CoChat
Sign Up
Cited by 238Cognitive Abilities and TestingCognitive and developmental aspects of mathematical skillsCognitive Functions and Memory

The DOI matches its Crossref record; 186 of the 208 references it cites were found in a registry or an index; none of the 128 sources its claims cite is checked for support yet.

Working memory training involves learning new skills

Open access
Gathercole, Susan E., Dunning, Darren L., Holmes, Joni, Norris, Dennis·2019·Journal of Memory and Language
DOI 10.1016/j.jml.2018.10.003PMC PMC6591133
Open this paper in CoChat — free account

Abstract

We present a new framework characterizing training-induced changes in WM as the acquisition of novel cognitive routines akin to learning a new skill. Predictions were tested in three studies analyzing the transfer between WM tasks following WM training. Study 1 reports a meta-analysis establishing substantial transfer when trained and untrained tasks shared either a serial recall, complex span or backward span paradigm. Transfer was weaker for serial recall of verbal than visuo-spatial material, suggesting that this paradigm is served by an existing verbal STM system and does not require a new routine. Re-analysis of published WM training data in Study 2 showed that transfer was restricted to tasks sharing properties proposed to require new routines. In a re-analysis of data from four studies, Study 3 demonstrated that transfer was greatest for children with higher fluid cognitive abilities. These findings suggest that development of new routines depends on general cognitive resources and that they can only be applied to other similarly-structured tasks.

Overview

AI snapshot

Written from the paper’s full text.

TL;DR

The paper proposes that working memory (WM) training produces transfer mainly when training requires learning a new cognitive routine, akin to acquiring a cognitive skill, rather than by generally expanding WM capacity. Across a meta-analysis and two re-analyses of child training data, transfer was strongest for tasks sharing paradigms thought to require new routines, such as visuo-spatial serial recall, complex span, and backward span.

Key findings

  • In Study 1, across 24 RCTs and 113 trained–untrained task pairs, mean transfer was small-to-moderate (d = 0.42); transfer was high when tasks shared a paradigm: serial recall matched d = 0.709, complex span matched d = 0.540, backward span matched d = 0.901.
  • Visuo-spatial serial recall showed larger matched transfer (d = 1.054) than verbal serial recall (d = 0.508), and complex span transfer was substantial within domain (d = 0.540) but absent cross-domain (d = 0.119).
  • In Study 2 (n = 106 children with low WM), adaptive Cogmed training produced transfer for dot matrix (d = 0.92), backward digits (d = 0.77), and Mr. X (d = 0.58), but weak digit span transfer (d = 0.29); verbal and performance IQ showed evidence for the null.
  • In Study 3 (n = 108 children), all four WM tests improved after adaptive training (dot matrix d = 1.27, backward digit span d = 0.96, Mr. X d = 0.75, digit span d = 0.37); baseline performance IQ predicted dot matrix and Mr. X (p < .005), backward digit span (p < .05), but not digit span.

Methods

  • Study 1: meta-analysis of randomized controlled trials of adaptive WM training with active control conditions, restricted to ordered recall tasks; 24 studies yielded 113 trained–untrained task pairs, analyzed with random-effects models, Cohen’s d, moderator analyses, heterogeneity, and publication-bias checks.
  • Study 2: re-analysis of two published studies of children aged 7–12 with low WM (n = 106) who completed adaptive or non-adaptive Cogmed RM; transfer tests were AWMA digit span, dot matrix, backward digit span, and Mr. X, plus WASI verbal and performance IQ; analyses used GLMs and Bayesian regressions.
  • Study 3: re-analysis of four WM training studies (n = 108 children, including low-WM, ADHD, and typically developing samples) all completing adaptive Cogmed RM; individual differences in post-training AWMA WM scores were predicted from baseline WM, verbal IQ, and performance IQ using GLMs and Bayesian regressions.

