A representational similarity analysis of cognitive control in color-word Stroop

Michael C. Freund, Julie M. Bugg and Todd S. Braver

Published in Journal of Neuroscience 23 June 2021, JN-RM-2956-20; DOI: https://doi.org/10.1523/JNEUROSCI.2956-20.2021

Biorxiv version: 10.1101/2020.11.22.392704

Please email Mike Freund at m.freund@wustl.edu with any questions or concerns.

---

PIPELINE SUMMMARY

This summary is broken down by section of the manuscript. The tables here connect analysis scripts to parts of the manuscript (Figures, Tables, text sections).

Estimating Coding Strength \beta

• scripts in ./code/behav/

script	manuscript	purpose
1_define_models.R	.csv files in rsa/mods/	Figure 1
2_estimate_rsms.R	.rds files in rsa/obsv	Method::Estimating Coding Strength \beta
3_regress_run_effects.R	.rds files in rsa/obsv with suffix tag _residual	Method::Estimating Coding Strength \beta
4_model_rsms.R	.csv files in rsa\stats	Method::Estimating Coding Strength \beta

Behavioral analyses

• scripts in ./code/behav/
• master file: _behav.rmd (this sources all scripts in order and generates report)
• view report: _behav.html

script	manuscript	purpose
prelim_behavioral_models.R	Method::Selection of Behavioral Measures...	models behavioral data of 'primary analysis set'
microphone_comparison.R	Method::Selection of Behavioral Measures...	estimates impact of microphone change on RT measures
prelim_behavioral_models_validation_set.R	Method::Selection of Behavioral Measures...	models RT data of 'validation set'

Group-level analyses

• scripts in ./code/group/
• master file: _group.rmd (this sources all scripts in order and generates report)
• view report: _group.html

script	manuscript	purpose
fpc_dissoc.R	Results::Group::Dorsomedial ...; Figure 2A; Tables A1--A3	tests hypotheses regarding group-level coding dissociations
mds.R	Figure 2B	dimensionality reduction to visualize DMFC (L), V1, and SomMot--Mouth geometries
visual_sm_dissoc.R	Results::Group::Sensitivity and control analyses; Figure A1	test SomMot--mouth and V1 coding for positive control analysis
fpc_dissoc_parcel.R	Results::Group::Sensitivity and control analyses; Figure A2	sensitivity test for group-level analysis (parcel-level)
fpc_dissoc_altdef.R	Results::Group::Sensitivity and control analyses; Table A4	sensitivity test for group-level analysis (alternate ROI definitions for DMFC and DLPFC)
noise_ceiling.R	Results::Group::Sensitivity and control analyses, stats inline	estimate noise ceilings per ROI
noise_ceiling_tost.R	Results::Group::Sensitivity and control analyses, stats inline	contrast noise ceiling estimates across ROIs, provide two one-sided test for equivalence

Individual (Difference) analyses

• scripts in ./code/indiv/
• master file: _indiv.rmd (this sources all scripts in order and generates report)
• view report: _indiv.html

script	manuscript	purpose
bivar_superparcel.R	Figure 3A; Figure A3;	create bivariate scatterplots Stroop ~ coding strength
single_roi.R	Results::Individual::Better-performing...; Table A5	test stroop*coding-strength relationship in each ROI*coding-scheme separately
wn_roi_contrast.R	Results::Individual::Better-performing...; Table A6	contrast stroop*coding-strength relationship between coding schemes (incongr., target), within-ROI
bn_roi_contrast.R	Results::Individual::Better-performing...; Table A7	contrast stroop*coding-strength relationship between ROIs, within coding schemes (incongr., target)
model_selection.R	Results::Individual::Model selection...; Figure 3B,C	elastic net model selection and validation-set prediction
bivar_allcorrs_table.R	Results::Individual::Model selection...; Table A9	largest 20% of stroop~coding-scheme correlations observed across all superparcels

Exploratory whole-cortex analyses

• scripts in ./code/explor/

script	manuscript	purpose
explor.rmd	Results::Exploratory; Figure 4, Table A10--12, Figure A6	test group-level coding of each scheme (target, distr, incongr.) in each MMP parcel
movregs_rsa.rmd	Results::Exploratory; Table A13	negative control: are coding schemes 'encoded' within movement regressors?

other scripts

'helpers'

scripts for sourcing, all located in ./code

• packages.R, strings.R, funs.R, read_atlases.R (depends: write_atlases.R), read_masks.R (depends: write_masks.R)

'prelim scripts'

scripts that set up initial things

• write_atlases.R (depends: ), write_masks.R (depends: )

---

MORE ABOUT REPO

./in

• contains behavioral data assembled by stroop-rsa/write_behav_stroop_rsa.R.
• no other script writes to this directory.
• underlay surfaces downloaded from BALSA: https://balsa.wustl.edu/sceneFile/show/7qP5m

./code

• primary analysis pipeline (bash, .R scripts)
• dynamic reports (.rmd files)
• reads from ./data, nil-bluearc (for 3D+t images), ./out
• writes to ./out
• ./behav: scripts / reports for generating behavioral analysis (RT and accuracy)
• ./group: scripts / reports for generating group-level analyses
• ./indiv: scripts / reports for generating individual difference analyses (brain~behavior)

./out

• all output of scripts are directed to this directory
• the one exeption is output related to fMRI GLMs, which is saved within ./glms
• ./rsa
  • ./mods: representational similarity models generated by define_models.R
  • ./obsv: observed similarity matrices. saved as arrays within .rds files.
  • ./stats: subject level fits and group statistics, saved in long-form .csvs
• ./summaries: various summary tables (.csvs) for QC, misc analyses, and things that might be read into a manuscript file.
• ./masks: functionally or anatomically defined brain masks, created by scripts within ./code/masks
• ./behav: output of scripts / reports for generating behavioral analysis (RT and accuracy)
• ./group: output of group-level analyses
• ./indiv: output of individual difference analyses (brain~behavior)

./glms

• contains AFNI GLM input (e.g., stimtime and movreg files), shell scripts, and output (e.g., .nii brick files)
• 3D+t images read from nil-bluearc
• GLMs fit on ccplinux1, and results merged with local directory via rsync