Online supplement for the paper "Neural detection of socially valued community members" (Morelli et. al, in prep)
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
Factor_analysis
Median_scripts
NoControlScripts
Nomination_matrices
Parametric_analyses
Prescan_ratings
Quartile_scripts
WS_prediction/scripts
data
fmri_tasks
masks
output_prediction
results
scripts
zSpace_scripts
.Rhistory
.gitignore
README.md

README.md

Neural detection of socially valued community members

This repository hosts the online supplement for the paper: "Neural detection of socially valued community members" (Morelli, Leong, Carlson, Kullar, & Zaki, in press)

For a preprint of the paper, please contact Sylvia Morelli at smorelli@uic.edu.

Social Network Nominations

Data

Nomination Matrices: Adjacency matrices of nominations for each of the 8 social network questions for the larger sample of 197 participants, as well as a matrix that represents the weighted average of these 8 questions

Analyses

Factor Analysis: Factor analysis on indegree for each of the eight questions, using the full sample (i.e., 97 participants)

Pre-Scan Ratings of Dorm Relationships

Pre-Scan Ratings: Anonymized data of scanner participants' ratings of each dorm member on various dimensions

Neuroimaging Tasks

Face Viewing

Face Selection Algorithm: Script for selecting 30 target faces for each participant based on their pre-scan ratings

Face Selection Files: 30 target faces selected for each participant produced by the face selection algorithm

Face Viewing Task: Main script to run the face-viewing task (but missing the folder of target photos to maintain anonymity)

Face Viewing Task Output: Recorded onsets & durations for stimuli, as well as button presses

Preprocessing scripts: SPM preprocessing scripts for all tasks (including face viewing)

First-level scripts for parametric analyses: SPM subject-level scripts for parametric modulation

First-level scripts for hub categories: SPM subject-level scripts used to generate hub categories (median split, terciles, & quartiles) for univariate and multivariate prediction analyses

Parametric_analyses: T maps for the parametric analyses reported in the paper and supporting appendix which can also be viewed in our NeuroVault Collection

Functional Reward Localizer

Modified Monetary Incentive Delay Task: Main script to run the modified MID (but missing the folder of photos to maintain anonymity)

MID Output: Recorded onsets & durations for stimuli, as well as button presses

Preprocessing scripts: SPM preprocessing scripts for all tasks (including MID)

First-level scripts:SPM subject-level scripts

Prediction Analyses

Data

Data for the prediction analyses reported in the paper can be downloaded here. Each participant's subfolder (SN_XXX) contains three pairs of .img/.hdr files. Each pair contains a t-map associated with a particular hub category:

  • spmT_0001 - High hub category
  • spmT_0002 - Middle hub category
  • spmT_0003 - Low hub category

ROI masks used for the analyses can be found here

Scripts

UnivariatePrediction.m: Follows a leave-one-participant-out cross-validation procedure to predict hub category from the average t-values of held-out data in a given ROI.

MultivariatePrediction.m: Follows a leave-one-participant-out cross-validation procedure to train a LASSO-PCR algorithm to predict hub category from neural patterns of held-out data in a given ROI.

Compare_RMSE.m: Compares univariate and multivariate prediction accuracy using root mean squared error (RMSE).

ParcelSearchLightAnalysis.m: Prediction analyses using whole-brain parcellation ROIs.

The following folders contain scripts for additional control analyses.

Median_scripts: Analyses when splitting data into two bins

Quartile_scripts: Analyses when splitting data into four bins

NoControlScripts: Analyses when not controlling for personal nomination and closeness

WS_prediction: Within-Subject Prediction Analyses

Comparing increase in response between terciles

Pattern Weights

Multivariate pattern weights learned by LASSO-PCR algorithm for each ROI can be found here ([roi_name.nii])

Dependencies

To run the prediction scripts, you will need to download the following toolboxes: