The disfluency effect revisited: processing demands moderate memory performance

“Desirable difficulties” describe encoding conditions that enhance memory. Recently, this principle has been extended to perceptual difficulties, such as hard-to-read fonts. Although several studies report a memory benefit of disfluency, other findings fail to replicate this effect or even show reversals, consistent with a resource-sharing account. The present study examined whether processing demands moderate the disfluency effect. We hypothesised that under low processing demands, disfluent fonts engage additional encoding processes, whereas under high demands, cognitive resources are consumed by processing disfluency itself. Across three experiments, participants performed word classification tasks with easy-to-read or hard-to-read fonts followed by a free recall test. In Experiment 1, using a task-switching procedure, hard-to-read font did not significantly affect recall performance, although it was descriptively higher for easy-to-read font. In Experiment 2, a low-demanding single-task procedure revealed no effect of font. In Experiment 3, hard-to-read font benefited recall in a low-demanding two-option, but not in a high-demanding four-option task, suggesting that processing demands moderated the disfluency effect. Together, the results imply that under sufficiently low processing demands, a disfluent font can benefit memory, in line with the desirable difficulties account. However, the findings suggest that this benefit may be difficult to reliably elicit.