Brain Networks Involved in Raven’s Standard Progressive Matrices (RSPM) and their Functions
Poster No:
1578
Submission Type:
Abstract Submission
Authors:
Riley Zurrin1, Samantha Wong2, Meighen Roes1, Chantal Percival1, Abhijit Chinchani3, Leo Arreaza1, Mavis Kusi4, Maiya Rasheed1, Zhaoyi Mo1, Vina Goghari4
Institutions:
1University of British Columbia, Vancouver, British Columbia, 2McGill, Montreal, Quebec, 3The University of British Columbia (UBC), Vancouver, British Columbia (BC), 4University of Toronto, Toronto, Ontario
First Author:
Co-Author(s):
Introduction:
A non-verbal estimate of fluid intelligence (Gf) is provided by the Raven's Standard Progressive Matrices (RSPM). Duncan and Owen proposed a neuroanatomical basis of Gf called the multiple demand system (MD system; Duncan & Owen, 2000) based on univariate functional magnetic resonance imaging (fMRI) analysis methodology. However, a network-based approach can provide (1) the anatomical configuration of the full set of brain networks involved, and (2) the task-induced BOLD signal changes associated with each anatomical configuration, which when combined with differential responses to task conditions, provide a neurological function for each network. In the current study we achieved this using carried out a dimensional, FIR-based analysis of the RSPM task, which utilized the smallest available anatomical measurement (i.e., voxels as opposed to parcels or a-priori regions of interest), and isolated task-timing-estimable variance prior to the dimension reduction.
Methods:
N = 56. Please see Figure 1 for task description). Constrained principal component analysis for fMRI (fMRI-CPCA) was used to extract the whole-brain task-based BOLD networks. It uses multivariate multiple regression to separate out task-timing-predictable variance in the BOLD signal from task-timing-unpredictable variance, and then PCA is applied to the former. Dominant sets of voxel-based component loadings are then interpreted spatially, alongside statistical assessment of temporal information in the task-induced BOLD changes. This method produces networks (task-timing dependent spatial patterns) that are common across all subjects, and allows the researcher to examine how each network's task-induced BOLD changes respond to subject and task/condition variables.
·Figure 1
Results:
Components were classified as Default Mode Network (DMN), Motor Responding Network (RESP), Multiple Demand Network (MDN), Multiple Demand Network and Default Mode Network (MDN/DMN), and Re-evaluation Network (RE-EV). We focus on the RESP, MDN, and (RE-EV) network here.
The anatomical depiction of RESP is presented in Figure 2A, and Figure 2B displays the estimated task-induced BOLD change pattern. The peak time point fell within approximately 11-14s for all conditions.
The anatomical depiction of MDN is presented in Figure 2C, and Figure 2D displays the estimated task-induced BOLD change pattern for MDN. The peak time point fell at approximately 9s for all conditions.
The anatomical depiction of RE-EV is presented in Figure 2E, and Figure 2F displays the estimated task-induced BOLD change pattern. The peak time point was the latest of all networks, falling within approximately 16-21s for all conditions. A significant difficulty × time interaction was caused by a more fluctuating curve for the medium condition, including the highest peak (see Figure 7B, 16-19s).
The Pearson correlations for the predictor weights averaged over the trial were computed between DMN, RESP, MDN, MDN/DMN, and RE-EV. The only correlations to achieve significance after correction were the positive correlations between MDN and RE-EV; namely, RE-EV Hard × MDN Medium (r = .43), and RE-EV Hard × MDN Hard (r = .44).
The anatomical depiction of RESP is presented in Figure 2A, and Figure 2B displays the estimated task-induced BOLD change pattern. The peak time point fell within approximately 11-14s for all conditions.
The anatomical depiction of MDN is presented in Figure 2C, and Figure 2D displays the estimated task-induced BOLD change pattern for MDN. The peak time point fell at approximately 9s for all conditions.
The anatomical depiction of RE-EV is presented in Figure 2E, and Figure 2F displays the estimated task-induced BOLD change pattern. The peak time point was the latest of all networks, falling within approximately 16-21s for all conditions. A significant difficulty × time interaction was caused by a more fluctuating curve for the medium condition, including the highest peak (see Figure 7B, 16-19s).
The Pearson correlations for the predictor weights averaged over the trial were computed between DMN, RESP, MDN, MDN/DMN, and RE-EV. The only correlations to achieve significance after correction were the positive correlations between MDN and RE-EV; namely, RE-EV Hard × MDN Medium (r = .43), and RE-EV Hard × MDN Hard (r = .44).
·Figure 2
Conclusions:
The MDN for solution searching peaked early in the trial (~9s peak), followed by RESP for response selection (~12s peak), and RE-EV for solution checking (~18s peak), (2) high activity in the MDN is correlated with high activity in the later-peaking RE-EV network, proposed to underpin cooperative searching (MDN) and checking (RE-EV) processes, supported by past work on other tasks (Lavigne, Menon, Moritz, & Woodward, 2020; Lavigne, Menon, & Woodward 2020; Lavigne et al., 2015), and provide overlap with the proposed abstraction/elaboration (MDN solving) and hypothesis testing (RE-EV checking) phases of the P-FIT (Jung & Haier, 2007; Colom et al., 2010).
Higher Cognitive Functions:
Reasoning and Problem Solving 2
Modeling and Analysis Methods:
Activation (eg. BOLD task-fMRI)
Connectivity (eg. functional, effective, structural) 1
fMRI Connectivity and Network Modeling
Methods Development
Keywords:
Cognition
Data analysis
Design and Analysis
FUNCTIONAL MRI
Multivariate
Statistical Methods
Other - Intelligence
1|2Indicates the priority used for review
Provide references using author date format
Lavigne, K. M., Menon, M., Moritz, S., & Woodward, T. S. (2020). Functional brain networks underlying evidence integration and delusional ideation. Schizophrenia research, 216, 302-309. https://doi.org/10.1016/j.schres.2019.11.038
Lavigne, K. M., Menon, M., & Woodward, T. S. (2020). Functional brain networks underlying evidence integration and delusions in schizophrenia. Schizophrenia bulletin, 46(1), 175-183. https://doi.org/10.1093/schbul/sbz032
Lavigne, K. M., Metzak, P. D., & Woodward, T. S. (2015). Functional brain networks underlying detection and integration of disconfirmatory evidence. NeuroImage, 112, 138–151. https://doi.org/10.1016/j.neuroimage.2015.02.043
Jung, R. E., & Haier, R. J. (2007). The parieto-frontal integration theory (P-FIT) of intelligence: Converging neuroimaging evidence. Behavioral and Brain Sciences, 30(2), 135-154. https://doi.org/10.1017/S0140525X07001185
Colom, R., Karama, S., Jung, R. E., & Haier, R. J. (2010). Human intelligence and brain networks. Dialogues in Clinical Neuroscience, 12(4), 489-501. https://doi.org/10.31887/DCNS.2010.12.4/rcolom