Changelog · how the belief was versioned
Processing Difficulty Effect
The claim that making information slightly harder to process (e.g., a degraded or unusual font) triggers deeper, more analytic processing and thereby improves memory or reasoning relative to equally informative but easy-to-read material.
# README · how it works
The proposed mechanism (Alter et al. 2007) is metacognitive: experienced difficulty/disfluency acts as a cue or 'alarm' that one's quick, intuitive (System 1) response may be wrong, recruiting slow, deliberate (System 2) analytic processing. Applied to memory (Diemand-Yauman et al. 2011), disfluency is cast as a 'desirable difficulty' (Bjork 1994) that engenders deeper, more elaborative encoding and hence better retrieval.
Release stability — failed. The memory/learning version of the effect has largely failed to replicate. Xie, Zhou & Liu's (2018) meta-analysis (25 articles, 3,135 participants) found essentially no effect on recall (d = -0.01) or transfer (d = 0.03). Independent tests of the 'Sans Forgetica' memory font found no benefit (Huff et al. 2022; Geller et al. 2020). Multiple direct/conceptual replications (e.g., Eitel & Kühl 2016; Meyer et al. 2015 on math problems) failed. The reasoning/judgment version (Alter et al. 2007) is more mixed and the failures are accumulating: large replications of hard-to-read-font effects on the CRT (Meyer et al. 2015) and on the framing effect (Korbach et al. 2017, lab N=158 / online N=271) found no reliable disfluency benefit. Note the 2018 meta-analysis itself was later criticized for coding/calculation errors (Weissgerber, Brunmair & Rummer 2021), though that critique does not overturn the broadly null picture.
Issues filed in production
#001 open · 2018
RMIT University launches 'Sans Forgetica,' a font marketed to boost memory
In October 2018, RMIT University (Melbourne) and agency Naked Communications released Sans Forgetica, the first typeface designed around 'desirable difficulty,' claiming a ~7% memory boost from a study of ~400 students. It was downloaded widely and covered globally. Independent replications (Geller et al. 2020; Huff et al. 2022) subsequently found no memory benefit, and the font was effectively retired/decommissioned.
#002 open · 2015
Disfluent fonts do not help students solve math problems
Meyer et al. (2015) ran large studies (including a high-powered replication) testing whether hard-to-read fonts improve performance on the Cognitive Reflection Test and math problems, motivated by the Alter et al. (2007) and Diemand-Yauman et al. (2011) claims; they found no disfluency benefit, contributing to the documented failure of the effect to generalize to real educational tasks.
Recommended migration
Distinguish what is robust from what is not: disfluency reliably makes learning feel harder (lower judgments of learning, more time spent) but does NOT reliably improve actual recall or transfer. Before adding friction, demand a real outcome measure (delayed test performance, comprehension, decisions) rather than perceived effort, and prefer evidence-backed desirable difficulties such as retrieval practice, spacing, and the generation effect, which change objective task demands rather than just font legibility. Treat any single font study as a non-replicated anecdote given the near-zero meta-analytic effect.
Xie, Zhou & Liu (2018) found recall d = -0.01 and transfer d = 0.03 across 3,135 participants, while disfluency lowered judgments of learning (d = -0.43) and raised study time (d = 0.52); Sans Forgetica showed no benefit in independent tests.
Catch it in the act
You are about to deliberately make something harder to read, slower, or uglier on the theory that 'people will pay more attention' or 'it'll stick better.' Watch for treating effort-that-feels-hard as proof of learning, and for citing the famous Comic Sans / Sans Forgetica font studies as settled fact.
Dependencies
- Desirable difficultiesparentRobert Bjork's (1994) framework that conditions slowing or complicating learning can improve long-term retention; the processing-difficulty/disfluency effect was framed as one instance of it.
- Levels-of-processing (depth-of-processing) effectmechanistically-linkedCraik & Lockhart's idea that deeper, more semantic encoding improves memory; the proposed mechanism for disfluency benefits.
- Processing fluency / fluency heuristicoppositeThe well-replicated finding that easy-to-process information feels more true, likeable, and familiar; the processing-difficulty effect is essentially the reversed prediction for memory and reasoning.
- Generation effectsiblingA robust desirable-difficulty manipulation (generating vs reading answers improves memory) that, unlike disfluency, changes objective task difficulty, not just subjective fluency.
- Distinctiveness (von Restorff) effecteasily-confusedUnusual fonts may be remembered better because they are distinctive/novel rather than because they are disfluent; a key confound flagged for the memory studies.