Abstract
When freely recalling events from the past, individuals tend to successively remember stimuli that were studied close together in time—a phenomenon known as temporal clustering. Temporal clustering is thought to occur because stimuli are encoded in relation to a slowly-drifting internal context; this internal context representation is then reinstated during recall, leading to clustered recall of stimuli that share a similar internal context. While several lines of evidence implicate the hippocampus in supporting internal context representations, there is limited evidence directly linking hippocampal drift during memory encoding to subsequent temporal clustering during recall. In a human fMRI experiment (n=38), we sought to influence the rate of internal context change during memory encoding and tested for corresponding effects on (a) temporal clustering and (b) hippocampal drift rate. To influence internal context, we manipulated the rate at which background scenes ‘switched’ while a list of words was encoded. Afterwards, subjects freely recalled as many words as possible. while switch rate had no effect on the total number of words recalled, it significantly influenced the degree of temporal clustering. Specifically, a higher switch rate was associated with less temporal clustering. Strikingly, this pattern of data was mirrored by drift rate in the hippocampus: a higher switch rate was associated with significantly lower hippocampal autocorrelation (more drift). Moreover, individual differences in hippocampal autocorrelation were positively correlated with temporal clustering. Collectively, these findings suggest that hippocampal drift rate during encoding and temporal clustering during recall each reflect a common internal context representation.
Significance Statement The hippocampus is thought to support a gradually-drifting internal context representation that allows memories to be organized in time. This putative internal context representation helps explain the phenomenon of temporal clustering—that events encoded nearby in time are clustered together during recall. Yet, there is surprisingly limited evidence directly linking the drift of hippocampal activity patterns to the phenomenon of temporal clustering. Here, we show that manipulating the rate of external context change during memory encoding induced parallel changes in hippocampal drift rate during encoding and temporal clustering during subsequent recall. Critically, hippocampal drift rate also predicted the degree of temporal clustering across individuals. These findings suggest that hippocampal drift rate and temporal clustering reflect a common internal context representation.
Footnotes
This work is dedicated to Sarah DuBrow, who passed away before manuscript completion, but whose impact continues to be felt by many. We thank J. Benjamin Hutchinson for insightful comments and feedback, and members of the Kuhl lab for helpful discussions. This work was supported by NIH NINDS NS089729 to B.A.K. and by NIH-NINDS F31NS126016 to G.W.
The authors declare no conflicts of interest.





