The asymmetry of working memory training transfer: A systematic review and meta-analysis

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

Perceptual learning studies report transfer from configuration to element learning but not the opposite.

“Perceptual learning studies similarly report transfer from configuration to element learning but not the opposite”

• cites Gong et al., 2022 sourcePartly supported

  The Discussion reports complete configuration-to-element transfer and an asymmetric pattern. It describes element-to-configuration transfer as weak, however, rather than absent, so “not the opposite” overstates the finding. (LLM, conf=0.88)

  Experimental — accuracy not yet measured

May combine two claims.

Both span and updating recruit frontoparietal systems, with updating eliciting stronger and more widespread engagement.

“Neuroimaging evidence likewise shows that both recruit frontoparietal systems, with updating eliciting stronger and more widespread engagement, which is consistent with greater cognitive demand (Rodríguez-Nieto et al., 2022; Wager & Smith, 2003).”

• cites Rodríguez-Nieto et al., 2022 sourcePartly supported

  Section 4.4 reports frontoparietal recruitment in both storage-related Recognition tasks and executive-component Updating tasks. It does not compare span tasks directly with updating; moreover, it describes the storage-related Recognition network as more distributed than the Manipulation and N-back networks, rather than the reverse. The proposed explanation in terms of greater cognitive demand is not established for a span-versus-updating comparison. (LLM, conf=0.83)

  Experimental — accuracy not yet measured
• cites Wager & Smith, 2003 sourceNot checked — no full text of the source

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

Task similarity has dominated transfer frameworks for more than a century.

“Since Thorndike’s common elements theory was established in the 1900 s, task similarity has dominated transfer frameworks for more than a century, centering on similarity as a primary determinant of training transfer likelihood (Waris et al., 2015; Woodworth & Thorndike, 1901).”

• cites Waris et al., 2015 sourcePartly supported

  The Introduction and Discussion (pages unavailable) treat cognitive overlap between training and transfer tasks as a predictor of transfer strength. They do not establish that task similarity has dominated transfer frameworks since Thorndike’s theory or for more than a century. (LLM, conf=0.84)

  Experimental — accuracy not yet measured
• cites Woodworth & Thorndike, 1901 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

Evidence on stimulus domain and transfer is unclear/competing.

“Some scholars argue that the novelty of visuospatial materials facilitates new learning and transfer (Gathercole et al., 2019; Norris et al., 2019), whereas competing evidence suggests stronger transfer following verbal WM training (Ni et al., 2023)”

• cites Gathercole et al., 2019 sourcePartly supported

  In Study 1, the paper proposes that training on spatial serial-recall tasks can develop new cognitive routines and produce robust transfer because an established visuospatial short-term-memory system may be absent. It does not attribute that effect to the novelty of visuospatial materials generally; its prediction concerns particular recall tasks, and the Study 1 Discussion says transfer across other kinds of visuospatial stimuli could not be tested. Page numbers were not available in our copy. (LLM, conf=0.82)

  Experimental — accuracy not yet measured
• cites Norris et al., 2019 sourcePartly supported

  The Discussion (page unavailable) proposes that the unfamiliar demands of a visual change-detection task may prompt a new cognitive routine that produces transfer. That supports a qualified link between novelty, learning and transfer, but not the broader claim that novelty of visuospatial materials generally facilitates transfer: the Results (page unavailable) found no substantial cross-domain transfer from circle-span training. (LLM, conf=0.82)

  Experimental — accuracy not yet measured
• cites Ni et al., 2023Not matched

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

May combine two claims.

Span performance primarily reflects storage/maintenance, whereas updating reflects maintenance plus additional manipulation/control components.

“Latent-variable and theoretical evidence suggests that span performance primarily reflects storage/maintenance, whereas updating reflects maintenance plus additional manipulation/control components (Aben et al., 2012; Burgoyne et al., 2025).”

