Map-like and graph-like representations of social knowledge in the hippocampus
Poster No:
806
Submission Type:
Abstract Submission
Authors:
DASOM KWON1,2, Luke Chang3, Eshin Jolly3, WON MOK SHIM1,2
Institutions:
1Sungkyunkwan University, Suwon-si, Korea, Republic of, 2Center for Neuroscience Imaging Research, Suwon-si, Korea, Republic of, 3Dartmouth College, Hanover, NH
First Author:
DASOM KWON
Sungkyunkwan University|Center for Neuroscience Imaging Research
Suwon-si, Korea, Republic of|Suwon-si, Korea, Republic of
Sungkyunkwan University|Center for Neuroscience Imaging Research
Suwon-si, Korea, Republic of|Suwon-si, Korea, Republic of
Co-Author(s):
Won Mok Shim
Sungkyunkwan University|Center for Neuroscience Imaging Research
Suwon-si, Korea, Republic of|Suwon-si, Korea, Republic of
Sungkyunkwan University|Center for Neuroscience Imaging Research
Suwon-si, Korea, Republic of|Suwon-si, Korea, Republic of
Introduction:
Humans and animals utilize knowledge structures based on either map-like or graph-like representations to facilitate adaptive behaviors [1, 2]. Two essential types of information crucial for social interactions, personality traits of each individual and relationships among them, can be organized into distinct structures [3, 4]. Personality traits can be structured like a map, indicated by coordinates, while relationships follow a graph-based structure, represented by connections and distance. The hippocampus has been previously proposed as a central hub for structuring relational knowledge [1, 2, 5, 6]. In this study, we examine how the hippocampus organizes these diverse types of social knowledge, each characterized by distinct representational structures, for the same group of social agents in naturalistic settings.
Methods:
We collected high spatial resolution (1.5mm isotropic voxels) 7T fMRI data while participants (N = 24) watched first-person movies and played a social interaction game within a custom-built Minecraft environment. They also assessed seven personality traits (dominance, warmth, extroversion, competence, trustworthiness, neuroticism, and agreeableness) and two relationships traits (closeness and trust) of characters, including themselves, across three consecutive episodes of a narrative story during fMRI scanning. In the Minecraft virtual world, participants interacted with six characters, each with distinct personality traits, making social decisions in the conversational context (as in the narrative script). To examine the neural representation of knowledge structures related to both social traits, we employed a representational similarity analysis [7]. Neural representational dissimilarity matrices (RDM) were generated from the rating task using beta coefficients from the general linear model (GLM), encompassing both personality traits and relationship distances.
Results:
First, participants exhibited a clear understanding of the intended personality (r = .62, s. d. across participants = .07) and relationship traits (r = .56, s. d. = .12) among characters, including themselves, as designed in our study. The rank correlation between the neural RDMs and social knowledge RDMs across three episodes revealed that the hippocampus represents both map-like knowledge of personality traits (anterior hippocampus: t(23) = 5.02, p < .001; posterior hippocampus: t(23) = 3.47, p = .002) and graph-like knowledge of relationships (anterior hippocampus: t(23) = 2.06, p = .051; posterior hippocampus: t(23) = 3.45, p = .002). Furthermore, hippocampal structures associated with personality and relationship traits were extracted from the parametric GLM using distance metrics in each knowledge structure. In the first episode, these structures were positively correlated (t(23) = 2.97, p = .007), consistent with behavioral data, indicating that initially, characters perceived as similar in personality were also thought to have closer relationships. However, this pattern did not persist through the second and third episodes (the second episode: t(23) = 1.03, p = .314; the third episode: t(23) = 1.33, p = .196), suggesting separate computational processes for the two distinct knowledge structures.
Conclusions:
Our social environment inherently encompasses diverse forms of relational knowledge that facilitate the formation of either map or graph representations. In this study, we demonstrated the role of the hippocampus in representing both map-like and graph-like knowledge structures for the same group of social agents. Our findings suggest distinct mechanisms in the hippocampus for structuring these two different knowledge structures, raising further research questions of how the hippocampus constructs these types of knowledge structures during ongoing social experiences.
Emotion, Motivation and Social Neuroscience:
Social Cognition 1
Social Interaction 2
Higher Cognitive Functions:
Higher Cognitive Functions Other
Keywords:
FUNCTIONAL MRI
Social Interactions
Other - Interactive video game;Knowledge structure;Decision making
1|2Indicates the priority used for review
Provide references using author date format
[2] Peer, M., Brunec, I. K., Newcombe, N. S., & Epstein, R. A. (2021). Structuring Knowledge with Cognitive Maps and Cognitive Graphs. Trends in Cognitive Sciences, 25(1), 37–54.
[3] Hassabis, D., Spreng, R. N., Rusu, A. A., Robbins, C. A., Mar, R. A., & Schacter, D. L. (2014). Imagine all the people: how the brain creates and uses personality models to predict behavior. Cerebral Cortex, 24(8), 1979–1987.
[4] Kwon, D., Jolly, E., Chang, L. J., & Shim, W. M. (2023). Neural representation of dynamic social interaction. Manuscript in preparation.
[5] Tolman, E.C. (1948). Cognitive maps in rats and men. Psychological Review. 55, 189–208.
[6] Whittington, J. C. R., McCaffary, D., Bakermans, J. J. W., & Behrens, T. E. J. (2022). How to build a cognitive map. Nature Neuroscience, 25(10), 1257–1272.
[7] Kriegeskorte, N., Mur, M., & Bandettini, P. (2008). Representational similarity analysis - connecting the branches of systems neuroscience. Frontiers in Systems Neuroscience, 2, 4.