plausible, since Gf and n-back performance draw upon overlapping cognitive and neural processes, including shared demands on WM and interference resolution (Burgess et al., 2011;Buschkuehl, Hernandez-Garcia, Jaeggi, Bernard, & Jonides, 2014;von Bastian & Oberauer, 2013).”
    Quote in the paperSubstantial claim2 of 3 sources checked
    • cites Burgess et al., 2011 source
    • cites Buschkuehl, Hernandez-Garcia, Jaeggi, Bernard, & Jonides, 2014 source

      The Discussion (page unavailable) proposes that n-back and reasoning tasks may engage overlapping brain regions and psychological processes. The paper also identifies working memory as relevant to both in its introductory text, but it does not establish a shared interference-resolution demand; the only retrieved mention of interference control is in a reference title. (LLM, conf=0.86)

      Experimental — accuracy not yet measured
    • cites von Bastian & Oberauer, 2013

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

  • Findingown data

    None of the training parameters examined (session length, number of sessions, starting n-back level, rate of improvement) were significant in the main regression analyses.

    “Although none of our analyzed parameters (session length, number of sessions, starting nback level, and rate of improvement) reached significance in our main regression analyses”
    Quote in the paperSubstantial claim
  • Findingown data

    Studies using passive controls showed significantly greater transfer than studies using active controls.

    “Although our meta-analysis did reveal significantly greater transfer in studies that used passive controls”
    Quote in the paperNot a substantial claim
  • Findingown data

    A fail-safe N analysis indicated that 59 null-effect studies would be needed to reduce the average effect size to .069.

    “Classic fail-safe N calculation: The observed studies, with weighted g = 0.24, led to a combined weighted ES of g = 5.78.Adding the 59 null effect studies would increase the sample size to n = 83.Average ES in the new sample would be: 5:78 83 ¼ :069 .”
    Quote in the paperSubstantial claim
  • Findingown data

    There was no difference in effect size between dual and single n-back training.

    “Additionally, we found no difference depending on whether studies used dual or single n-back, which corroborates previous findings (Jaeggi et al., 2014;Jaeggi et al., 2010).”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites Jaeggi et al., 2014 source
    • cites Jaeggi et al., 2010 source
  • It may state two separate claims as one. 28
  • Probably not a substantial claim. 17
  • The quote only partly states this claim. 2

  • It may state two separate claims as one.

    Premisecitation

    The literature on transfer from working memory training to fluid intelligence is mixed, with some promising studies and rigorous replications failing to find transfer.

    “However, evidence on transfer of improved WM into more general cognitive domains such as fluid intelligence (Gf) has been more equivocal.”
    • Also stated in: § Abstract · plain
      “Despite many promising studies (e.g., Jaeggi, Buschkuehl, Jonides, & Perrig, 2008;Rudebeck, Bor, Ormond, O'Reilly, & Lee, 2012;Stephenson & Halpern, 2013), the aforementioned quote reflects results from other well-controlled, rigorous attempts at replication that have failed to find transfer (Redick et al., 2013;Thompson et al., 2013).”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites Redick et al., 2013 source
    • cites Thompson et al., 2013 source

  • It may state two separate claims as one.

    Premisecitation

    Brain imaging indicates overlapping activity in frontal and parietal regions, suggesting shared neural resources for working memory and fluid intelligence.

    “neuroimaging evidence has demonstrated functional overlap in the lateral prefrontal and parietal cortices, implicating shared neural resources that underlie both constructs (Burgess, Gray, Conway, & Braver, 2011).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Burgess, Gray, Conway, & Braver, 2011 source

  • It may state two separate claims as one.

    Premisecitation

    A dose-dependent relationship between amount of training and transfer has been described and replicated in some but not all studies.