• cites Aben et al., 2012 sourcePartly supported

  In “Simple Span Versus Complex Span” (page unavailable), the paper describes simple span tasks as primarily involving maintenance and lists n-back among working-memory tasks involving maintenance and manipulation. It also says span tasks are not process-pure: complex spans add processing demands, so the claim that span performance generally reflects storage or maintenance is too broad. (LLM, conf=0.85)

  Experimental — accuracy not yet measured
• cites Burgoyne et al., 2025 sourceSupported

May combine two claims.

The cognitive operations and neural correlates of both span and updating paradigms are relatively well characterized.

“This design reduced conceptual heterogeneity and strengthened interpretability because the cognitive operations and neural correlates of both paradigms are relatively well characterized (Aben et al., 2012; Burgoyne et al., 2025).”

• cites Aben et al., 2012 sourcePartly supported

  The Simple Span Versus Complex Span section describes what span and n-back tasks require, and Neurobiological Findings discusses brain activity during memory tasks, including n-back. But the paper cautions that these tasks are not process-pure, that their underlying constructs are not well established, and that imaging findings are difficult to interpret. It therefore does not fully support the claim that the operations and neural correlates of both paradigms are relatively well characterized. (LLM, conf=0.87)

  Experimental — accuracy not yet measured
• cites Burgoyne et al., 2025 sourcePartly supported

  The Discussion and Conclusion distinguish the cognitive demands of complex span and updating tasks, particularly maintenance versus disengagement. The paper mentions n-back’s use in fMRI research, but the passages do not characterize the neural correlates of both paradigms. Page numbers were not available in our copy. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

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

Compared with nondominant-limb training, dominant-limb training results in more bilateral motor activation and stronger transfer, and a related pattern has been reported for discrete-to-rhythmic movement transfer.

“For example, compared with nondominant-limb training, dominant-limb training results in more bilateral motor activation and stronger transfer (Teixeira, 2000), and a related pattern has been reported for discrete-to-rhythmic movement transfer (Ikegami et al., 2010; Schaal et al., 2004).”

• cites Teixeira, 2000 sourceNot checked — no full text of the source
• cites Ikegami et al., 2010 sourceSupported
• cites Schaal et al., 2004 sourceNot checked — no full text of the source

May combine two claims.

Transfer has been observed from high-precision visual perception training to relatively low-precision WM tasks, but not in the reverse direction.

“Consistent with this interpretation, transfer has been observed from high-precision visual perception training to relatively low-precision WM tasks (Jia et al., 2021), but not in the reverse direction (Cai et al., 2022).”

• cites Jia et al., 2021 sourceSupported
• cites Cai et al., 2022 sourceNot checked — no full text of the source

May combine two claims.

The training-dose categories were defined empirically from the distribution of training hours rather than from trajectory-based markers of learning, so the pattern should not be taken as direct evidence regarding fast versus slow learning phases in WM training.

“However, this pattern should not be taken as direct evidence regarding fast versus slow learning phases in WM training (Ericson & Klingberg, 2023; Yamashita et al., 2015), because our training-dose categories were defined empirically from the distribution of training hours rather than from trajectory-based markers of learning.”

• cites Ericson & Klingberg, 2023 sourceSupported
• cites Yamashita et al., 2015 sourcePartly supported

  Split into 2 atoms — 1 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

  Experimental — accuracy not yet measured

May combine two claims.

Core executive functions are separable but partially overlapping, and demanding cognitive tasks can recruit overlapping domain-general control systems.

“This rationale is consistent with evidence that core executive functions are separable but partially overlapping (Friedman & Miyake, 2017; Miyake et al., 2000) and that demanding cognitive tasks can recruit overlapping domain-general control systems (Duncan, 2010; Fedorenko et al., 2013).”

• cites Friedman & Miyake, 2017 sourceSupported
• cites Miyake et al., 2000 sourceNot checked — no full text of the source
• cites Duncan, 2010 sourceNot checked — no full text of the source

May combine two claims.

