rsa_project

Analysis of gambling fmri data

file directory

• Untitled0.ipynb inspecting the gambling data

• a4bres.rds results of analysis4.R

• temp.RData Temporary file (obv.) Currently: results for analysis 6, in progress.

• temp_pvals.rds results of analysis 4??

• temp_script1.R Not really a temporary script: used for producing files roi/data.rds, roi/cluster.rds from Poldrack's 3d data and the clustering results in the roi/ folder.

• a6source.R new bootstrap functions developed for analysis 6 in order to fit subjects individually

• analysis1.R Incomplete. Attempt to apply ANOVA to voxels/copes.

• analysis2.R Classification of copes. Finding which ROIs have signal. Pooling across subjects boosts signal. DIMENSION REDUCTION.

• analysis3.R Computing distance-feature matrices (without testing, without ordination.)

• analysis4_sim_diagnosis.R At first we got wrong p-values in analysis 4, this simulation study helped lead to realizing that the heterogeneity of covariances was the root of the issue.

• analysis4.R first time we applied bootstrap to get p-values for metric equivalence

• analysis5.R first use of Procrustes

• analysis4b.R update analysis4 by using unbiased test stat

• analysis4b_sim.R check if the approach in analysis 4b controls type I error, power etc

• analysis6.R contrast with analysis 4: pool subjects first then compute test stat

• analysis6_sim.R (check if the approach in analysis 4b controls type I error, power etc). By way of comparing analysis4b_sim.R and analysis6_sim.R we conclude that pooling subjects (the 4b approach) is more powerful.

• analysis7.R apply T-test to distance matrices

• analysis7_sim.R check if the approach in analysis 7 controls type I error, power etc

• fun_plot.R cool 3d plot! totally unrelated to the project though...

• graphics_a5.R Plots which try to align MDS distance matrices to the natural parameter distance grid. (Hard to explain..)

• indscal_source.R Find latent parameterizations from distances (when you don't have parametrically generated stimuli...)

• instructions.txt This was an email written by Russell Poldrack on how to get the data from TACC and the experimental parameters.

• make_doppelganger_B.R See make_doppelganger.R. This one tries to imitate the data after it has already been reduced in dimensionality.

• make_doppelganger.R Fits a model to the data in order to generate a synthetic dataset with similar properties.

• prepare_dim_reduced.R An important source file used by many analyses. Further processes the data matrix, including PCA.

• rsa_boot_source.R A file copied over from github/snarles/fmri then modified. Contains code for bootstrap-based tests of distance equivalence!

• script_output.txt Output of scripts used to extract ROIS from parametric map data.

• source1.R some functions for working with the nifti 3d brain scans.

doppel/

Includes synthetic datasets. (see make_goppelganger.R).

roi/

Various files and scripts involved in extracting ROIs from the parametric brain map. Also contains the data matrix used in many analyses.

theory/

Some simulations, mostly about factor analysis (indscal.)