Distinct encoding and post-encoding representational formats contribute to episodic sequence memory formation

Abstract

In episodic encoding, an unfolding experience is rapidly transformed into a memory representation that binds separate episodic elements into a memory form to be later recollected. Here, we sought to investigate the representational format that accounts for successful memory formation of episodic sequences. We combined RSA and multivariate decoding analysis on scalp EEG recordings while healthy female (N = 17) and male (N = 13) human participants encoded trial-unique combinations of face-object-scene picture triplets that were subsequently recalled in a test. Our findings revealed an ongoing process of image category integration throughout sequence encoding, that this process accumulated as a function of picture order in the sequence and that the degree of picture category integration throughout sequence encoding was not associated with episodic retrieval. We also found that episodic sequences that were later remembered were associated with an increased trial-specific neural similarity between EEG patterns elicited during picture sequence encoding and in the offset period, thereby indicating that a rapid memory reinstatement at the episodic offset supported the rapid transformation of an episodic sequence into a bound memory representation. The present findings demonstrate the time course and the functional relevance of the different representational formats that take place during online encoding and immediate offline periods that support the formation of memories for episodic sequences.

Retrieval of human aversive memories involves reactivation of gamma activity patterns in the hippocampus that originate in the amygdala during encoding

“…These patterns of neuronal activity originating in the amygdala during encoding were replayed in the hippocampus during both the encoding, around 0.4 s after stimulus offset, and retrieval of the same information. The latency of encoding reactivation is consistent with findings from scalp EEG studies, showing a higher degree of reactivation of encoding patterns to pictures later successfully vs unsuccessfully recalled during the post-encoding period, around 0.5 s from stimulus offset 19,41 .…”

“…Conversely, there is growing evidence for putative alterations in memory representations within various brain regions and memory stages [14][15][16][17] . Thus, some questions remain regarding the interpretation of reactivation-related activity, such as whether BOLD signal measuring patterns of population activity could fail to isolate more transient [14][15][16][17][18][19] activity related to the content of memory. An additional question is whether reactivation patterns spread into different brain areas during a period of putative consolidation rather than remaining focal 17,20 .…”

Study-Phase Reinstatement

“…To test differences statistically, we fit the following statistical model: In addition to predicting reinstatement as a function of recall status, this model allowed for interactions with serial position in order to test the possibility that earlier serial positions showed greater effects of reinstatement, as suggested by the literature on overt rehearsal [25]. This analysis conceptually replicates past work on reinstatement (such as [2] and [40]) but applied to individual lists in a free recall task. As in the analyses of the inter-stimulus intervals, we find a main effect of recall status such that remembered items are more similar to the delay interval than forgotten items ( F (1, 28.43) = 30.12, p ≤.001.…”

Study-Phase Reinstatement: Encoding Spontaneous Thoughts as Memories

“…In addition to predicting reinstatement as a function of recall status, this model allowed for interactions with serial position in order to test the possibility that earlier serial positions showed greater effects of reinstatement, as suggested by the literature on overt rehearsal [25]. This analysis conceptually replicates past work on reinstatement (such as [2] and [42]) but applied to individual lists in a free recall task. As in the analyses of the inter-stimulus intervals, we find a main effect of recall status such that remembered items are more similar to the delay interval than forgotten items (F (1, 28.43) = 30.12, p ≤ .001.…”