Visual novelty can facilitate within-paradigm transfer, but paradigm-specific benefits associated with visual novelty may not support or even hinder cross-paradigm generalization.

“Notably, this pattern differs from evidence that visual novelty can facilitate within-paradigm transfer (Gathercole et al., 2019; Norris et al., 2019), perhaps because the paradigm-specific benefits associated with visual novelty may not support or even hinder cross-paradigm generalization.”

• cites Gathercole et al., 2019 sourcePartly supported

  The General discussion (page unavailable) reports transfer when trained and untrained tasks share a paradigm, including visuo-spatial serial recall. It does not identify visual novelty as the factor facilitating that transfer. The suggestion about cross-paradigm generalization is framed as the citing authors’ own explanation, rather than a finding attributed to this source. (LLM, conf=0.82)

  Experimental — accuracy not yet measured
• cites Norris et al., 2019 sourcePartly supported

  The Results (Transfer data; page unavailable) show transfer from color-change training to an untrained line-orientation task using the same paradigm. The Discussion describes the change-detection demands as unfamiliar and proposes a new cognitive routine as one possible explanation, but it does not establish that visual novelty caused the transfer. The cross-paradigm explanation follows the citations as the citing authors’ own speculation. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

May combine two claims.

The transfer advantage does not simply scale with the amount of training, which is consistent with previous meta-analytic evidence showing no reliable dose-transfer relationship.

“This suggests that the transfer advantage does not simply scale with the amount of training, which is consistent with previous meta-analytic evidence showing no reliable dose-transfer relationship (Au et al., 2015; Soveri et al., 2017a).”

• cites Au et al., 2015Not matched

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

• cites Soveri et al., 2017a sourceSupported

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

Computational work highlights the importance of feature-invariant representations for transfer.

“This interpretation is broadly consistent with computational work highlighting the importance of feature-invariant representations for transfer (Johnston & Fusi, 2023; Yosinski et al., 2014).”

• cites Johnston & Fusi, 2023 sourceSupported
• cites Yosinski et al., 2014Not matched

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

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

One possible explanation is that updating ability matures later than span ability does, leaving greater room for updating-based training to benefit young individuals.

“One possible explanation is that updating ability matures later than span ability does, leaving greater room for updating-based training to benefit young individuals (Ahmed et al., 2022; Skalaban et al., 2022).”

• cites Ahmed et al., 2022 sourcePartly supported

  The Method section describes backward digit span as requiring manipulation and updating, and the Discussion reports that backward-span performance emerged at age 5, compared with age 3 for forward span. This supports a later onset for an updating-demanding task, but the paper does not establish when updating ability matures relative to span ability or test whether updating-based training benefits young people more. (LLM, conf=0.84)

  Experimental — accuracy not yet measured
• cites Skalaban et al., 2022 sourcePartly supported

  The Results (Development of working memory; page unavailable) show that n-back performance peaks in the late teens and early twenties, supporting continued development of working memory in young people. The paper does not compare the maturation of updating and span abilities or test whether updating-based training benefits young people. (LLM, conf=0.77)

  Experimental — accuracy not yet measured

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

Clinical conditions are heterogeneous, and emerging evidence suggests that transfer patterns may vary across etiologies.

“This question is especially important given the heterogeneity of clinical conditions and emerging evidence that transfer patterns may vary across etiologies (Zhang et al., 2025).”

• cites Zhang et al., 2025 sourcePartly supported

  The Introduction describes heterogeneity in intervention effects within schizophrenia, and the Discussion reports near-transfer from working-memory training but no far-transfer to other executive-function domains. The paper does not compare clinical etiologies or say that transfer patterns vary across them; its findings are limited to patients with schizophrenia. (LLM, conf=0.83)

  Experimental — accuracy not yet measured

May combine two claims.