    “Furthermore, we have previously described a dose-dependent relationship between training and transfer, such that more training leads to more gain (Jaeggi et al., 2008).”
    • Also stated in: § Abstract · plain
      “This relationship has been replicated in some studies (Basak, Boot, Voss, & Kramer, 2008;Dahlin, Backman, Neely, & Nyberg, 2009;Stepankova et al., 2014;Tomic & Klauer, 1996), but not others (Redick et al., 2013).”
    Quote in the paperSubstantial claim2 of 3 sources checked
    • cites Jaeggi et al., 2008 source
    • cites Basak, Boot, Voss, & Kramer, 2008 source

      The General Discussion reports that transfer benefits were mostly observed after 23.5 hours of video-game training, while about half that time was not enough. This supports a relationship between training time and transfer in this study, but the paper does not establish that the relationship was replicated in some studies but not others. (LLM, conf=0.86)

      Experimental — accuracy not yet measured
    • cites Redick et al., 2013 source

  • It may state two separate claims as one.

    Findingown data

    Remuneration correlated negatively with fluid intelligence transfer, but the effect became non-significant after outlier trimming (p from .05 to .22), while the slope stayed at -.07.

    “Although we did find a negative correlation, the effect lost significance (from p = .05to p = .22)after trimming outliers (SOM, Table S1).Nevertheless, the slope of the effect (-.07) remained the same”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim.

    Findingown data

    The meta-analysis included 20 studies that met the inclusion criteria.

    “In total, we included 20 studies in our analyses that met our criteria”
    • Also stated in: § Study selection · plain
      “In the end, 20 studies remained in the final meta-analysis (Fig. 2).”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    Publication bias was assessed, and Egger's regression found no relationship between standard error and effect size (p = .68).

    “Egger's regression (Egger, Davey Smith, Schneider, & Minder, 1997), which is based on the association between standard error and ES, revealed no relationship (p = .68).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Egger, Davey Smith, Schneider, & Minder, 1997 source

  • It may state two separate claims as one. The quote only partly states this claim.

    Findingown data

    Passive control groups were mostly from international studies (10/12), whereas active control groups were mostly from U.S. studies (3/12 international).

    “Passive control groups were mostly enrolled in international studies (10/12), as compared with active control groups (3/12)”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    In multiple regression, remuneration and international status individually contributed to the passive-control effect; together they reduced the effect size gain from .31 to .09, making the p-value nonsignificant.

    “Both individually appeared to contribute toward the effect observed in passive controls and, together, reduced the ES gain caused by the condition of having a passive control from .31 to .09,rendering the p-value nonsignificant.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: For the posttest effect size calculation, the authors assumed that treatment and control groups did not differ at pretest.

  • It may state two separate claims as one.

    Findingown data

    The meta-analysis drew on a narrow sample: healthy young adults aged 18-50, mostly college undergraduates, with a mean age of 22.85 years (SD = 2.60).

    “First is that these results are based on a restricted range of the general population: healthy, young adults between 18 and 50 years of age who were largely at the peak of cognitive functioning.Within this restricted range of young adults, the majority of participants were college undergraduates, thereby skewing our samples even younger (mean age ± SD: 22.85 ± 2.60).”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    Heterogeneity was low for the posttest treatment-control effect (I2 = 6.92%, p = .37) and some for the treatment pre-post effect (I2 = 27.88%, p = .08).

    “Some heterogeneity (I 2 = 27.88%,p = .08)was found in Tx ES, which only looked at improvements within the treatment groups, but this heterogeneity was largely controlled out in Post ES (I 2 = 6.92%, p = .37),which directly compares treatment to control groups at posttest.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Findingown data

    The analysis covered 98 fluid intelligence measures, with 559 trained participants and 463 controls.

    “In total, we analyzed data from 98 Gf outcome measures among 559 n-back trained participants and 463 controls.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Findingown data

    With outliers removed, session length and initial n-back level each showed marginal negative associations with posttest effect size (p = .06; b = -.03 and b = -.33).

    “No other predictors reached significance, but session length (p = .06,b = -.03) and starting n-back level (p = .06,b = -.33) both correlated marginally and negatively with Post ES when outliers were removed (SOM, Table S1).”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: For the posttest effect size calculation, the authors assumed that treatment and control groups did not differ at pretest.

  • It may state two separate claims as one.

    Findingown data

    The remuneration effect lost significance when outliers were trimmed (p = .22, b = -.07), although the slope remained the same.

