CoChat
Sign Up
Cited by 13Cognitive Abilities and TestingNeural and Behavioral Psychology StudiesCognitive Functions and Memory

Which “working memory” are we talking about? Complex span tasks versus N-back

Open access
Burgoyne, Alexander P., Frank, David J., Macnamara, Brooke N.·2025·Psychonomic Bulletin & Review
DOI 10.3758/s13423-024-02622-0
Open this paper in CoChat — free accountOpen PDF

Abstract

Psychologists and neuroscientists often use complex span tasks or the n-back to measure working memory capacity. At first glance, both tasks require many cognitive processes attributed to the construct, including the maintenance of information amidst interference. Nevertheless, evidence for their convergent validity is mixed. This poses consequences for the interpretation of working memory performance in cognitive neuroscience, developmental psychology, applied psychology, and executive functioning research. We recruited a large and diverse sample using a multisite approach (N = 1,272; community and university participants) and had them complete multiple working memory capacity, updating, and fluid intelligence tests. We found strong evidence for a dissociation between complex span and n-back tests, and more broadly, between working memory capacity and updating factors. Observed correlations between complex span and n-back performance were modest (r̄ = .25), and at the latent level, the two factors only shared 20% of their variance. Each explained unique variance in fluid intelligence, and each was more strongly related to fluid intelligence than to each other, with updating measures demonstrating stronger relations to fluid intelligence. These results were interpreted via the disengagement hypothesis. What distinguishes updating measures from working memory capacity measures is their relative emphasis on disengagement from outdated information; disengagement drives their strong relation with fluid intelligence because problem-solving requires generating hypotheses but also discarding those discovered to be false. We suggest that researchers who want to measure and draw conclusions about working memory capacity or updating should not use complex span tasks and the n-back interchangeably.

AI snapshot

Written from the paper’s full text.

TL;DR

In a multisite sample (N = 1,272), complex span and n-back tasks showed only modest observed correlation (r̄ = .25) and latent factors shared 20% of their variance, indicating a dissociation. Updating/n-back measures related more strongly to fluid intelligence than complex span/working memory capacity measures, which the authors interpret via the disengagement hypothesis.

Key findings

  • Observed complex span–n-back correlations ranged from r = .17 to .32 (r̄ = .25); the two n-back tasks correlated r = .71, while complex span tasks intercorrelated r = .33 to .57 (r̄ = .43).
  • At the latent level, complex span and n-back factors correlated r = .45 [.39, .51], sharing 20.25% of variance; constraining this correlation to 1 worsened fit, Δχ²(1) = 514.61, p < .001. Broader working memory capacity and updating factors correlated r = .57 [.51, .62] (R² = 32.5%) and could not be set equal without loss in fit, Δχ²(1) = 412.52, p < .001.
  • Fluid intelligence correlated r = .66 [.61, .72] with complex span and r = .74 [.70, .78] with n-back, with n-back significantly stronger, Δχ²(1) = 5.22, p = .022. Broader updating correlated r = .84 [.81, .87] with fluid intelligence versus r = .76 [.71, .81] for working memory capacity, Δχ²(1) = 7.36, p = .007.
  • In SEM, both complex span (β = .42) and n-back (β = .55) explained unique variance in fluid intelligence; n-back explained significantly more unique variance, Δχ²(1) = 5.22, p = .022, and together they explained 67.9%. Broader updating (β = .60) explained more unique variance than working memory capacity (β = .40), Δχ²(1) = 7.36, p = .007, together explaining 80.2%.

Methods

  • Multisite individual-differences study with N = 1,272 university and community participants aged 18–39 from five universities and the greater Cleveland, Ohio, community.
  • Participants completed working memory capacity tasks (symmetry span, rotation span, reading span, letter-number sequencing), updating tasks (letter three-back, spatial three-back, tone monitoring, keep track), and fluid intelligence tasks (Raven’s Advanced Progressive Matrices, letter sets, number series, Cattell’s Culture Fair Test).
  • Analyses used exploratory factor analysis, confirmatory factor analysis, and structural equation modeling with maximum likelihood estimation, robust standard errors, and full information maximum likelihood for missing data; variables were standardized prior to estimation. Data cleaning began with 1,407 participants and retained 1,272 after exclusions.