The asymmetric transfer patterns persisted when dosage thresholds were increased to 7 h and 10 h, although statistical significance was attenuated.

“The asymmetric transfer patterns persisted when dosage thresholds were increased to 7 h and 10 h, although the statistical significance was attenuated (see Appendix E, Table S5 (OSM)).”

May combine two claims.

Follow-up analyses stratifying tasks into dual versus single n-back and simple versus complex span consistently replicated the asymmetric transfer pattern, excluding subtype variation as the cause.

“These analyses consistently replicated the asymmetric transfer pattern (see OSM Table S3), excluding the possibility that the asymmetry was caused by subtype variation.”

May combine two claims.

The updating-to-span transfer effect was quantitatively greater in non-adults than in adults, but this difference was not statistically significant (F(1, 154) = 2.072, p = .152).

“This developmental divergence may stem from a quantitatively greater updating-to-span transfer effect in non-adults than in adults (F(1, 154) = 2.072, p =.152), whereas the span-to-updating transfer effect did not differ between the two populations (F(1, 41) = 1.498, p =.228).”

May combine two claims.

Egger's regression test confirmed a nonsignificant intercept (t(202) = −0.087, p = .931), indicating no systematic publication bias in the effect-size estimation.

“The funnel plot exhibited notable symmetry (Fig. 3), and Egger’s regression test confirmed a nonsignificant intercept (t(202) = −0.087, p =.931), indicating that there was no systematic publication bias in the effect-size estimation.”

May combine two claims.

No significant differences emerged for either transfer direction between the verbal and nonverbal stimulus groups (Fs < 0.026, ps > .873).

“No significant differences emerged for either transfer direction between the two stimulus groups (Fs < 0.026, ps >.873), and we proposed that the asymmetry observed in the verbal training group may be attributable to its smaller standard error (verbal: SE =.060 vs. nonverbal: SE =.114).”

May combine two claims.

Updating training showed broader transfer to non-WM outcomes, including intelligence-related measures, whereas span training showed little evidence of such effects.

“Updating training showed broader transfer to non-WM outcomes, including intelligence-related measures, whereas span training showed little evidence of such effects.”

May combine two claims.

Developmental stage modulated the transfer asymmetry: compared with adults, non-adults exhibited greater updating-to-span transfer, leading to more significant asymmetry.

“Second, developmental stage modulated the transfer asymmetry: Compared with adults, non-adults exhibited greater updating-to-span transfer, leading to more significant asymmetry.”

May combine two claims.

The transfer advantage was evident at low training doses, and increasing training duration was not associated with progressively greater transfer effects in either direction.

“First, the transfer advantage was evident at low training doses, and increasing training duration was not associated with progressively greater transfer effects in either direction.”

Classified as probably not a substantial claim.

Asymmetry-like patterns have been suggested across studies.

“asymmetry-like patterns have been suggested across studies (Cai et al., 2022; Schwarb et al., 2016)”

• cites Cai et al., 2022 sourceNot checked — no full text of the source
• cites Schwarb et al., 2016 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

n-back training has been reported to transfer to syntactic processing more than syntactic training transfers back to WM outcomes.

“n-back training has been reported to transfer to syntactic processing more than syntactic training transfers back to WM outcomes (Roque-Gutierrez & Ibbotson, 2023)”

• cites Roque-Gutierrez & Ibbotson, 2023 sourceNot checked — no full text of the source

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

Motor-learning studies report stronger transfer from dominant- to nondominant-limb training.

“motor-learning studies report stronger transfer from dominant- to nondominant-limb training (Teixeira, 2000)”

• cites Teixeira, 2000 sourceNot checked — no full text of the source

Classified as probably not a substantial claim. May combine two claims.

Publication bias was assessed using funnel plot, Egger's regression test, and fail-safe number.

“To determine the risk of publication bias, we first generated a funnel plot by plotting individual study effect sizes (x-axis) against their standard errors (y-axis).”

