Repository used for the paper (IN SUBMISSION): Sequence models for by-trial decoding of cognitive strategies from neural data
Following research on cognitive processing stages and localizing their onset:
-
Borst, Jelmer P., and John R. Anderson. “The Discovery of Processing Stages: Analyzing EEG Data with Hidden Semi-Markov Models.” NeuroImage 108 (March 2015): 60–73. https://doi.org/10.1016/j.neuroimage.2014.12.029.
-
Anderson, John R., Qiong Zhang, Jelmer P. Borst, and Matthew M. Walsh. “The Discovery of Processing Stages: Extension of Sternberg’s Method.” Psychological Review 123, no. 5 (October 2016): 481–509. https://doi.org/10.1037/rev0000030.
-
Weindel, Gabriel, Leendert van Maanen, and Jelmer P. Borst. “Trial-by-Trial Detection of Cognitive Events in Neural Time-Series.” Imaging Neuroscience 2 (December 16, 2024): 1–28. https://doi.org/10.1162/imag_a_00400.
The code for this project consists of three repositories:
- The one you are currently viewing, containing notebooks that interact with the other two repositories.
- A forked version of HMP.
- A Python package (HMP-AI), created to make working with HMP and S4/ML models easier.
I used a Docker image on a remote server to run the code, Docker scripts and miscellaneous install scripts (some specific to UU server architecture) can be found in the /docker folder. /docker/requirements.txt can be used as a guide to install Python packages needed by the code.
Install the linked versions of HMP and HMP-AI by executing pip install -e . from their respective directories.
The /weindel and /boehm folders contain the analysis notebooks for their respective datasets. Everything can be executed in order, but heed instructions in the individual notebooks. Raw files should be available at a location that corresponds to an environment variable: $DATA_PATH, $DATA_PATH/sat2 for Weindel, and $DATA_PATH/sat1 for Boehm.