Familiarity From Discrepancy: Temporal Constraints on Relative Fluency in Recognition Memory.

Processing fluency is widely assumed to influence recognition memory through attributional mechanisms. Relative fluency accounts propose that familiarity arises not from processing ease per se, but from the discrepancy between expected and experienced fluency. However, the temporal locus of this discrepancy remains unclear: must expected non-fluency coincide with target processing to shape recognition judgments, or can degradation effects be explained solely by compensatory retrieval processes? Across two experiments, we manipulated perceptual degradation and its locus relative to the target stimulus. In Experiment 1, degradation increased false recognition when applied to target words but reduced false recognition when applied to contextual primes, revealing a locus-dependent crossover interaction. Experiment 2 dissociated perceptual degradation from retrieval availability by manipulating whether new words had studied associates. Contrary to a retrieval account, degradation produced the largest increase in false recognition when no associate was available. Across studies, degradation reduced correct recognition of old words but selectively amplified false recognition when temporally aligned with target evaluation. These findings challenge a pure compensatory retrieval explanation and support a temporally constrained relative fluency account. Familiarity appears to emerge from the attribution of prediction–perception discrepancies occurring at the moment of target processing, rather than from retrieval or absolute fluency alone.