Memories Between Us: Autobiographical Memory Sharing with Friends and Strangers
(2026) PSYP01 20261Department of Psychology
- Abstract
- Autobiographical remembering is reconstructive and frequently occurs in social
interaction, yet it remains unclear whether the relationship between narrator and
listener shapes how memories are expressed and subsequently remembered.
This study examined autobiographical memory sharing with friends versus
strangers and evaluated the use of a large language model (GPT-5.5) for coding
narrative data. Fifty-six adults (28 narrators, 28 listeners) participated in a
repeated-measures memory-sharing paradigm involving both relationship
conditions. Narrators shared positive autobiographical memories, with a nonshared memory serving as a baseline for pre-post comparisons. Narratives were
coded for episodic richness and self-event connections... (More) - Autobiographical remembering is reconstructive and frequently occurs in social
interaction, yet it remains unclear whether the relationship between narrator and
listener shapes how memories are expressed and subsequently remembered.
This study examined autobiographical memory sharing with friends versus
strangers and evaluated the use of a large language model (GPT-5.5) for coding
narrative data. Fifty-six adults (28 narrators, 28 listeners) participated in a
repeated-measures memory-sharing paradigm involving both relationship
conditions. Narrators shared positive autobiographical memories, with a nonshared memory serving as a baseline for pre-post comparisons. Narratives were
coded for episodic richness and self-event connections (SECs), and listeners
later recalled the memories they had heard. Relationship condition did not
significantly affect episodic richness during narration, pre-post change, or
listener recall. Pre-post change in SECs likewise did not differ across friend,
stranger, and baseline conditions. During narration, narrators produced more
SECs when speaking to strangers than friends, although this effect was no longer
significant after accounting for pre-session SECs and should therefore be
interpreted cautiously. GPT coding showed framework-dependent performance,
with stronger reliability and human agreement for total episodic richness than
for some other coding frameworks. Overall, the findings provide little evidence
that a single socially situated recall produces broad or persistent changes in
autobiographical memory, while suggesting that relationship context may
selectively influence self-related framing. LLM-based narrative coding showed
promise but requires framework-specific validation. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/student-papers/record/9250836
- author
- Jalk Carlé, Simone Mikaela LU
- supervisor
- organization
- course
- PSYP01 20261
- year
- 2026
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- Autobiographical memory, memory sharing, reconstructive memory, interpersonal relationships, large language models.
- language
- English
- id
- 9250836
- date added to LUP
- 2026-09-11 14:21:08
- date last changed
- 2026-09-11 14:21:08
@misc{9250836,
abstract = {{Autobiographical remembering is reconstructive and frequently occurs in social
interaction, yet it remains unclear whether the relationship between narrator and
listener shapes how memories are expressed and subsequently remembered.
This study examined autobiographical memory sharing with friends versus
strangers and evaluated the use of a large language model (GPT-5.5) for coding
narrative data. Fifty-six adults (28 narrators, 28 listeners) participated in a
repeated-measures memory-sharing paradigm involving both relationship
conditions. Narrators shared positive autobiographical memories, with a nonshared memory serving as a baseline for pre-post comparisons. Narratives were
coded for episodic richness and self-event connections (SECs), and listeners
later recalled the memories they had heard. Relationship condition did not
significantly affect episodic richness during narration, pre-post change, or
listener recall. Pre-post change in SECs likewise did not differ across friend,
stranger, and baseline conditions. During narration, narrators produced more
SECs when speaking to strangers than friends, although this effect was no longer
significant after accounting for pre-session SECs and should therefore be
interpreted cautiously. GPT coding showed framework-dependent performance,
with stronger reliability and human agreement for total episodic richness than
for some other coding frameworks. Overall, the findings provide little evidence
that a single socially situated recall produces broad or persistent changes in
autobiographical memory, while suggesting that relationship context may
selectively influence self-related framing. LLM-based narrative coding showed
promise but requires framework-specific validation.}},
author = {{Jalk Carlé, Simone Mikaela}},
language = {{eng}},
note = {{Student Paper}},
title = {{Memories Between Us: Autobiographical Memory Sharing with Friends and Strangers}},
year = {{2026}},
}