Passive Spatial Learning in Humans: Neural Alignment during Naturalistic Navigation
Poster No:
1062
Submission Type:
Abstract Submission
Authors:
Xinyu Liang1, Joern Alexander Quent1, Liangyue Song1, Yueting Su1, Deniz Vatansever1
Institutions:
1Institute of Science and Technology for Brain-inspired Intelligence (ISTBI), Fudan University, Shanghai, China
First Author:
Xinyu Liang
Institute of Science and Technology for Brain-inspired Intelligence (ISTBI), Fudan University
Shanghai, China
Institute of Science and Technology for Brain-inspired Intelligence (ISTBI), Fudan University
Shanghai, China
Co-Author(s):
Joern Alexander Quent
Institute of Science and Technology for Brain-inspired Intelligence (ISTBI), Fudan University
Shanghai, China
Institute of Science and Technology for Brain-inspired Intelligence (ISTBI), Fudan University
Shanghai, China
Liangyue Song
Institute of Science and Technology for Brain-inspired Intelligence (ISTBI), Fudan University
Shanghai, China
Institute of Science and Technology for Brain-inspired Intelligence (ISTBI), Fudan University
Shanghai, China
Yueting Su
Institute of Science and Technology for Brain-inspired Intelligence (ISTBI), Fudan University
Shanghai, China
Institute of Science and Technology for Brain-inspired Intelligence (ISTBI), Fudan University
Shanghai, China
Deniz Vatansever
Institute of Science and Technology for Brain-inspired Intelligence (ISTBI), Fudan University
Shanghai, China
Institute of Science and Technology for Brain-inspired Intelligence (ISTBI), Fudan University
Shanghai, China
Introduction:
Learning of spatial environments plays a vital role in the survival of both animal and human species. Emerging evidence has highlighted a "navigational neural network" in supporting the encoding and internal representation of a cognitive map of our environments during active navigation [1]. However, the contribution of this network to passive encoding (i.e. observer effect) of spatial environments remains unclear [2]. Recent work has shown that passive learning from lecture recordings could evoke shared neural representations across participants, with greater alignment to experts' responses linked to better learning outcomes [3]. Here, we conducted a virtual reality experiment using 3T fMRI with a naturalistic navigation video to investigate the neural underpinnings of passive spatial learning.
Methods:
A total of 48 participants (mean age = 23.69 years, SD = 2.15, F/M ratio = 30/18) underwent 3T fMRI scanning (TR = 0.8 s, TE = 37 ms, 2 mm iso voxels). Participants were asked to learn the locations of 6 different objects while watching a video of an agent navigating within a circular virtual arena (Fig. 1a). Subsequently, participants performed an Object Location Memory (OLM) task within the same arena, in which they actively searched and retrieved the hidden objects. We assessed participants' passive learning outcomes by analyzing their performance during the initial presentation of all objects, focusing specifically on the accuracy and placement error (Fig. 1b). Based on performance, we defined a fast-learner group comprising of four participants with high accuracy (≥3 trials) and low placement error (<30 vm), categorizing the remaining participants as part of the slow-learner group. The HCP style fMRI data was minimally preprocessed using HCP pipelines (Qunex) [4]. To calculate the shared neural activity patterns across slow and fast-learners during video watching, we employed an inter-subject pattern correlation framework [3,5]. In total, 360 cortical regions from multi-model parcellation [6] and 16 subcortical regions from Desikan-Killiany atlas were included. For each region, we derived a neural alignment score to fast-learners by correlating the shared response model (SRM) pattern in each slow-learner with the mean pattern across fast-learners in each time point, and then temporally averaged within each individual (Fig. 1c). To test if the neural alignment could predict learning outcomes, we calculated correlation between alignment scores and average placement error. Statistical significance was assessed using a one-side non-parametric permutation testing.
·Figure 1. Neural alignment during passive naturalistic navigation.
Results:
The neural alignment results revealed shared neural activity patterns centered within the occipital visual, motor, medial orbital frontal and anterior temporal cortices as well as subcortical regions such as the bilateral hippocampus and thalamus (Fig. 1c). When projected onto the cortical connectivity gradient space [7], the effects highlighted the distribution of the neural alignment scores both at the intermediate zones and at the endpoints of the unimodal-to-transmodal gradient. The alignment scores from the bilateral visual pathway, visual motion area, attention and navigation related regions could significantly predict average placement error, indicating better performance (FDRp < .05) (Fig. 2a). Moreover, neural alignment in the hippocampus, amygdala, and other cortical regions involved in cognitive control and navigation also displayed association with improved learning outcomes.
·Figure 2. Neural alignment to fast-learners during naturalistic navigation task predicts subsequent learning outcomes.
Conclusions:
Our results revealed that neural alignment to fast-learners within the hippocampus and visual regions during passive spatial learning could predict subsequent memory performance. Collectively, our findings not only highlight the potential value of using naturalistic navigation in investigating spatial learning and memory, but also provide vital evidence for the neural mechanisms underlying passive learning of spatial environments.
Learning and Memory:
Working Memory
Learning and Memory Other 1
Modeling and Analysis Methods:
Methods Development
Multivariate Approaches
Novel Imaging Acquisition Methods:
BOLD fMRI 2
Keywords:
Learning
Other - Spatial Navigatioin; Naturalistic Stimuli; fMRI; Neural Alignment
1|2Indicates the priority used for review
Provide references using author date format
2. Chrastil, E. R. & Warren, W. H. Active and passive contributions to spatial learning. Psychon Bull Rev 19, 1–23 (2012).
3. Meshulam, M. et al. Neural alignment predicts learning outcomes in students taking an introduction to computer science course. Nat Commun 12, 1922 (2021).
4. Ji, J. L. et al. QuNex—An integrative platform for reproducible neuroimaging analytics. Frontiers in Neuroinformatics 17, (2023).
5. Chen, J. et al. Shared memories reveal shared structure in neural activity across individuals. Nat Neurosci 20, 115–125 (2017).
6. Glasser, M. F. et al. A multi-modal parcellation of human cerebral cortex. Nature 536, 171–178 (2016).
7. Margulies, D. S. et al. Situating the default-mode network along a principal gradient of macroscale cortical organization. Proc Natl Acad Sci USA 113, 12574–12579 (2016).