Limitations

  • The paper states that only two n-back tests were administered (spatial and verbal), so the n-back latent factor may contain more method-specific variance than if three n-back tests had been used; a broader updating factor with tone monitoring and keep track was modeled to mitigate this.
  • The paper states that complex span tasks had lower internal consistency reliability (.74, .76, .70) than n-back tasks (.95, .95), and that reliability could be increased by adding more trials; however, adding letter-number sequencing and keep track/tone monitoring made reliabilities more balanced and increased the dissociation.
  • Evident from the text: reading span had a coding error that randomized set sizes, required an adapted scoring approach, had lower reliability (ω = .60), and showed the weakest correlations with n-back tasks; the broader CFA model including additional measures had less-than-adequate fit and a large residual variance for keep track (.89).

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

Integrity signals

Registry

Crossref · 10.3758/s13423-024-02622-0

Cites

39 of 39 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

10 of 44 cited sources checked · 9 supported · 0 flagged · accuracy not yet measured

  • Premisecitation

    Evidence on whether complex span and n-back tasks converge on the same ability is mixed.

    “Nevertheless, evidence for their convergent validity is mixed.”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites Schmiedek et al., 2009 source
    • cites Redick and Lindsey, 2013 source
  • Premisecitation

    Kane et al. (2007) found that correlations between n-back and operation span ranged from nonsignificant to weak, with all rs at or below .22.

    “Indeed, correlations between the n-back and operation span ranged from nonsignificant to weak (all rs ≤ .22).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Kane et al., 2007 source
  • Interpretationcitation

    Because reading span uses verbal memory items, its weak associations with n-back align with Redick and Lindsey's meta-analytic finding of weak n-back-complex span relations for verbal memoranda.

    “Reading span uses verbal memory items; as such, these results align with Redick and Lindsey's (2013) meta-analytic finding of relatively weak associations between n-back tasks and complex span that used verbal memoranda.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Redick and Lindsey, 2013 source
  • Findingown data

    Complex span and n-back were each more strongly related to fluid intelligence than to each other.

    “each was more strongly related to fluid intelligence than to each other”
    Quote in the paperSubstantial claim
  • Findingown data

    Among the three complex span tasks, reading span had the weakest correlation with the two n-back tasks, with rs of .17 and .18.

    “Of the three complex span tasks, reading span had the weakest correlation with the two n-back tasks (rs of .17 and .18).”
    Quote in the paperSubstantial claim
  • Classified as probably not a substantial claim. 24
  • May combine two claims. 36

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

    Interpretationcitation

    Both maintenance and disengagement matter for problem-solving, but their relative contribution could explain the different relations of complex span and n-back with fluid intelligence and their less-than-perfect correlation.

    “Both maintenance and disengagement are important for problem-solving (e.g., Carpenter et al., 1990), but the relative contribution of each could be driving the different relationships of complex span tests and the n-back with fluid intelligence, as well as their less-than-perfect correlation with each other (e.g., Kane et al., 2007).”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites Carpenter et al., 1990 source
    • cites Kane et al., 2007 source

      The Discussion (page unavailable in our copy) reports that n-back and complex span shared only 2%–5% of their variance and predicted largely independent variance in fluid intelligence. This supports their weak correlation and different relationships with intelligence. The paper does not attribute those differences specifically to the relative contributions of maintenance and disengagement; it instead discusses differences in retrieval and recognition demands. (LLM, conf=0.84)

      Experimental — accuracy not yet measured

  • Classified as probably not a substantial claim.

    Premisecitation

    Redick and Lindsey (2013) concluded that complex span and n-back tasks should not be used interchangeably.

    “They concluded that complex span and n-back tasks should not be used interchangeably.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Redick and Lindsey, 2013 source

  • May combine two claims.

    Premisecitation

    Kane et al. (2007) found that two-back performance did not correlate significantly with operation span or Raven's matrices, while three-back correlated descriptively more strongly with Raven's than with operation span.

    “Two-back performance did not correlate significantly with either measure, whereas three-back performance correlated descriptively more strongly with Raven's matrices than with operation span.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Kane et al., 2007 source

  • May combine two claims.

    Premisecitation

    Kane et al. (2007) found that operation span and n-back each accounted for unique variance in fluid intelligence, suggesting a dissociation.