May combine two claims.

Sensitivity analyses excluding older adult samples yielded virtually identical results, indicating that the adult pattern was not driven by the inclusion of older participants.

“To ensure stability, sensitivity analyses excluding older adult samples were conducted and yielded virtually identical results, indicating that the adult pattern was not driven by the inclusion of older participants (see Table S7, OSM).”

Classified as probably not a substantial claim. May combine two claims.

One possible explanation is that updating training, which requires continuous monitoring and replacement of WM content, involves broader or more integrative processing than span training does, and therefore supports wider transfer.

“One possible explanation is that updating training, which requires the continuous monitoring and replacement of WM content, involves broader or more integrative processing than span training does, and therefore supports wider transfer.”

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

The asymmetry observed in the verbal training group may be attributable to its smaller standard error (verbal SE = .060 vs. nonverbal SE = .114).

“No significant differences emerged for either transfer direction between the two stimulus groups (Fs < 0.026, ps >.873), and we proposed that the asymmetry observed in the verbal training group may be attributable to its smaller standard error (verbal: SE =.060 vs. nonverbal: SE =.114).”

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

Two authors independently coded the studies, with an overall agreement rate of 98%.

“the first author coded all the studies, and a second author independently coded them again. Intercoder reliability was exceptionally high, with an overall agreement rate of 98%.”

May combine two claims.

Auditory temporal discrimination training can transfer to visual tasks with lower precision demands, whereas the reverse pattern is not observed, presumably because the trained representations are insufficiently fine grained.

“Consistent with this view, auditory temporal discrimination training can transfer to visual tasks with lower precision demands, whereas the reverse pattern is not observed, presumably because the trained representations are insufficiently fine grained (Bratzke et al., 2012).”

• cites Bratzke et al., 2012 sourceNot checked — no full text of the source

Classified as probably not a substantial claim. May combine two claims.

The literature search was systematic, covering English and Chinese databases up to September 30, 2024.

“We conducted a systematic search across both English- and Chinese-language databases.”

• Also stated in: § Literature search and inclusion criteria · plain

  “The search covered all relevant studies published up to September 30, 2024.”

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

A related possibility is that more demanding training establishes more precise and stable representations, which in turn support transfer.

“A related possibility is that more demanding training establishes more precise and stable representations, which in turn support transfer.”

May combine two claims.

Taken together, the design features support the conclusion that the observed asymmetry is theoretically meaningful: WM transfer appears to be shaped not only by task similarity but also by the relative level and organization of cognitive demand.

“Taken together, these design features support the conclusion that the observed asymmetry is theoretically meaningful: WM transfer appears to be shaped not only by task similarity but also by the relative level and organization of cognitive demand.”

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

The study was preregistered: an a priori analysis plan was uploaded to OSF before initial analyses.

“Before conducting the initial analyses, we uploaded an a priori analysis-plan document to the Open Science Framework (OSF).”

May combine two claims.

WM transfer appears to be shaped not only by whether two tasks overlap, but also by how their cognitive demands are structured relative to one another.

“In other words, WM transfer appears to be shaped not only by whether two tasks overlap, but also by how their cognitive demands are structured relative to one another.”

May combine two claims.

The asymmetry extended to broader non-WM outcomes, suggesting that the transfer advantage of higher-demand training may generalize beyond paradigm-specific WM effects.

“Importantly, this asymmetry also extended to broader non-WM outcomes, suggesting that the transfer advantage of higher-demand training may generalize beyond paradigm-specific WM effects.”

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

Asymmetric transfer may reflect a broader property of learning rather than a WM-specific phenomenon.

“Similar directional asymmetries have been reported in motor learning (Ikegami et al., 2010; Teixeira, 2000), raising the possibility that asymmetric transfer reflects a broader property of learning rather than a WM-specific phenomenon.”

May combine two claims.