Limitations

  • The paper states that Study 1’s independent feature coding is limited because matched features are often correlated, making it hard to identify the critical feature; for example, matched digits were confounded with backward span, and verbal material was confounded with shared paradigm.
  • The paper states that Study 2’s participants were all children with low WM, so generalization to other populations or ages is undetermined; Study 3’s sample was cognitively heterogeneous and unusual, so generalization requires further data.
  • The paper also notes insufficient data for updating paradigms, limited generalization evidence for backward span because most verbal pairs differed only in response modality, and unresolved limits on intermediate-level routine transfer; evident from the text, the meta-analysis includes heterogeneous studies and some moderator analyses had fewer than 10 cases for features.

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

Integrity signals

Registry

Crossref · 10.1016/j.jml.2018.10.003

Cites

186 of 186 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

0 of 128 cited sources checked · 0 supported · 0 flagged

  • Probably not a substantial claim. 54
  • It may state two separate claims as one. 77
  • The quote only partly states this claim. 11
126 of 126

  • The authors state this as an assumption.

    Training on unfamiliar working memory tasks leads to development of novel cognitive routines that control the sequence of cognitive processes required to perform the task.

    “First, training on unfamiliar WM tasks will lead to the development of novel cognitive routines that control the sequence of cognitive processes required to perform the task.”

  • The authors state this as an assumption.

    These routines can be applied only to other tasks with common structures, and only then will transfer occur.

    “Second, these routines can be applied only to other tasks with common structures and only then will transfer occur.”

  • The authors state this as an assumption.

    Transfer following working memory training is restricted to cases where the training tasks require the establishment or refinement of a cognitive routine that is not already fully developed.

    “A core assumption is that transfer following WM training is restricted to cases where the training tasks require the establishment or refinement of a cognitive routine that is not already fully developed.”

  • The authors state this as an assumption.

    The contribution of rehearsal to verbal serial recall develops over the early school years and becomes fully functional by about seven years.

    “But most importantly for the present purposes, it is reasonable to assume that the contribution of rehearsal to verbal serial recall develops over the early school years and becomes fully functional by seven years or so.”

  • The authors state this as an assumption.

    Established processes involved in encoding verbal item and order information are not amenable to further training.

    “A key assumption of the framework is that established processes such as those involved in encoding verbal item and order information should not be amenable to further training.”

  • The paper relies on this without support.

    The field lacks detailed accounts of the cognitive changes that take place during training.

    “the field lacks detailed accounts of the cognitive changes that take place.”

  • The authors state this as an assumption.

    For most individuals, the recoding strategies require new routines that can transfer to other WM tasks with similar stimulus content.

    “We assume that for most individuals these strategies require the development of new routines that have the potential to transfer to other WM tasks with similar stimulus content.”

  • The paper relies on this without support.

    Most WM training programs include complex WM tasks combining storage with additional processing demands.

    “Most training programs include complex WM tasks that combine the temporary storage demands of simple serial recall tasks with additional processing requirements such as changing the order of items at recall, updating the contents of WM, or handling irrelevant distraction.”

  • The paper relies on this without support.

    In all relevant training studies known to the authors, trained and untrained complex span tasks use different distractor activities.

    “Indeed, the distractor activities employed in the trained and untrained complex span tasks differ in all relevant training studies familiar to the authors.”

  • The paper relies on this without support.

    Contemporary WM training programs usually involve multiple training activities with differing materials and fewer than 20 hours of training in total.

    “In contrast, contemporary training programs usually involve multiple training activities with differing materials and involved fewer than 20 h of training in total.”

  • The paper relies on this without support.

    In most complex span tasks, distractor activities are unrelated to the to-be-recalled items.

    “In most complex span tasks, the distractor activities are unrelated to the to-be-recalled stimulus items.2”

  • The paper relies on this without support.

    The studies combined in the meta-analysis are sufficiently comparable for a random-effects synthesis despite variation in training programs, samples, and outcome measures.

    “Due to variation in studies (e.g. type of training program used, age of sample, outcome measures used) a random effects model was chosen for all analyses.”

  • The paper relies on this without support.

    The coding of task features used in the meta-analysis validly captures the cognitive demands relevant to transfer.