    “Furthermore, operation span and the n-back each accounted for unique variance in fluid intelligence, suggesting a dissociation between the two measures.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Kane et al., 2007 source

  • May combine two claims.

    Premisecitation

    Redick and Lindsey's (2013) meta-analysis of 20 studies (N = 2,178) found that the complex span-n-back correlation was stronger for visuospatial than verbal items, with an overall weak meta-analytic correlation of .20.

    “Across 20 studies (total N = 2,178), they found that the strength of the correlation between complex span and n-back performance was stronger for visuospatial memory items than for verbal memory items.Overall, however, they observed only a relatively weak meta-analytic correlation, r̄ = .20.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Redick and Lindsey, 2013 source

  • Classified as probably not a substantial claim.

    Premisecitation

    In Friedman and Miyake's (2017) model of inhibition, complex span and updating tasks are described as measuring closely related constructs.

    “Though in Friedman and Miyake's (2017) model of inhibition, complex span and updating tasks are described as measuring closely related constructs”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Friedman and Miyake's (2017) source

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

    Premisecitation

    Kane et al. (2007) identified recall versus recognition as one candidate source of variance distinguishing these measures.

    “Kane et al. (2007) identified recall versus recognition as one candidate.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Kane et al., 2007 source

  • May combine two claims.

    Findingown data

    The study found strong evidence for a dissociation between complex span and n-back tests, and more broadly between working memory capacity and updating factors.

    “We found strong evidence for a dissociation between complex span and n-back tests, and more broadly, between working memory capacity and updating factors.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Both the n-back and the complex span factors were significantly more strongly related to fluid intelligence than to each other, Δχ2(1) = 92.52 and Δχ2(1) = 51.60, both p < .001.

    “Furthermore, both the n-back and complex span factors were significantly more strongly related to fluid intelligence than to each other, ∆χ 2 (1) = 92.52,p < .001,and ∆χ 2 (1) = 51.60,p < .001,respectively.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    In that factor analysis, the two n-back tasks loaded .84 and .84 on the first factor, and the three complex span tasks loaded .82, .69, and .47 on the second factor, with negligible cross-loadings.

    “The two n-back tasks loaded highly on the first factor (loadings of .84 and .84)and the three complex span tasks loaded highly on the second factor (loadings of .82,.69,and .47[reading span]).Cross-loadings were negligible (range: -.02 to .03).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    In the structural equation model, both the complex span (β = .42) and the n-back (β = .55) regression paths predicting fluid intelligence were statistically significant, indicating that each factor explained unique variance.

    “Both regression paths were statistically significant (complex span β = .42;n-back β = .55),indicating that both factors explained unique variance.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    In the study sample, the complex span and n-back latent factors correlated r = .57 [.51, .62] (R2 = 32.5%) and could not be constrained to equality without a significant loss of fit, Δχ2(1) = 412.52, p < .001.

    “We found that the two factors correlated r = .57[.51, .62](R 2 = 32.5%) and could not be set equal to each another without loss in fit, ∆χ 2 (1) = 412.52,p < .001.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The three complex span tasks had intercorrelations from .33 to .57 (mean .43), while the two n-back tasks correlated .71 with each other.

    “The three complex span tasks had intercorrelations ranging from r = .33to r = .57(r̄ = .43),whereas the two n-back tasks correlated r = .71with each another.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Among the 1,272 university and community participants, updating was strongly related to fluid intelligence (r = .84) and, to a lesser extent, working memory capacity (r = .61).

    “Our results from 1,272 university and community participants revealed that updating was strongly related to fluid intelligence (r = .84)and, to a lesser extent, working memory”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The confirmatory factor analysis including additional working memory capacity and updating measures fit less than adequately, χ2(19) = 248.09, p < .001, CFI = .919, TLI = .880, RMSEA = .097, 90% CI [.087, .108], SRMR = .064, with a large residual variance for keep track (.89) that was retained on conceptual grounds.

    “Model fit was less than adequate: χ 2 (19) = 248.09,p < .001;CFI = .919,TLI = .880,RMSEA = .097,90% CI [.087, .108],SRMR = .064;there was a large residual variance for keep track (.89), but it was retained on conceptual grounds as a measure of updating.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    Complex span tasks and the n-back are two commonly used task types for assessing individual differences in working memory capacity.

