A Latent Factor Analysis of Working Memory Measures Using Large-Scale Data

Abstract

Working memory (WM) is a key cognitive system that is strongly related to other cognitive domains and relevant for everyday life. However, the structure of WM is yet to be determined. A number of WM models have been put forth especially by factor analytical studies. In broad terms, these models vary by their emphasis on WM contents (e.g., visuospatial, verbal) vs. WM processes (e.g., maintenance, updating) as critical, dissociable elements. Here we conducted confirmatory and exploratory factor analyses on a broad set of WM tasks, half of them numerical-verbal and half of them visuospatial, representing four commonly used task paradigms: simple span, complex span, running memory, and n-back. The tasks were selected to allow the detection of both content-based (visuospatial, numerical-verbal) and process-based (maintenance, updating) divisions. The data were collected online which allowed the recruitment of a large and demographically diverse sample of adults (n = 711). Both factor analytical methods pointed to a clear division according to task content for all paradigms except n-back, while there was no indication for a process-based division. Besides the content-based division, confirmatory factor analyses supported a model that also included a general WM factor. The n-back tasks had the highest loadings on the general factor, suggesting that this factor reflected high-level cognitive resources such as executive functioning and fluid intelligence that are engaged with all WM tasks, and possibly even more so with the n-back. Together with earlier findings that indicate high variability of process-based WM divisions, we conclude that the most robust division of WM is along its contents (visuospatial vs. numerical-verbal), rather than along its hypothetical subprocesses.

Δ9-Tetrahydrocannabinol (THC) impairs visual working memory performance: A randomized crossover trial

“…Here, we used the "Discrete Whole-Report" task, a visuospatial working memory task in which participants encode many items simultaneously and recall them at test. This and similar visuospatial working memory tasks are reliable [32,56], correlate well with other measures of working memory [46,47], and predict individual differences in general fluid intelligence [47]. Thus, we believe that visuospatial working memory tasks (i.e., whole-report, partial-report, and change detection) are well-suited for characterizing deficits in the ability to temporarily store information in mind.…”

“…Most critically, based on our down-sampling analysis, we believe that the very short task length of digit span (forward and backward conditions, <5 min) critically reduces its sensitivity. In addition, simple span tasks (e.g., the forward digit span condition) load onto a general working memory factor at the latent level [46] relatively weakly, and typically do not predict individual differences in general fluid intelligence [47][48][49][50][51], but also see [52]. N-back tasks correlate somewhat weakly with other working memory tasks [53][54][55] and have relatively poor statistical reliability [51], potentially making it difficult to detect effects across treatment conditions.…”

“…N-back tasks correlate somewhat weakly with other working memory tasks [53][54][55] and have relatively poor statistical reliability [51], potentially making it difficult to detect effects across treatment conditions. On the other hand, n-back tasks still load well onto a general WM factor at the latent level [46] and are particularly useful for investigating the executive function component of WM (vs. the storage component).…”

Is Bilingualism Associated with Enhanced Executive Functioning in Adults? A Meta-Analytic Review_Pre-Print

“…Paap & Sawi, 2014; Waris et al, 2017). Also, for many EF tasks, validity information is 14" completely lacking.…”

Bilingualism and working memory performance: Evidence from a large-scale online study

“…In the present study, we set out to investigate the associations between bilingualism and WM in large, diverse, and well-matched groups of monolinguals, early bilinguals, and late bilinguals. We measured WM with three composite scores that reflect the previously reported latent structure of WM in this sample [ 43 ]. To the best of our knowledge, this is the first study that combines the use of an extensive WM task battery and composite scores based on the latent structure of the WM tasks in a study of BEA.…”