Are you interested in analyzing EEG when one reads a book, listens to music while walking around, navigates a city or appreciates a piece of art?
Such naturalistic paradigms, where the experimenter does not control the subjects sensory input, are becoming more popular, but offer very difficult challenges for data analysis and interpretation.
I will discuss two of the main problems:

1.  If two events (e.g. fixations on a word) follow each other too quickly, the brain response of the first event will overlap with the response to the second one.
   This is also true for traditional designs, e.g. the response to a stimulus will overlap the response of a button press.
   Furthermore, in multimodal research you might have auditory and visual inputs in the same experiment, which will also overlap.
2. There are numerous uncontrollable variables in naturalistic paradigms, for instance saccade amplitudes, response times, fixation position, or local luminance.
   These variables might be different between our conditions of interest and we have to adjust for that.

In this talk, I will present an integrated framework to analyze EEG data from such paradigms influenced by these problems.
I will be using the unfold toolbox (www.unfoldtoolbox.org) which is based on linear deconvolution and non-linear multiple regression (GAM).
The toolbox offers intuitive syntax using formulas (e.g ERP ~ this + cat(is) + spl(endid,5)) and plenty of tutorials and documentation.
I will introduce the problem and their proposed solution in an intuitive and tutorial-style.
I will use combined EEG/Eye Tracking data as an exemplary case.