    “Two types of tasks are frequently used to measure individual differences in working memory capacity-complex span tasks and the n-back”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites Engle et al., 1999 source
    • cites Owen et al., 2005 source

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

    Premisecitation

    According to the executive attention view, the attentional component primarily drives the predictive validity of working memory capacity measures.

    “it is primarily the attentional component that drives the predictive validity of working memory capacity measures (e.g., Daneman & Carpenter, 1980;Engle et al., 1999).”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites Daneman & Carpenter, 1980 source
    • cites Engle et al., 1999 source

  • May combine two claims.

    Premisecitation

    In complex span tasks, attention supports maintaining information amid distraction and interference, and this is a major source of individual differences.

    “Within the context of complex span tasks, attention supports the maintenance of information amidst distraction and interference, and this is a major source of individual differences in performance.”
    Quote in the paperSubstantial claim

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

    Study design and conductown data

    A sensitivity check using scores before data cleaning found the same pattern of significant results.

    “As a sensitivity check, we also analyzed scores prior to data cleaning and found the same pattern of significant results.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisereasoning

    In n-back tasks, memory set size is typically small; maintenance is necessary, but individual differences are hypothesized to stem from disengagement from outdated items.

    “Unlike complex span tasks, the memory set size used in n-back tasks is typically small (e.g., two or three items); maintenance is necessary, but the source of individual differences in performance is hypothesized to stem from differences in the ability to disengage from those outdated items that make one susceptible to far lures.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Schmiedek et al. (2009) found a latent correlation of .96 between complex span and updating factors, which was not significantly different from 1.0.

    “Using latent variable modeling, they found a correlation of r = .96between factors representing complex span and updating performance-this correlation was not significantly different from 1.0.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Schmiedek et al., 2009 source

  • May combine two claims.

    Premisecitation

    In n-back tasks, subjects must rapidly update working memory to disengage from outdated information in addition to maintaining n items.

    “One challenge is that, in addition to having to maintain n items, subjects must also rapidly update the contents of working memory to disengage from outdated information (Oberauer, 2009;Shipstead et al., 2016;Szmalec et al., 2011)”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites Oberauer, 2009 source
    • cites Shipstead et al., 2016 source
    • cites Szmalec et al., 2011 source

  • May combine two claims.

    Premisenone given

    In fluid intelligence tests, subjects must generate hypotheses and discard incorrect ones; failure to disengage leads to perseveration and impedes problem-solving.

    “When subjects attempt to solve novel problems in fluid intelligence tests, they must generate hypotheses and discard incorrect hypotheses.An inability to disengage from outdated hypotheses leads to perseveration and stymies problem-solving.”
    Quote in the paperSubstantial claim

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

    Premisecitation

    Complex span tasks have often been used to measure working memory capacity in cognitive psychology.

    “In the decades that followed, complex span tasks have been frequently used to measure working memory capacity in cognitive psychology (Engle et al., 1999;Kyllonen & Christal, 1990;Miyake et al., 2001).”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites Engle et al., 1999 source
    • cites Kyllonen & Christal, 1990 source
    • cites Miyake et al., 2001 source

  • Classified as probably not a substantial claim.

    Premisecitation

    The disengagement hypothesis treats maintenance and disengagement as two critical factors linking executive functioning to fluid intelligence.

    “This framework views maintenance and disengagement as two critical factors linking executive functioning to fluid intelligence (Burgoyne & Engle, 2020;Burgoyne et al., 2019;Hambrick & Altmann, 2015;Shipstead et al., 2016).”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites Burgoyne & Engle, 2020 source
    • cites Burgoyne et al., 2019 source
    • cites Shipstead et al., 2016 source

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

    Premisenone given

    Attention supports updating by removing no-longer relevant memory items from focus.

    “Attention is also hypothesized to support performance on updating tasks; however, its primary function is to remove no-longer relevant memory items from focus.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    In Schmiedek et al. (2009), two of three complex span measures had weak, nonsignificant correlations with n-back performance, yet the latent constructs were perfectly correlated.

