Sex differences in verbal working memory performance emerge at very high loads of common neuroimaging tasks

• Tested sex effects on behavior in 3 classic working memory (WM) neuroimaging tasks.
• Increased WM load above levels used in prior research that found no sex differences.
• Found sex differences within each task and for a cross-task WM construct/factor.
• Sex differences emerged at high loads even in groups matched at lower loads.
• Informs prior reports of WM-related sex differences in brain structure and function.

Working memory (WM) supports a broad range of intelligent cognition and has been the subject of rich cognitive and neural characterization. However, the highest ranges of WM have not been fully characterized, especially for verbal information. Tasks developed to test multiple levels of WM demand (load) currently predominate brain-based WM research. These tasks are typically used at loads that allow most healthy participants to perform well, which facilitates neuroimaging data collection. Critically, however, high performance at lower loads may obscure differences that emerge at higher loads. A key question not yet addressed at high loads concerns the effect of sex. Thoroughgoing investigation of high-load verbal WM is thus timely to test for potential hidden effects, and to provide behavioral context for effects of sex observed in WM-related brain structure and function. We tested 111 young adults, matched on genotype for the WM-associated COMT-Val108/158Met polymorphism, on three classic WM tasks using verbal information. Each task was tested at four WM loads, including higher loads than those used in previous studies of sex differences. All tasks loaded on a single factor, enabling comparison of verbal WM ability at a construct level. Results indicated sex effects at high loads across tasks and within each task, such that males had higher accuracy, even among groups that were matched for performance at lower loads.