The events of the COVID-19 Pandemic forced many psychologists to abandon lab-based approaches and embrace online experimental techniques. Although lab-based testing will always be the gold standard of experimental precision, several protocols have evolved to enable supervised online testing for paradigms that require direct observation and/or interaction with participants. However, many tasks can be completed online in an unsupervised way, reducing reliance on lab-based resources (e.g., personnel and equipment), increasing flexibility for families, and reducing participant anxiety and/or demand characteristics. The current project demonstrates the feasibility and utility of unsupervised online testing by incorporating a classic change-detection task that has been well-validated in previous lab-based research. In addition to serving as proof-of-concept, our results demonstrate that large online samples are quick and easy to acquire, facilitating novel research questions and speeding the dissemination of results. To accomplish this, we assessed visual working memory (VWM) in 4- to 10-year-old children in an unsupervised online change-detection task using arrays of 1–4 colored circles. Maximum capacity (max K) was calculated across the four array sizes for each child, and estimates were found to be on-par with previously published lab-based findings. Importantly, capacity estimates varied markedly across array size, with estimates derived from larger arrays systematically underestimating VWM capacity for our youngest participants. A linear mixed effect analysis (LME) confirmed this observation, revealing significant quadratic trends for 4- through 7-year-old children, with capacity estimates that initially increased with increasing array size and subsequently decreased, often resulting in estimates that were lower than those obtained from smaller arrays. Follow-up analyses demonstrated that these regressions may have been based on explicit guessing strategies for array sizes perceived too difficult to attempt for our youngest children. This suggests important interactions between VWM performance, age, and array size, and further suggests estimates such as optimal array size might capture both quantitative aspects of VWM performance and qualitative effects of attentional engagement/disengagement. Overall, findings suggest that unsupervised online testing of VWM produces reasonably good estimates and may afford many benefits over traditional lab-based testing, though efforts must be made to ensure task comprehension and compliance.
Proactive Interference Does Not Meaningfully Distort Visual Working Memory Capacity Estimates in the Canonical Change Detection Task