    “However, the effect lost significance when outliers were trimmed (p = .22,b = -.07;SOM, Table S1), although the slope remains the same.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: For the posttest effect size calculation, the authors assumed that treatment and control groups did not differ at pretest.

  • It may state two separate claims as one.

    Findingown data

    The overall treatment-control difference in fluid intelligence at posttest was small and significant (g = 0.24, SE = 0.07), and larger than the baseline difference (g = -0.003, SE = 0.08; p = .03).

    “The treatment/control group difference in Gf at posttest (g = 0.24, SE = 0.07) is significantly greater than the treatment/control group difference at baseline (g = -0.003,SE = 0.08; group difference: p = .03).”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: For the posttest effect size calculation, the authors assumed that treatment and control groups did not differ at pretest.

  • It may state two separate claims as one.

    Findingown data

    Transfer occurred equally to verbal and visual fluid intelligence tasks, though only a minority of studies used verbal reasoning (n = 5), making this finding preliminary.

    “Transfer seemed to occur equally to both verbal and visual Gf tasks, although there were only a minority of studies that employed any form of verbal reasoning (n = 5), making this finding preliminary.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim.

    Findingown data

    The shortest session length in the sample was 18.5 minutes.

    “Importantly, the shortest session length in our sample was only 18.5 min”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    The baseline effect was near zero, which the authors take to mean the groups did not differ before training.

    “With a baseline ES of essentially 0, we conclude that no preexisting differences are present at the group level between treatment and control groups.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    Treatment groups showed large improvements (Tx ES), which is calculated independently of the control group.

    “large improvements were found in Tx ES (Table 1), which is calculated independently of the control group.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Findingown data

    International studies had a larger average posttest effect size (0.44, SE = .10) than U.S. studies (0.06, SE = .09), and the difference was significant at p < .01.

    “international (outside the U.S.) studies demonstrated an average ES of .44 (SE = .10),while U.S. studies only averaged .06(SE = .09).The difference is significant at the p < .01level.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: For the posttest effect size calculation, the authors assumed that treatment and control groups did not differ at pretest.

  • It may state two separate claims as one.

    Findingown data

    The regression analysis gave equivocal support to the hypothesis that remuneration reduces intrinsic motivation and harms training outcomes.

    “Our regression analysis equivocally supports this hypothesis.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. The quote only partly states this claim.

    Findingown data

    All twenty studies in the meta-analysis were published or completed from 2008 through 2013.

    “The 20 studies included here were all completed between 2008 and 2013.”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingown data

    Studies using passive controls showed more net transfer (g = 0.44, SE = .10) than those using active controls (g = 0.06, SE = .09), a difference significant at p < .01.

    “Second, studies that used passive controls demonstrated more net transfer (g = 0.44, SE = .10)than those with active controls (g = 0.06, SE = .09),a difference that is also statistically significant at p < .01.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: For the posttest effect size calculation, the authors assumed that treatment and control groups did not differ at pretest.

  • It may state two separate claims as one.

    Findingown data

    Alternative regression models with trimmed outliers showed marginally significant negative trends between fluid intelligence transfer and both starting n-back level (p = .06) and session length (p = .06).

    “the SOM contains alternate regression models with trimmed outliers that revealed marginally significant negative trends between Gf transfer and both starting n-back level (p = .06)and session length (p = .06;Table S1).”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Findingown data

    Geographic region and remuneration together explain a sizable portion of the passive control effect and reduce it to statistical insignificance.

    “These factors, individually and together, account for a sizable portion of the passive control effect and reduce its estimated effect to the point of statistical insignificance (Table 3).”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Interpretationown data

    The attrition difference suggests cultural differences in dropout that may reflect general compliance and investment in research.

    “suggesting cultural differences in attrition that may be reflective of issues surrounding general compliance and investment in research.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it.

    Interpretationreasoning

    Even small improvements in fluid intelligence can have profound societal ramifications.

    “even small improvements can have profound societal ramifications”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: Fluid intelligence is a fundamental cognitive ability that underpins many life functions.