The findings support the hypothesis that cognitive/neural plasticity during developmental stages amplifies both the magnitude and the asymmetry of transfer effects.

“These findings support the hypothesis that cognitive/neural plasticity during developmental stages amplifies both the magnitude and the asymmetry of transfer effects.”

May combine two claims.

The asymmetry was most evident under lower training dosages, in non-adult samples, and in studies using verbal materials, indicating important boundary conditions for the proposed account.

“In addition, this asymmetry was most evident under lower training dosages, in non-adult samples, and in studies using verbal materials, indicating important boundary conditions for the proposed account.”

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

Cross-task generalization in frontoparietal systems is enhanced when tasks require active manipulation rather than simple maintenance, indicating that manipulations promote more abstract representations.

“Support for this possibility comes from evidence that cross-task generalization in frontoparietal systems is enhanced when tasks require active manipulation rather than simple maintenance, indicating that manipulations promote more abstract representations (Shi et al., 2025).”

• cites Shi et al., 2025 sourceNot checked — no full text of the source

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

Differences in how sensory pathways project to shared neural systems may contribute to such asymmetries and thus provide a complementary mechanism to the resource-based account proposed here.

“Differences in how sensory pathways project to shared neural systems may contribute to such asymmetries and thus provide a complementary mechanism to the resource-based account proposed here.”

May combine two claims.

Differences in cognitive demand across domains are difficult to quantify using conventional behavioral measures, limiting direct tests of the resource-based account.

“In addition, differences in cognitive demand across domains (e.g., working memory vs. inhibition; Liu et al., 2023) are difficult to quantify using conventional behavioral measures, limiting direct tests of the resource-based account.”

• cites Liu et al., 2023 sourceNot checked — no full text of the source

May combine two claims.

Several methodological limitations of the present meta-analysis warrant consideration, including the greater number of studies examining updating-to-span than span-to-updating transfer, the relatively small number of studies in some moderator analyses, and the possibility that certain updating tasks were insufficiently sensitive to detect transfer from span training.

“First, several methodological limitations of the present meta-analysis warrant consideration, including the greater number of studies examining updating-to-span than span-to-updating transfer, the relatively small number of studies in some moderator analyses, and the possibility that certain updating tasks were insufficiently sensitive to detect transfer from span training.”

May combine two claims.

The supplementary analysis using an alternative storage-to-manipulation versus manipulation-to-storage framework revealed a similar trend but lacked enough statistical power to detect the transfer asymmetry.

“Because only a small number of studies could be recoded using an alternative storage-to-manipulation versus manipulation-to-storage framework, our supplementary analysis revealed a similar trend but lacked enough statistical power to detect the transfer asymmetry (see Table S9, OSM).”

May combine two claims.

By showing that transfer depends not only on task similarity but also on relative cognitive demand, the findings extend traditional accounts of WM training transfer and support a resource-based perspective, according to which training on more demanding tasks transfers more effectively to less demanding tasks.

“By showing that transfer depends not only on task similarity but also on relative cognitive demand, these findings extend traditional accounts of WM training transfer and support a resource-based perspective, according to which training on more demanding tasks transfers more effectively to less demanding tasks.”

May combine two claims.

The present meta-analysis provides the first systematic evidence for asymmetric transfer in WM training: updating-to-span transfer yields reliable gains, whereas span-to-updating transfer remains negligible.

“In conclusion, the present meta-analysis provides the first systematic evidence for asymmetric transfer in WM training: updating-to-span transfer yields reliable gains, whereas span-to-updating transfer remains negligible.”

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

One possible explanation is that verbal processing may facilitate more abstract or relational representations, whereas nonverbal materials may more often elicit concrete and feature-specific coding.

“One possible explanation is that verbal processing may facilitate more abstract or relational representations, whereas nonverbal materials may more often elicit concrete and feature-specific coding (Alfred & Kraemer, 2017).”