    “Notably, two of the three complex span measures had weak, nonsignificant correlations with n-back performance, yet at the latent level, the constructs were perfectly correlated.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Schmiedek et al., 2009 source

  • May combine two claims.

    Interpretationown data

    Reading span had lower reliability than the other two complex span tests, which could have attenuated its correlations.

    “Reading span also had lower reliability than the other two complex span tests, which could have attenuated its correlations.”
    Quote in the paperSubstantial claim

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

    Study design and conductown data

    The final analytical sample consisted of 1,272 participants retained after data cleaning.

    “The analyses reported below are based on the analytical sample of 1,272 participants who were retained throughout the data cleaning procedure.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    Shipstead et al. (2016) found that the negative correlation between fluid intelligence and n-back false alarms strengthened as lure position increased, indicating that lower fluid intelligence individuals were more likely to mistake far lures as targets.

    “Shipstead et al. (2016) found that the negative correlation between fluid intelligence and n-back false alarms strengthened as the lure position increased, indicating that individuals with lower fluid intelligence were more likely to mistake far lures as targets.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Shipstead et al., 2016 source

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

    Study design and conductown data

    Scores more than three standard deviations worse than the mean were removed, with 190 observations removed.

    “Scores that were three standard deviations worse than the mean were removed (n obs.= 190).”
    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 conductreasoning

    Keep track was used as an updating indicator because it requires rapid disengagement, and the manuscript's conclusions are robust to excluding it.

    “In the following analyses, keep track served as an indicator of updating because it requires rapid disengagement, but we note that the conclusions of the manuscript are robust to its exclusion.”
    Quote in the paperNot a substantial claim

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

    Study design and conductown data

    Participants who were noncompliant during the study were removed from the data, with n = 123.

    “We removed participants' data if they were noncompliant during the study (e.g., using their phone while completing a task; n subjects = 123).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    The study provides strong evidence for a dissociation between complex span and n-back tasks and, more generally, between working memory capacity and updating measures.

    “we found strong evidence for a dissociation between complex span and n-back tasks, and more generally, between working memory capacity and updating measures.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The latent factors estimated from the task batteries validly represent the underlying constructs of working memory capacity, updating, and fluid intelligence.

  • May combine two claims.

    Study design and conductreasoning

    The study's n-back tasks contained as many lure trials as target trials, so relying on mere familiarity would not be an effective strategy because it would produce a high false alarm rate.

    “Although recognition could support performance in some n-back tasks, our n-back tasks had as many lure trials as target trials; relying on mere familiarity would not be an effective strategy as it would result in a high false alarm rate.”
    Quote in the paperSubstantial claim

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

    Study design and conductreasoning

    This study was not preregistered, but it followed the same data preparation and cleaning procedures as the larger project's preregistration.

    “This study was not preregistered, but we followed the same data preparation and cleaning procedures as outlined in the preregistration.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisenone given

    Working memory capacity measures such as complex span tasks emphasize maintenance, and attention control serves to protect memory items from interference caused by the secondary task.

    “Working memory capacity measures (e.g., complex span tasks) emphasize maintenance; the role of attention control is to protect memory items from the consequences of interference from the secondary task.”
    Quote in the paperSubstantial claim

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

    Interpretationnone given

    These tasks may differ not only in their affordance of recall versus recognition but also in their relative demands on maintenance versus disengagement.

    “These tasks may differ not only in their affordance of recall versus recognition, but also in their relative demands on maintenance versus disengagement.”
    Quote in the paperSubstantial claim

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

    Interpretationreasoning

    Alternative theoretical accounts are also consistent with the study's results.

    “Nevertheless, there are alternative theoretical accounts that are also consistent with our results.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    The broader updating factor incorporating tone monitoring and keep track likely reduced the influence of common method variance, and models using this factor showed stronger evidence for a dissociation than models with only complex span and n-back tasks.

    “The influence of common method variance was likely reduced in this factor, and yet models using this factor revealed stronger evidence for a dissociation than the models that included just complex span and n-back tasks.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationcitation

    In Frischkorn et al.'s study, the nonupdating indicators were only weakly correlated with one another (rs of .16, .20, and .27), two showed ceiling effects (accuracy of 95% and 91%), and they had imbalanced and generally low loadings (.6, .3, .3) on a common factor.