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Interpretationcitation

    Cultural effects may contribute to differences between international and U.S. participants and between researchers' methodological practices.

    “These effects may contribute to differences not only between international and U.S. research participants, but also between methodological practices of researchers.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim.

    Study design and conductown data

    The regression results should be treated as preliminary because of heavy outlier influence and small sample sizes.

    “A note of caution is warranted here regarding these regression results: given the heavy influence of outliers (SOM, Table S1) and the relatively small sample sizes, these results should be viewed as preliminary”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductown data

    It is unclear what would happen below the 18.5-minute data range.

    “so it is not clear what would happen below this data range.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Interpretationown data

    The starting n-back trend suggests that participants who begin with more room to improve (lower day-1 n-back level) may gain the most.

    “The former suggests that those who start with more room to improve (i.e., lower n-back level on day 1) may also gain the most.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Premisecitation

    There is a substantial body of literature on culture's effects on cognition.

    “There is a substantial body of literature available on the effects of culture on cognition (cf.Muggleton & Banissy, 2014).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Muggleton & Banissy, 2014 source

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Interpretationown data

    The precise nature of remuneration's effects on n-back training remains unclear.

    “Therefore, while the precise nature of the effects of remuneration for n-back training are still unclear”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Premisecitation

    Passive control groups account for practice effects from repeated testing but not for motivational or Hawthorne effects that come with participating in an intervention.

    “While this does control for important test-retest practice effects, it does not control for potential motivational or Hawthorne effects associated with being enrolled in an intervention study (Shipstead, Redick, & Engle, 2012).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Shipstead, Redick, & Engle, 2012 source

  • Probably not a substantial claim. The paper’s conclusions may not depend on it.

    Study design and conductreasoning

    The precise reasons for the observed regional differences are not immediately clear because the meta-analysis includes a plurality of cultures.

    “However, due to the plurality of cultures (mostly European and American) represented in our meta-analysis, the precise reasons for the observed regional differences are not immediately clear.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Study design and conductown data

    Coding was done independently by two teams and had high interrater reliability (94.14% agreement, κ = 0.88).

    “Percent agreement on the coding (interrater reliability) was high (94.14%),with κ = 0.88 (SE = 0.03) using a conservative expected agreement estimate of 50% (i.e., match or no match).”
    Quote in the paperNot a substantial claim

  • It may state two separate claims as one.

    Interpretationown data

    The training effect may extend beyond matrix reasoning tests and reflect a broader gain in fluid intelligence.

    “thereby suggesting that the transfer of n-back training may not be restricted to matrix reasoning tests (Shipstead et al., 2012) but, rather, may reflect a more global improvement in Gf.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Shipstead et al., 2012 source

  • It may state two separate claims as one.

    Interpretationown data

    The lack of baseline difference rules out preexisting cultural differences or participant differences due to selective U.S. universities.

    “thereby ruling out preexisting differences between cultures or differences in participant characteristics due to the highly selective universities in which certain U.S. studies were conducted.”
    Quote in the paperSubstantial claim

  • It may state two separate claims as one.

    Premisecitation

    In those studies, active controls showed no gain from baseline and were not better than passive controls, which suggests the null transfer result did not depend on control type.

    “It is important to note, however, that their active control groups did not improve over baseline, nor did they outperform their associated passive control groups, suggesting that the failure to find transfer was irrespective of control type.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites Harrison et al., 2013 source
    • cites Redick et al., 2013 source

  • Probably not a substantial claim.

    Interpretationown data

    This meta-analysis of 20 studies, 30 treatment groups, 24 control groups, and 98 fluid intelligence measures is a first step toward providing extraordinary evidence for n-back training's effect on fluid intelligence.

    “this meta-analysis of 20 different studies, 30 different treatment groups, and 24 different control groups, with data collected from 98 different Gf outcome measures, is a first step toward providing this extraordinary evidence.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

    Premisecitation

    The literature on rewards and compensation and performance is extensive and sometimes mixed.

    “Moreover, the effects of rewards and compensation on performance have an extensive body of literature behind them, which at times can also show mixed results (Cameron, Banko, & Pierce, 2001).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Cameron, Banko, & Pierce, 2001 source
40 of 40

  • The authors state this as an assumption.

