whyhowlocalizer-analysis

Software for analyzing single-subject fMRI data on the Why/How Localizer using SPM12

Before you begin

• The main function to run on your end is wrapper_level1_whyhow.m. See below to learn about the various arguments you can pass to it to customize the analysis, and see the included example_usage.m file to run a model on real data (in the "example_data" folder).
• Make sure you have SPM12 and included subfolder support on your MATLAB search path before running
• If you plan to use Weighted Least Squares (WLS) estimation (I do this by default), you need to have a copy of the Robust Weighted Least Squares toolbox in the toolbox folder within your SPM12 directory. A copy of the toolbox is included in the file rWLS_v4.0_SPM12.zip.

These varargins are used for finding the necessary files and directories

NAME	DEFINITION
studydir	full path to directory containing subject data folders
behavpat	search pattern for finding behavioral data within subject dirs, e.g., 'behav/whyhow*mat'
epipat	search pattern for finding functional data file(s) within each run, e.g., 'sw*nii'
nuisancepat	search pattern for finding nuisance regressor file within each run, e.g., 'rp*txt'
runpat	search pattern for finding run directories within subject dirs, e.g.,'raw/BOLD_WhyHow*'
subpat	search pattern for finding subject directories within studydir, e.g., 'Subject*'
brainmask	full path to brain mask to use (leave empty for none)

These varargins are used to specify relevant details about the image and behavioral data being modeled

NAME	DEFINITION
is4D	flag for 4D image file (0=No, 1=Yes)
nskip	number of initial TRs that have been removed (for adjusting stimulus onsets)
TR	acquisition repetition time (in seconds)
yesnokeys	keys corresponding to yes/no responses (e.g., [1 2])

These varargins are used to configure the model and estimation methods

NAME	DEFINITION
basename	base name for the analysis (e.g., 'WhyHow_SmoothedData')
model	string specifying model to use ('2x2' for full design, '1x2' to collapse faces/hands factor
incl_err	flag to include parametric covariate modeling blockwise variation in # of errors (0=No, 1=Yes)
incl_rt	flag to include parametric covariate modeling blockwise variation in response time (0=No, 1=Yes)
armethod	autocorrelation removal method (0=None, 1=AR(1), 2=Weighted Least Squares (WLS), 3=FAST)
HPF	high-pass filter cutoff to use (in seconds)
maskthresh	implicit masking threshold (proportion of globals), default = 0.8
fcontrast	flag to compute omnibus F-contrast (useful for feature selection, e.g., in PPI analysis) (0=No, 1=Yes)
run_it_now	flag to run analysis now (0=No, 1=Yes)