Lotta Pesonen¹ (), Jozsef Fiser¹, Máté Lengyel^1,2; ¹Central European University, ²University of Cambridge

Understanding the relationship between statistical learning and recognition memory is essential for explaining how environmental input is encoded and recalled. While statistical learning has been extensively studied in visual paradigms, recognition memory research has typically involved semantically rich stimuli such as words and images of natural scenes. The present study bridges these fields by examining how previous findings in recognition memory generalise to paradigms where simple, non-semantic visual stimuli is used. In seven experiments, participants completed recognition tasks with shape stimuli. In the item recognition task (n = 150), participants distinguished between previously viewed shapes and novel lures. In the association recognition task (n = 200), participants judged whether pairs of shapes were previously viewed or recombined from familiar shapes. Unlike traditional paradigms, our study used a small stimulus set and a single familiarization-test block. The stimulus presentation times during familiarization and inventory sizes were varied between participants. Prior research suggests item recognition depends on both familiarity and recollection, while association judgments rely mainly on recollection. Our findings confirm this for item recognition but reveal that associative memory judgments are primarily driven by familiarity, contrary to prior studies. This challenges earlier models, indicating that recognition processes are influenced not only by task type but also by stimulus characteristics and experimental design. Moreover, reducing the study list length in the association task further amplified familiarity’s role, contradicting previous claims linking list length solely to recollection. These results underscore the impact of contextual factors on recognition memory and suggest a more prominent role for familiarity in tasks involving non-semantic stimuli.