    For the posttest effect size calculation, the authors assumed that treatment and control groups did not differ at pretest.

    “under the assumption that no group differences existed at pretest.”

  • The paper relies on this without support.

    Fluid intelligence is a fundamental cognitive ability that underpins many life functions.

    “Since Gf is a fundamental cognitive skill that underlies a wide range of life functions”

  • The paper relies on this without support.

    The healthy young adults in the meta-analysis are representative of society's workforce.

    “particularly given the healthy young adults in our analyses, representative of society's workforce.”

  • The paper relies on this without support.

    Prior claims about n-back training's effect on fluid intelligence are usually based on single studies with methodological limitations.

    “Although claims to date are ordinarily based on single studies, each with various methodological limitations”
4 of 4

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?

Statistics recomputed

Every reported test result, recomputed from its test statistic and degrees of freedom. A result shows as not adding up only once an independent check has confirmed it.

  • Consistentt(34) = 2reported p = .05 · recomputed p = .054

    “t(34) = 2.0, p = .05”

  • Consistentt(14) = 2.23reported p = .04 · recomputed p = .043

    “t(14) = 2.23, p = .04”

2 of 2

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.

  • Passive-control standard error conflicts with the reported subgroup significance

    Minor

    The table gives treatment-group effects of .25 (SE .10) and .54 (SE .90) for active- and passive-control studies. A difference of .29 with those standard errors cannot support the stated p = .04 comparison; the passive-control standard error may be mistyped.

1 of 1

Citations

Cites

Flagged first, then by citations

2 of the 73 references couldn’t be matched to a known paper.

Research works printed but not linked: 2

  1. Borenstein, M., Higgins, J., & Rothstein, H. (2005). Comprehensive meta-analysis version 2. Englewood: Biostat.
  2. Dunst, C. J., Hamby, D. W., & Trivette, C. M. (2004). Guidelines for Calculating Effect Sizes for Practice-Based Research Syntheses. Centerscope, 3(1), 1–10.

Cited by

Sorted by citations

Related

Metadata

Authors
Au, J., Sheehan, E., Tsai, N., Duncan, G. J., Buschkuehl, M., & Jaeggi, S. M.
Affiliations
  • University of California, Irvine (US)
  • MIND Research Institute (US)
Corresponding author
Au, Jacky
ORCID
Venue
Psychonomic Bulletin & Review
Publisher
Springer Science and Business Media LLC
Published
2014-08-08
Volume
22
Issue
2
Pages
366-377
ISSN
  • 1069-9384 (ISSN-L)
  • 1531-5320
DOI
10.3758/s13423-014-0699-x
OpenAlex
W1985948545
Semantic Scholar
669a9f1a189a86de5adae42b90991148d90e53fc
Cited by
727
References
73
Citations by year
2026: 33 · 2025: 39 · 2024: 42 · 2023: 45 · 2022: 57 · 2021: 66 · 2020: 97 · 2019: 73 · 2018: 78 · 2017: 81 · 2016: 63 · 2015: 47 · 2014: 4
FWCI
7.44
Type
journal article
Language
en
Open access
Licence
www.springer.com/tdm (tdm)
Locations
  • Psychonomic Bulletin & Review (publishedVersion)
  • PubMed (publishedVersion)
  • eScholarship (California Digital Library) (submittedVersion)
Topics
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Cognitive Abilities and Testing
  • Social Sciences › Psychology › Developmental and Educational Psychology › Cognitive and developmental aspects of mathematical skills
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Cognitive Functions and Memory
Keywords
working memory training, fluid intelligence, cognitive training, transfer of training, cognitive plasticity, working memory
MeSH terms
Adolescent, Adult, Aptitude, Attention, Female, Humans, Intelligence, Learning, Male, Memory, Short-Term, Middle Aged, Practice, Psychological, Transfer, Psychology, Controlled Clinical Trials as Topic, Young Adult

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.

A language model drew the claims from the paper on 2026-10-06, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-06 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (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