Dissociable episodic memory performance and subjective experience across the aphantasia imagery spectrum

Mental imagery is often assumed to support episodic memory, yet whether this relationship depends on conscious access to imagery, and how reduced imagery differs from total aphantasia, remains unclear. The present study compared objective episodic memory performance, subjective trial-by-trial vividness, confidence, and response times in people reporting normal imagery, reduced imagery, or total aphantasia, across younger and older age groups. Participants completed an incidental-encoding episodic memory task requiring identification of visual or spatial changes in targets within scenes, alongside trial-by-trial vividness and confidence ratings. Across the full imagery spectrum, trial-level fluctuations in vividness were related to memory performance, whereas trait-level and averaged vividness measures were not. Moreover, reduced imagery and total aphantasia showed dissociable patterns of episodic memory performance and subjective experience. In the reduced-imagery group, vividness was a stronger predictor of accuracy and response times, vividness judgements were slower, and confidence was lower, suggesting greater metacognitive uncertainty when imagery is weak and inconsistent. Controls outperformed reduced and aphantasic groups on visual but not spatial changes, and both groups responded more slowly than controls; confidence did not differ between control and aphantasic groups. Age-related memory decline was observed in the control and reduced groups but not in the aphantasic group. Exit questionnaire data highlighted limited awareness of aphantasia among typical imagers and a persistent belief among aphantasic participants that their memory was inferior. Together, these findings show that reduced imagery should be distinguished from total aphantasia rather than treating both as a single population, and that trial-level fluctuations in vividness reveal relationships with episodic memory that trait-level and averaged measures do not capture.