Main content



Loading wiki pages...

Wiki Version:
This OSF directory contains the data and code specific to "People represent mental states in terms of rationality, social impact, and valence: Validating the 3d Mind Model". This study draws on data from four previous studies. The OSF repositories for all of those studies are linked for convenience. The data dictionary below describes the material within this repostiory: * alldims.csv - average ratings for 58 dimensions of mental state representation across 60 mental states * alldims_withpcs.csv - same as 'alldims.csv' but additionally containing four variables with PC scores from a PCA conducted as part of "Study 1A" in the paper (Tamir, Thornton, Contreras & Mitchell, 2016, PNAS) * analyze_msb_ms_dim.R - analysis script for behavioral mental state similarity judgement task conducted on * mega_analysis.R - analysis script for conducting a mega-analysis on the neural dissimilarity data * ms_dim.R - analysis script conducting a bi-cross-validated PCA on the 58 dimensions in alldims.csv, followed by a cross-validated model selection procedure using the neural dissimilarity matrices * ms_dim_results.Rdata - results of ms_dim.R model selection procedure, stored due to long computation time * msb_msdim_data.csv - behavioral data from the similarity judgement task on * neural_dissimilarity_matrices/ - directory containing neural representational dissimilarity matrices from Studies 1A-D. 'origsorted' = Study 1A, 'targ' = Study 1B, 'adapt' = Study 1C, and 'pic' = Study 1D. Studies 1B and 1D also indicate the target person (self, close other, or far other) * parcel_analysis.R - wholebrain parcel-based version of the mega-analysis * parcel_data/ - files for parcel-based mega-analysis. Each subdirectory contains csv files containing vectorized neural neural dissimilarity matrices * msadapt_parcelsim/ - Study 3 * msorig_parcelsim/ - Study 1 * mspic_parcelsim/ - Study 4 * mstarg_parcelsim - Study 2 * state_vectors.csv - word embedding vectors for 166 mental state terms extracted from a fasttext model pretrained on the common crawl * state_files/ - files containing ordered lists of the states presented in each study. Naming conventions shared with the neural dissimliarity matrices. states60 contains the original (misordered) condition order from Study 1, used for resorting in the parcel analysis. * text_similarity.R - analysis script for conducting representational similarity analysis on the text state vectors * univariate_ratings166.csv - contains average ratings of 16 dimensions collected as part of Study 1A. These dimensions are included in alldims.csv, but that file contains only the 60 states using in Studies 1A-C, whereas this file contains all 166 originally rated because all of those states were featured in the behavioral study on and the text analysis. * whole_brain_cluster_labels_PCA=100_k=200.nii.gz - MRI data file containing neurosynth parcellation used in wholebrain mega-analysis