Answers are grounded in this paper and cite their sources.
Memories are often reactivated in one context and only later judged with respect to where they originally occurred. We tested how an intervening retrieval context shapes neural representations and later source memory. In two fMRI tasks, participants (both male and female) encoded objects within contextual backgrounds, completed an immediate retrieval test in which object cues appeared against either the original background or a different background, and subsequently judged each object’s original context on a neutral screen. Shifted cueing contexts reduced source accuracy and biased errors toward the intervening context, despite no detectable difference in immediate recognition or associative-memory performance. Representational similarity analyses showed that encoding patterns reflected original-context structure, whereas immediate-cue-period patterns were broadly organized by the cued (i.e., currently visible) context. On shifted trials, activity patterns showed greater similarity to encoding-derived templates for the cued context than for the original context. Across participants, greater differentiation of original-context similarity between consistent and shifted cueing in posterior hippocampus, parahippocampal cortex, and dorsal anterior cingulate cortex was associated with better source memory after a contextual shift. Within participants, precuneus contextual engagement and dorsal anterior cingulate context alignment predicted later source choices, with similar relationships after adjustment for immediate recognition responses. Later source attribution therefore depended on the contextual state expressed when an item was re-encountered, not simply on reinstatement of the original event. Significance statement Remembering is often described as bringing up the past, but memories are frequently reactivated in new contexts before we judge where they came from. We show that this intervening context can bias later source attribution even when that context is no longer visible at final test. Using two fMRI tasks, representational similarity analysis and trial-level source-choice models, we found that cue-period activity was organized by the context present during cueing and that specific context-state signals predicted later memory choices, whereas item reinstatement did not. Source-memory errors can therefore reflect contextual information encountered after the original event, not only failures to recover the original context.
No takes yet. Share an insight, caveat, or question.
Hermiller et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: