What changed, what was learned, what failed.

The research question for Study 01 is fixed: how separating and jointly modelling EEG- and EMG-related information affects decoding performance under reduced, ear-based sensor configurations.

This site went up as a structure rather than a showcase. Most sections are scaffolding with labelled placeholders, because there is nothing verified to put in them yet. The intent is that adding a result later is a content change, not a rebuild.

Next: fix the dataset selection, then write the pre-registration before touching a model.

A recurring temptation in reduced-channel work is to take a handful of scalp electrodes that happen to sit near the ear and describe the result as ear-EEG. The electrode type, the skin contact, the reference placement and the achievable inter-electrode distance are all different in a real around-ear or in-ear array.

Treating a scalp subset as a stand-in gives an upper bound at best. Cerebral will label the distinction everywhere it appears, including on this site.