    “One limitation of the coding method is that although individual features are coded and analyzed independently of one another, matched features are often highly or indeed perfectly correlated with one another.”

  • The paper relies on this without support.

    The active control conditions in the included studies adequately control for nonspecific training effects.

    “RCTs of adaptive WM training with active control training conditions.”

  • The authors state this as an assumption.

    Cognitive routines are hierarchically organized into subroutines that control the execution of processes for different task components.

    “Our starting position was that routines are hierarchically organized, consisting of subroutines that control the execution of processes necessary to accomplish different task components (e.g., encoding, maintenance, distractor processing, retrieval).”

  • The paper relies on this without support.

    New routines are needed only when existing processes and mechanisms cannot satisfy the demands of the task.

    “These are only needed when existing processes and mechanisms are not sufficient to satisfy the demands of the task.”

  • The paper relies on this without support.

    Performance IQ indexes the fluid cognitive abilities or general attentional resources relevant to constructing new routines.

    “Higher performance IQ scores were uniquely associated with the magnitude of transfer to the three WM tests (visuo-spatial serial recall, visuo-spatial complex span and backward span) hypothesized to require new routines.”

  • The paper relies on this without support.

    The four WM training studies' samples can be combined for individual differences analysis.

    “To achieve the statistical power and heterogeneity required to examine individual differences in transfer following training, we re-analyzed data from four WM training studies in which measures of IQ as well as WM were obtained before and after training.”
18 of 18

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?

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.

  • Effect sizes below .5 are labeled large

    Minor

    The auditory and visual matched effect sizes are called large, although the paper's stated convention places the moderate benchmark at .5 and the large benchmark at .8.

  • Cross-modal response data are both reported and denied

    Minor

    The manual-response results report a task pair with spoken recall at transfer and manual recall during training, then state that no cross-modal task-pair data are available.

  • Study 1 reports both 20 and 24 studies

    Minor

    The participant breakdown accounts for 20 studies (11 plus nine), but the discussion says the meta-analysis included 24 studies.

3 of 3

Citations

Cites

Flagged first, then by citations

Show all 208 references

22 of the 208 references couldn’t be matched to a known paper.

Research works printed but not linked: 1

  1. Chase, 1981, Skilled memory, Cognitive Skills and their Acquisition, 141

Cited by

Sorted by citations

See all 238

Related

See all

Metadata

Authors
Gathercole, S. E., Dunning, D. L., Holmes, J., & Norris, D.
Affiliations
MRC Cognition and Brain Sciences Unit (GB)
Corresponding author
Gathercole, Susan E.
ORCID
Venue
Journal of Memory and Language
Publisher
Elsevier BV
Published
2019-04
Volume
105
Pages
19-42
ISSN
0749-596X (ISSN-L)
DOI
10.1016/j.jml.2018.10.003
PMC
PMC6591133
OpenAlex
W2902394180
Semantic Scholar
4f0858b4d100dd2fffebd2b8f2232c34d70dd12c
Cited by
225
References
208
Citations by year
2026: 15 · 2025: 25 · 2024: 35 · 2023: 31 · 2022: 36 · 2021: 30 · 2020: 34 · 2019: 18
FWCI
27.18
Type
journal article
Language
en
Update notices
erratum (2019-06-01)
Open access
Licence
Locations
  • Journal of Memory and Language (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • UEA Digital Repository (University of East Anglia) (acceptedVersion)
  • PubMed Central (acceptedVersion, cc-by)
  • UEA Digital Repository (University of East Anglia) (submittedVersion)
  • Apollo (University of Cambridge) (submittedVersion, cc-by)
  • Apollo (cc-by)
Funders
  • Wellcome Trust (104908/Z/14/Z, 107496)
  • Leverhulme Trust (F00/224/AI)
  • Medical Research Council
  • Economic and Social Research Council (RES-000-23-0979, RES-000-)
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, transfer of training, skill acquisition, serial recall, complex span, short-term memory, fluid intelligence, children

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 2026-10-08, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-08 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

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