Hippocampal pattern separation/completion and cognitive map during naturalistic stimuli
Poster No:
2329
Submission Type:
Abstract Submission
Authors:
lili sun1, Xia Liang1
Institutions:
1Harbin Institute of Technology, Harbin, Heilongjiang
First Author:
Co-Author:
Introduction:
The hippocampus, known for supporting episodic memory, could represent the traces for individual experiences can vary in their relations, becoming more distinct (pattern separation) or more similar (pattern completion) to each other[1,2]. Moreover, it also has the ability to abstract common elements and infer relationships among information from diverse experiences, which enables the construction of structured knowledge about the information, forming 'cognitive maps'[3,4]. However, we still lack a detailed understanding of whether and how hippocampal pattern separation/completion contribute to cognitive map formation during movie stimuli. Here, using ultra-high-resolution fMRI data from the Human Connectome Project (HCP), we first investigated how the hippocampal circuit exhibits pattern separation/completion at different time scales. Then, we constructed movie semantic networks and hippocampal subfield BOLD networks to represent cognitive map respond to movie stimuli. Finally, we studied how these movie networks relate to pattern separation/completion.
Methods:
All data used here come from184 subjects of the HCP 7T release in this study (13 video clips)[5]. We segmented the hippocampus into the dentate gyrus (DG) and cornu ammonis (CA) 1-3 subfields, then focused on two input-output pairs (i.e., DG→CA3 and CA3→CA1) along the hippocampal tri-synaptic circuit. For each clip, sliding window correlation was conducted to evaluate temporal similarity of input/output fMRI patterns for each input-output pair[6]. Further, we compared input and output patterns to derive the ratio of pattern separation/completion at each window size (range from 1TR to 50TR) and used paired t-tests to assess the significance of pattern separation or completion across all subjects. To explore whether the hippocampal BOLD network acquired the topological properties of the movie semantic network, we constructed networks for hippocampal BOLD signal and semantic features respectively, and compared the global and local efficiency of these networks with their corresponding random networks. Moreover, we used the Floyd-Warshall algorithm to calculate the shortest path distance between nodes in the BOLD networks and evaluated the changes in path distance along the DG-CA3-CA1 circuit. We next calculated Pearson correlation of input-output difference matrix between temporal similarity and shortest distance, and Pearson correlation coefficients were transformed to z-scores using Fisher's transformation.
Results:
We found that when the window size was longer than 2 TRs, DG→CA3 showed significant pattern separation than completion in ten movie clips, while CA3→CA1 pair exhibited significant pattern completion in all movie clips (p<0.05, Fig.1).
We found that these networks created from movie semantic and hippocampal subfield BOLD signals showed significantly lower global efficiency but significantly higher local efficiency compared to random networks (p<0.05).
For the DG→CA3 pair that showed significant pattern separation effect, we observed that the mean shortest path distance between all node pairs within the CA3 BOLD network increased significantly compared to the DG. In contrast, for the CA3→CA1 pair that showed significant pattern completion effect, the mean shortest path distance between all node pairs within the BOLD network of the CA1 decreased significantly compared to of the CA3 (p<0.05, Fig.2).
We found that these networks created from movie semantic and hippocampal subfield BOLD signals showed significantly lower global efficiency but significantly higher local efficiency compared to random networks (p<0.05).
For the DG→CA3 pair that showed significant pattern separation effect, we observed that the mean shortest path distance between all node pairs within the CA3 BOLD network increased significantly compared to the DG. In contrast, for the CA3→CA1 pair that showed significant pattern completion effect, the mean shortest path distance between all node pairs within the BOLD network of the CA1 decreased significantly compared to of the CA3 (p<0.05, Fig.2).
·Pattern separation and completion effect under different window lengths in the hippocampal circuit
·Results of the relationship between pattern separation/completion and shortest distance and at a 10% threshold
Conclusions:
The hippocampal circuits could underpin pattern separation/completion at relatively short time scales (>2s) and learned the topological property from the semantic network during movie stimuli. Moreover, significant changes in the shortest distances between nodes in the BOLD network of the hippocampal pair were observed. These findings may contribute to our understanding of how pattern separation and completion connect with cognitive map organization during naturalistic stimuli, as well as the distinct functions played by various hippocampal subfields in these mechanisms.
Brain Stimulation:
Non-invasive Electrical/tDCS/tACS/tRNS
Higher Cognitive Functions:
Higher Cognitive Functions Other 2
Learning and Memory:
Long-Term Memory (Episodic and Semantic)
Modeling and Analysis Methods:
Activation (eg. BOLD task-fMRI)
Novel Imaging Acquisition Methods:
BOLD fMRI 1
Keywords:
Data analysis
Learning
Memory
MRI
Other - naturalistic stimuli
1|2Indicates the priority used for review
Provide references using author date format
Marr D. (1971), 'Simple memory: a theory for archicortex'. Philos Trans R Soc Lond B Biol Sci, 262(841): 23-81
Stachenfeld K L. (1971), 'Gershman S J. The hippocampus as a predictive map[J]. Nature Neuroscience, 20(11): 1643-+.
Stoewer P. (1971), 'Neural network based successor representations to form cognitive maps of space and language'. Sci Rep, 12(1): 11233.
Van Essen D C. (1971), 'The WU-Minn Human Connectome Project: an overview'. Neuroimage, 80: 62-79.
Shakil S. (1971), 'Evaluation of sliding window correlation performance for characterizing dynamic functional connectivity and brain states'. Neuroimage,133: 111-128.