    “the “nonupdating” indicators were only weakly correlated with each other (rs of .16,.20,and .27),two suggested ceiling effects (accuracy of 95%”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Frischkorn et al. (2022) source

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

    Premisereasoning

    Unreliability reduces the shared variance among measures and could have consequences for the latent factor intended to capture that common variance.

    “unreliability reduces the shared variance among measures and could have consequences for the latent factor attempting to capture this common variance.”
    Quote in the paperNot a substantial claim

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

    Premisecitation

    Frischkorn et al. (2022) analyzed a sample of 111 young adults and concluded that updating-specific variance was not related to fluid intelligence or working memory capacity.

    “These findings stand in contrast to the results of Frischkorn et al. (2022), who analyzed a sample of 111 young adults and concluded that updating-specific variance was not related to fluid intelligence or working memory capacity.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Frischkorn et al. (2022) source

  • May combine two claims.

    Interpretationcitation

    In Frischkorn et al.'s study, the loadings of the updating-specific indicators were not credibly different from zero and the best-fitting models did not include updating-specific variance.

    “Across several models, the loadings for the updating-specific indicators were not credibly different from zero, and the bestfitting models did not include updating-specific variance.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Frischkorn et al. (2022) source

  • May combine two claims.

    Interpretationown data

    Because updating correlated with fluid intelligence more strongly than working memory capacity did, disengagement may be especially crucial for problem-solving.

    “Our structural equation models demonstrated that updating correlated with fluid intelligence more strongly than working memory capacity, suggesting that disengagement may be especially crucial for problem-solving.”
    Quote in the paperSubstantial claim

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

    Study design and conductnone given

    The study was approved by each university's Institutional Review Board.

    “This study was approved by each university's Institutional Review Board.”
    Quote in the paperNot a substantial claim

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

    Interpretationcitation

    The general principles of maintenance and disengagement can be incorporated into other working memory frameworks, including the binding hypothesis and Cowan's embedded-process model.

    “the general principles of maintenance and disengagement can be incorporated into other working memory frameworks, including the binding hypothesis as well as Cowan's embedded-process model (Cowan et al., 2020).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Cowan et al., 2020 source

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

    Study design and conductnone given

    This study was part of a larger project that was pre-registered; this study itself was not preregistered, but the same data preparation and cleaning procedures as outlined in the preregistration were followed.

    “This study was part of a larger project that was pre-registered (https:// osf. io/ c39y6).This study was not preregistered, but we followed the same data preparation and cleaning procedures as outlined in the preregistration.”
    Quote in the paperNot a substantial claim

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

    Premisecitation

    The binding hypothesis views the working memory system as reflecting the ability to rapidly build, maintain, and update bindings or relations between memory items.

    “the “binding hypothesis” (e.g., Wilhelm et al., 2013) views the working memory system as reflecting the ability to rapidly build, maintain, and update bindings or relations between memory items.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Wilhelm et al., 2013 source

      Not checked in the free run.

  • May combine two claims.

    Study design and conductreasoning

    Only two n-back tests were administered, one with spatial and one with verbal memory items, so the n-back latent factor may contain more method-specific variance than if three n-back tests had been used.

    “The first limitation of this work is that we administered only two n-back tests, one with spatial memory items and another with verbal memory items.As a result, our n-back latent factor may contain more method-specific variance than if we had administered three n-back tests.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    Updating measures such as the n-back emphasize disengagement, and attention control serves to remove outdated items from focus.

    “In contrast, updating measures (e.g., n-back) emphasize disengagement; the role of attention control is to remove outdated items from focus (Shipstead et al., 2016).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Shipstead et al., 2016 source

  • May combine two claims.

    Premisecitation

    In n-back tasks subjects make continuous recognition judgments, whereas in complex span tasks subjects attempt to recall memory items at the end of each trial.

    “In the n-back, subjects make continuous recognition judgments, whereas in complex span tasks, subjects attempt to recall memory items at the end of each trial.”
    Quote in the paperSubstantial claim

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

    Interpretationreasoning

    When problem-solving, the ability to generate, test, and keep track of hypotheses depends on maintaining that information despite interference.

    “When problem-solving, one's ability to generate, test, and keep track of hypotheses depends on the ability to maintain this information despite interference.”
    Quote in the paperNot a substantial claim
53 of 53

  • The paper relies on this without support.

