Time-resolved representational similarity analysis reveals integrated and separated neural patterns of overlapping events
Liu, Zhenghao LU ; Johansson, Mikael LU
and Bramao, Ines
LU
(2023)
International Conference on Learning and Memory
- Abstract
- Episodic memory allows the flexible retrieval of commonalities and idiosyncrasies of overlapping life events. For example, seeing a woman in the city with your colleague's daughter may form an integrated memory representation involving the woman and your colleague. However, you may also keep a specific representation of the city event to talk with your colleague about the circumstances of having met her daughter. The present study investigates how overlapping events are coded in the brain to support integrated and separated memory representations. In particular, we examined if integrated representations can be formed without losing episodic details for the individual events, i.e., information required for source memory judgments.... (More)
- Episodic memory allows the flexible retrieval of commonalities and idiosyncrasies of overlapping life events. For example, seeing a woman in the city with your colleague's daughter may form an integrated memory representation involving the woman and your colleague. However, you may also keep a specific representation of the city event to talk with your colleague about the circumstances of having met her daughter. The present study investigates how overlapping events are coded in the brain to support integrated and separated memory representations. In particular, we examined if integrated representations can be formed without losing episodic details for the individual events, i.e., information required for source memory judgments. Participants (N = 39) were presented with Sims videos simulating real-life events while their electroencephalograph (EEG) was recorded. First, they encountered videos of two interacting characters (person A and B). These were followed by a new set of videos showing a new character interacting with an old one (person C and B). Participants were asked to integrate the overlapping events to infer the indirect relationships between the characters (i.e., AC) and to keep information about the event-specific relationships (i.e., AB and CB). In a final test, we measured memory for all associations. Source memory was tested by asking participants if the two characters had directly interacted with each other. Participants’ source memory accuracy was higher when they also correctly inferred the indirect relationships across episodes. Representational similarity analysis (RSA), used to investigate how the brain represents overlapping episodes, showed that the AB and CB episodes were characterized by systematic neural pattern similarities and dissimilarities. Importantly, the neural similarities observed were predictive of both future inference and memory for the source. The data show that neural pattern similarities between overlapping episodes support memory integration across event boundaries, while pattern dissimilarities may separate events. This study provides novel evidence that integrated and separated representations may coexist to support multiple memory functions. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/3c8d5285-b658-4993-8451-f2feb55ca1f1
- author
- Liu, Zhenghao
LU
; Johansson, Mikael
LU
and Bramao, Ines
LU
- organization
- publishing date
- 2023
- type
- Contribution to conference
- publication status
- published
- subject
- conference name
- International Conference on Learning and Memory
- conference dates
- 2023-04-26 - 2023-04-30
- project
- When remembering affects new learning: Temporal dynamics of memory integration revealed by EEG and machine learning techniques
- language
- English
- LU publication?
- yes
- id
- 3c8d5285-b658-4993-8451-f2feb55ca1f1
- date added to LUP
- 2025-03-20 13:03:51
- date last changed
- 2025-05-06 03:05:47
@misc{3c8d5285-b658-4993-8451-f2feb55ca1f1,
abstract = {{Episodic memory allows the flexible retrieval of commonalities and idiosyncrasies of overlapping life events. For example, seeing a woman in the city with your colleague's daughter may form an integrated memory representation involving the woman and your colleague. However, you may also keep a specific representation of the city event to talk with your colleague about the circumstances of having met her daughter. The present study investigates how overlapping events are coded in the brain to support integrated and separated memory representations. In particular, we examined if integrated representations can be formed without losing episodic details for the individual events, i.e., information required for source memory judgments. Participants (N = 39) were presented with Sims videos simulating real-life events while their electroencephalograph (EEG) was recorded. First, they encountered videos of two interacting characters (person A and B). These were followed by a new set of videos showing a new character interacting with an old one (person C and B). Participants were asked to integrate the overlapping events to infer the indirect relationships between the characters (i.e., AC) and to keep information about the event-specific relationships (i.e., AB and CB). In a final test, we measured memory for all associations. Source memory was tested by asking participants if the two characters had directly interacted with each other. Participants’ source memory accuracy was higher when they also correctly inferred the indirect relationships across episodes. Representational similarity analysis (RSA), used to investigate how the brain represents overlapping episodes, showed that the AB and CB episodes were characterized by systematic neural pattern similarities and dissimilarities. Importantly, the neural similarities observed were predictive of both future inference and memory for the source. The data show that neural pattern similarities between overlapping episodes support memory integration across event boundaries, while pattern dissimilarities may separate events. This study provides novel evidence that integrated and separated representations may coexist to support multiple memory functions.}},
author = {{Liu, Zhenghao and Johansson, Mikael and Bramao, Ines}},
language = {{eng}},
title = {{Time-resolved representational similarity analysis reveals integrated and separated neural patterns of overlapping events}},
url = {{https://lup.lub.lu.se/search/files/211889232/LEARNMEM2023_poster_final.pdf}},
year = {{2023}},
}