• cites Alfred & Kraemer, 2017 sourceNot checked — no full text of the source

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

The possible mechanisms are not mutually exclusive: broader neural recruitment, more precise representations, or more abstract representations may jointly contribute to asymmetric transfer.

“Importantly, these possible mechanisms are not mutually exclusive: broader neural recruitment, more precise representations, or more abstract representations may jointly contribute to asymmetric transfer.”

May combine two claims.

The asymmetry is especially pronounced under lower training doses, in younger samples, and with verbal materials, indicating important boundary conditions for transfer.

“Moreover, the asymmetry is especially pronounced under lower training doses, in younger samples, and with verbal materials, indicating important boundary conditions for transfer.”

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

In the present dataset, some verbal materials also involved richer semantic processing (e.g., sentences), which may further support abstraction.

“In the present dataset, some verbal materials also involved richer semantic processing (e.g., sentences), which may further support abstraction.”

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

Increasing manipulation complexity, such as in executive n-back tasks, leads to broader transfer outcomes.

“This finding is also consistent with training findings showing that increasing manipulation complexity, such as in executive n-back tasks, leads to broader transfer outcomes (Cao et al., 2026).”

• cites Cao et al., 2026 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

The converging patterns raise the possibility that asymmetric transfer reflects a broader tendency for training that recruits more extensive or integrative processing to transfer more readily than training that recruits less extensive or integrative processing.

“These converging patterns raise the possibility that asymmetric transfer reflects a broader tendency for training that recruits more extensive or integrative processing to transfer more readily than training that recruits less extensive or integrative processing.”

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

Because the present meta-analysis focused on healthy populations, it remains unclear whether the observed asymmetry can be generalized to clinical groups.

“Second, because the present meta-analysis focused on healthy populations, it remains unclear whether the observed asymmetry can be generalized to clinical groups.”

Probably not a substantial claim: may not pass S1 (asserted), S3 (load-bearing).

A third possibility is that the observed asymmetry reflects differences along a storage-manipulation continuum such that training requiring active manipulation promotes broader transfer than training focused primarily on maintenance does.

“A third possibility, motivated by recent WM evidence, is that the observed asymmetry reflects differences along a storage–manipulation continuum such that training requiring active manipulation promotes broader transfer than training focused primarily on maintenance does, resulting in the transfer asymmetry.”

May combine two claims.

Overall, the moderator findings suggest that the transfer advantage is not only systematic but also context dependent.

“Overall, these moderator findings suggest that the transfer advantage is not only systematic but also context dependent.”

Probably not a substantial claim: may not pass S1 (asserted), S3 (load-bearing). May combine two claims.

Longer training may induce fatigue and reduce motivation, potentially offsetting any benefits of additional practice.

“Au et al. further proposed that longer training may induce fatigue and reduce motivation, potentially offsetting any benefits of additional practice.”

• cites Au et al., 2015Not matched

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

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

Roque-Gutierrez and Ibbotson (2023) reported an ability-based transfer advantage in the WM-linguistic domain, indicating that transfer asymmetry in higher-order cognitive training requires more study.

“Roque-Gutierrez and Ibbotson (2023), for example, reported an ability-based transfer advantage in the WM–linguistic domain, indicating that transfer asymmetry in higher-order cognitive training requires more study.”

• cites Roque-Gutierrez and Ibbotson, 2023 sourceNot checked — no full text of the source

May combine two claims.

Children and adolescents may have greater transfer potential because their cognitive systems remain under development, and prior meta-analytic findings show stronger cross-domain transfer from WM training in younger populations.

“This pattern is consistent with evidence that children and adolescents may have greater transfer potential because their cognitive systems remain under development (Liu et al., 2023), as well as with prior meta-analytic findings of stronger cross-domain transfer from WM training in younger populations (Sala et al., 2019b).”

• cites Liu et al., 2023 sourceNot checked — no full text of the source
• cites Sala et al., 2019bNot matched

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