Voxel-wise encoding model of navigation behavior

Analysis & voxel-time-course simulation code for: Behavior-dependent directional tuning in the human visual-navigation network. Nau, Navarro Schröder, Frey, Doeller. 2020. Nature Communications

This code simulates voxel time courses, builds various encoding models of virtual head direction (vHD), trains them using cross-validated ridge regression, and finally tests them on held-out data. The model also estimates the vHD-tuning width for each voxel (similar to population receptive field mapping). Requires SPM12.

Click HERE for a visual depiction of the pipeline.

How to run the code

1. Download code
2. Open the script: "vHD_simulation.m"
3. Set path to SPM
4. Adjust the number of parallel workers to match your computer specs
5. Click "run"

To get a first immpression, I recommend reducing the number of voxels for the first runs to speed tings up.

Log file

The code makes use of the navigation data of a sample participant. The log file contains two relevant variables: "headDir", the virtual head direction over time (higher temporal resolution than the imaging data) and "TR_idz", the linear indizes corresponding to each value in headDir split into TRs/functional images. The file contains data of all 5 scanning runs.

Output and results

The script will visualize the results of the simulations as shown in the paper's SFig. 4C.

Adapting this code to analyse fMRI data

You can easily adapt this simulated code for new fMRI-data analyses by replacing the simulated voxel time courses for real ones (e.g. taken from an ROI). In this case, preprocess & clean your data beforehand incl. nuisance regression of head- motion parameters...).

Questions?

If you have any comments or questions, please reach out to me: matthias.nau[at]nih.gov