Human Memories Can Be Linked by Temporal Proximity

Abstract

Real-world memories involve the integration of multiple events across time, yet the mechanisms underlying this integration is unknown. Recent rodent studies show that distinct memories encoded within a few hours, but not several days, share a common neural ensemble, and a common fate whereby later fear conditioning can transfer from one memory to the other. Here, we tested if distinct memories could be linked by temporal proximity in humans. 74 young adults encoded two memories (A and B) close (3-h) or far apart (7-day) in time. One day after encoding the second memory (B), Memory A was updated by pairing it with electric shock (i.e., fear conditioning). We tested whether the memory and fear associated with Memory B would be stronger in the 3-h, compared with the 7-day condition. Results were generally consistent with rodent studies, where we found heightened Memory B fear expression when the two memories were encoded close, but not far apart, in time. Furthermore, there was less forgetting of Memory B in the 3-h compared to 7-day condition. Our results suggest that temporally proximal memories may be linked, such that updating one experience updates the other.

Keywords: fear conditioning; memory linking; memory updating; neuronal ensemble; temporal proximity.

FIGURE 1

We investigated if memories encoded close in time are linked, such that manipulations of one (association of fear, memory improvement) generalized to the other. When A and B were encoded close in time (3-h) we expected higher fear response and less forgetting than when they were encoded 7 days apart.

FIGURE 2

Study design: the experiment consisted of two encoding sessions, a fear conditioning session and a final recognition test session, where memory for B items as well as fear response for A and B are measured. Context is represented by the colored shapes: green rectangle = context 1, yellow ellipse = context 2, gray hexagon = context 3 (novel). A and B refer to the content being tested (animals or tools dependent on counterbalance).

FIGURE 3

Fear response of CS+ vs. CS– trails for (A) skin conductance response, (B) evoked heart rate (as a beat per minute difference from baseline) and (C) fear potentiated startle response during fear acquisition session. CS: conditioned stimulus.

FIGURE 4

D-prime differences (final recognition test – immediate test) between 3-h and 7-day timing conditions at final test across contexts. Significantly higher memory in 3-h condition in the novel room, but not in the shock or encoding room. ^∗Statistically significant at the 0.05 level. ns: non-significant at the 0.05 level.

FIGURE 5

Group differences between 3-h and 7-day timing conditions at final test for (A) heart rate and (B) fear potentiated startle response. Significantly greater fear in Context 1 and 2 for heart rate and in the novel Context 3 for startle response. ^∗Statistically significant at the 0.05 level. ns: non-significant at the 0.05 level.

FIGURE 6

(A) Skin conductance response [muS], (B) heart rate [BPM], (C) startle response [uV/s], for memory A during fear conditioning correlated with skin conductance of Memory B at final test, across context. Note that when removing the potential skin conductance response outliers (i.e., those >1.25), significance remained for all but the difference between correlations in the novel room. FC, Fear Conditioning Session; FT, Final Test Session.