Motion-corrected eye tracking improves gaze accuracy during visual fMRI experiments

(Dec 4, 2025) Minor edit notice:

A minor code issue was identified in the notebook analysis/scripts/nonlinear_mocet/motion_types.ipynb. This does not affect any figures in the notebooks or the results reported in the paper. As described in the Methods section (“Motion-Corrected Eye Tracking (MoCET) – Nonlinear motion-correction”), the autocorrelation scores used in Supplementary Figure 3 should be computed using lags of one to five TRs. The current notebook version uses only a one-TR lag. However, all figures included in the paper and notebooks were generated using the correct implementation. If desired, users may update the function momentum_score() to the corrected version below:

# analysis/scripts/nonlinear_mocet/motion_types.ipynb
def momentum_score(x):
x = np.asarray(x)
x_mean = np.mean(x)
correlation_coefficients = []
for i in range(1, 6):
correlation_coefficients.append(np.corrcoef(x[i:] - x_mean, x[:-i] - x_mean)[0, 1])
return np.mean(correlation_coefficients)

This corrected function reflects the version used to generate all published figures and results. Codes in the github were revised.

Dataset Description: 

This repository contains the data and code used in "Park, J., Jeon, J. Y., Kim, R., Kay, K. & Shim, W. M. Motion-corrected eye tracking improves gaze accuracy during visual fMRI experiments", which investigates the impact of head motion on gaze accuracy in fMRI experiments and proposes a method for drift correction.

Usage:

Please see the basic instruction: https://github.com/jwparks/mocet

Restrictions:

Raw eye tracking video, Neuroimaging data and gameplay data are not included due to privacy and storage constraints.

This dataset is intended for research purposes only.