The absence of voluntary visual imagery, termed aphantasia, offers a unique lens to investigate the role of visual imagery in various cognitive processes, including face recognition. Given its low prevalence (2–4%), many studies increasingly rely on online experiments to recruit adequate sample sizes. In this study, we compared two age- and gender-matched cohorts of aphantasics and typical imagers using a face recognition paradigm with one cohort tested in-lab and the other online. Findings based on recognition accuracy reveal differing results between the cohorts. While the lab-based sample shows significant benefits for imagery in recognition accuracy, particularly when faces were presented inversely, these benefits were less pronounced in the online cohort. To explore this further, we fitted a hierarchical drift diffusion model to our data, allowing us to examine decision-making processes in greater detail. Across collapsed cohorts, we observe more decisive evidence accumulation, i.e., higher drift rates, for the imagery group. Additionally, intra-cohort differences emerged in boundary separation, reflecting the degree of caution exhibited during decision-making. Imagers tested in the controlled lab-environment, along with aphantasics from both online and lab settings, demonstrated consistent decision-making patterns. In contrast, the online imagery group relied on smaller amounts of evidence, resulting in shorter response times and lower recognition accuracy compared to the other three groups. our findings highlight the importance of addressing motivational factors influencing participation commitment and implementing methodological adaptations when conducting online research, particularly in the context of neurodivergent populations, and ensuring the appropriate selection of control groups.