Charting neural representations behind human spatial learning
Poster No:
1064
Submission Type:
Abstract Submission
Authors:
Liangyue Song1,1, Joern Alexander Quent2, Xinyu Liang3, Yueting Su3, Deniz Vatansever3
Institutions:
1Fudan University, Shanghai, China, 2Fudan University, Shanghai, NA, 3Fudan University, Shanghai, Shanghai
First Author:
Co-Author(s):
Introduction:
Spatial navigation is a core cognitive process that relies on the coordinated activity of a complex and widespread neural circuitry [1]. Notably, a set of brain regions centered on the medial temporal, parietal and prefrontal cortices are now suggested to exhibit grid-like activity patterns during spatial navigation [2], providing mechanistic insights into the use of cognitive maps in the navigation of learned spatial environments. However, there is notable scarcity of research on the encoding of spatial memory and thus the learning process itself. Here, using an ecologically valid virtual reality environment, we aimed to map the neural mechanism underlying the naturalistic encoding of spatial memory in humans.
Methods:
A group of 55 participants (mean age = 23.70, SD = 2.21, F/M ratio = 34/21) were scanned at 3T MRI via HCP style data acquisition protocols, while performing two runs of an Object Location Memory (OLM) paradigm. In this task, participants were asked to navigate in a virtual arena in order to learn and memorize the locations of six hidden objects through multiple cycles of memory retrieval and update (Fig. 1a). Memory performance and learning in each trial was measured via placement error i.e. the the Euclidean distance between the estimated and correct object location (Fig. 1b), which was subsequently employed as a parametric modulator in a GLM-based analysis of the task fMRI data. Our main analysis of interest was to investigate brain regions which showed greater activity with a reduction in placement error, indicating the onging learning process. Results were further interrogated to unveil their distribution across the cortical and subcortical organization of the human brain.
Results:
Behavioral results revealed a significant reduction in the placement error within and between runs, demonstrating a learning effect in the memorization of object locations. Using placement error as the parametric modulator, we identified a host of regions responsive to improved memory performance, suggesting their involvement in spatial learning. These results encompassed regions that are highly connected to learning such as the hippocampus, as well as regions associated with grid-like activity patterns during navigation, such as the entorhinal cortex (EC), orbital prefrontal cortex (OFC) and temporal cortex (Fig. 1c).
Network partitioning revealed that regions belonging to the Default Mode (DMN) and Somatomotor Networks (SMN) contributed most to the significant clusters (FDRp < 0.05, Fig. 2ai). This was further supported by cortical gradient projection, where significant clusters were positioned along transmodal and somatomotor endpoints of the principal connectivity gradients (Fig. 2aii). In a between-subject analysis, the role that the DMN plays in spatial learning was further highlighted by the significant correlation between modulation of regions within this network and task performance (p<0.05, Fig. 2aiii). While at the subcortical level, the results revealed that the learning related modulation was localized in the bilateral anterior hippocampus, left amygdala, and the ventral striatum (Fig. 2bi). The modulation effect from the left hippocampus positively correlated with performance improvement from Run 1 to Run 2 (cor=0.27, p=0.04), indicating its contribution to spatial learning (Fig. 2bii).
Network partitioning revealed that regions belonging to the Default Mode (DMN) and Somatomotor Networks (SMN) contributed most to the significant clusters (FDRp < 0.05, Fig. 2ai). This was further supported by cortical gradient projection, where significant clusters were positioned along transmodal and somatomotor endpoints of the principal connectivity gradients (Fig. 2aii). In a between-subject analysis, the role that the DMN plays in spatial learning was further highlighted by the significant correlation between modulation of regions within this network and task performance (p<0.05, Fig. 2aiii). While at the subcortical level, the results revealed that the learning related modulation was localized in the bilateral anterior hippocampus, left amygdala, and the ventral striatum (Fig. 2bi). The modulation effect from the left hippocampus positively correlated with performance improvement from Run 1 to Run 2 (cor=0.27, p=0.04), indicating its contribution to spatial learning (Fig. 2bii).
Conclusions:
Our findings indicate that spatial learning in a virtual reality environment engages medial temporal lobe regions that often exhibit grid-like activity patterns, suggesting a potential role for cognitive map dynamics in this encoding process. Moreover, the additional involvement of regions within the Default Mode Network adds an intriguing layer to our findings, highlighting the active contribution of this transmodal network to spatial learning.
Higher Cognitive Functions:
Higher Cognitive Functions Other
Learning and Memory:
Learning and Memory Other 1
Modeling and Analysis Methods:
Activation (eg. BOLD task-fMRI) 2
Novel Imaging Acquisition Methods:
BOLD fMRI
Keywords:
Cognition
Learning
Memory
Other - Virtual Reality
1|2Indicates the priority used for review
Provide references using author date format
[2] Doeller CF, Barry C, Burgess N (2010), ‘Evidence for grid cells in a human memory network.’ Nature, 463(7281):657-61.
[3] Glasser MF et al. (2016), ‘The Human Connectome Project's neuroimaging approach’, Nature Neuroscience, 19:1175–1187.
[4] Worsley KJ et al. (2009), ‘A Matlab toolbox for the statistical analysis of univariate and multivariate surface and volumetric data using linear mixed effects models and random field theory.’ NeuroImage, 47(S1):S39-S41.