A toolbox for deconvolution of overlapping EEG (Pupil, LFP etc.) signals and (non)-linear modeling
New in 1.2 (December 2020): More automatic cleaning tools (ASR, Entropy-Based). Many Bugfixes, better Documentation and better error-codes.
Download our reference paper Ehinger & Dimigen 2019 (peerJ).
We recently published a new preprint on the analysis of Eyetracking/EEG data, with unfold playing a prominent role Dimigen & Ehinger 2019
If you use the toolbox, please cite us as: Ehinger BV & Dimigen O, Unfold: An integrated toolbox for overlap correction, non-linear modeling, and regression-based EEG analysis, peerJ, https://peerj.com/articles/7838/
Find a twitter thread explaining the general idea here or have a look at Figure 1 of our paper
Adjust for overlap between subsequent potentials using linear deconvolution
Massive-Univarite Modeling (rERP) using R-style formulas, e.g.
Non-linear effects using regression splines (GAM), e.g.
Model multiple events, e.g. Stimulus, Response and Fixation
Use temporal basis functions (Fourier & Splines)
(Optional) regularization using glmnet
Temporal Response Functions (TRFs)
We are developing a sister-toolbox for Julia. It is not feature-par with matlab-unfold yet, e.g. splines are missing and no plotting tools. But it allows to specify more flexible basisfunctions, e.g. different types for different events and different lengths. It also fully supports MixedModels (but that part is still untested in practice, besides unittests).
(deprecated) The addon unmixed, allowing to use mixed models can be found in its alpha version version here
Continuous data in EEGLAB 12+ format
Unfold toolbox Download it on GitHub
To get started, best is to start with the 2x2 ANOVA-Design tutorial Quickstart: 2x2 ANOVA