    Complex span and n-back tasks are ostensibly measures of the same construct, so their relation can be evaluated as convergent validity.

    “demonstrating poor convergent validity for two sets of tasks that ostensibly measure the same construct.”

  • The paper relies on this without support.

    The latent factors estimated from the task batteries validly represent the underlying constructs of working memory capacity, updating, and fluid intelligence.

    “Using structural equation modeling, we specified the complex span and n-back factors as correlated predictors of fluid intelligence (Fig. 7).”
2 of 2

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?

  • 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).”

    From The asymmetry of working memory training transfer: A systematic review and meta-analysis (2026) · cites this paper as Burgoyne et al., 2025

    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

  • 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).”

    From The asymmetry of working memory training transfer: A systematic review and meta-analysis (2026) · cites this paper as Burgoyne et al., 2025

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.

  • Consistentχ²(5) = 11.7reported p = .039 · recomputed p = .039

    “χ 2 (5) = 11.70,p = .039”

  • Consistentχ²(1) = 2.54reported p = .111 · recomputed p = .111

    “χ 2 (1) = 2.54, p = .111”

  • Consistentχ²(13) = 89.88reported p < .001 · recomputed p < .001

    “χ 2 (13) = 89.88,p < .001”

  • Consistentχ²(4) = 1.93reported p = .748 · recomputed p = .749

    “χ 2 (4) = 1.93, p = .748”

  • Consistentχ²(19) = 248.09reported p < .001 · recomputed p < .001

    “χ 2 (19) = 248.09,p < .001”

  • Consistentχ²(24) = 62.07reported p < .001 · recomputed p < .001

    “χ 2 (24) = 62.07, p < .001”

  • Consistentχ²(51) = 450.56reported p < .001 · recomputed p < .001

    “χ 2 (51) = 450.56,p < .001”

7 of 7

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.

  • Three fluid-intelligence tasks are said to comprise four tests

    Minor

    The text calls these three tasks but lists four: Raven's matrices, letter sets, number series, and Cattell's test.

  • Keep-track and tone-monitoring reliabilities are reversed

    Minor

    The limitations section assigns .92 to keep track and .59 to tone monitoring; Table 1 assigns .59 to keep track and .92 to tone monitoring.

  • Reading-span reliability differs from Table 1

    Minor

    The limitations section gives .70 as the third complex-span reliability, but Table 1 reports .60 for reading span.

3 of 3

Cites

Flagged first, then by citations

3 of the 42 references couldn’t be matched to a known paper.

Research works printed but not linked: 2

  1. Ekstrom, R. B., French, J. W., Harman, H. H., & Dermen, D. (1976). Manual for kit of factor-referenced cognitive tests: 1976. Educational Testing Service.
  2. Raven, J. C., & Court, J. H. (1998). Raven’s progressive matrices and vocabulary scales (Vol. 759). Oxford Psychologists Press.

Cited by

Sorted by citations

Related

Authors
Burgoyne, A. P., Frank, D. J., & Macnamara, B. N.
Affiliations
  • Georgia Institute of Technology (US)
  • Youngstown State University (US)
  • Purdue University West Lafayette (US)
Corresponding author
Burgoyne, Alexander P.
ORCID
Venue
Psychonomic Bulletin & Review
Publisher
Springer Science and Business Media LLC
Published
2024-12-04
Volume
32
Issue
3
Pages
1337-1351
ISSN
  • 1069-9384 (ISSN-L)
  • 1531-5320
DOI
10.3758/s13423-024-02622-0
OpenAlex
W4405023182
Cited by
12
References
42
Citations by year
2026: 7 · 2025: 4 · 2024: 1
FWCI
5.15
Type
journal article
Language
en
Open access
Licence
Locations
  • Psychonomic Bulletin & Review (publishedVersion)
  • PubMed (publishedVersion)
Topics
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Cognitive Abilities and Testing
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Neural and Behavioral Psychology Studies
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Cognitive Functions and Memory
Keywords
working memory capacity, complex span, N-back, updating, fluid intelligence, convergent validity, information maintenance, interference
MeSH terms
Adolescent, Adult, Female, Humans, Intelligence, Male, Memory, Short-Term

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