Author: Jonas

Objec­tive. Con­ven­tion­al, mul­ti-chan­nel scalp elec­troen­cephalog­ra­phy (EEG) allows the iden­ti­fi­ca­tion of the attend­ed speak­er in con­cur­rent-lis­ten­ing (‘cock­tail par­ty’) sce­nar­ios. This implies that EEG might pro­vide valu­able infor­ma­tion to com­ple­ment hear­ing aids with some form of EEG and to install a lev­el of neu­ro-feed­back. Approach. To inves­ti­gate whether a listener’s atten­tion­al focus can be detect­ed from sin­gle-chan­nel hear­ing-aid-com­pat­i­ble EEG con­fig­u­ra­tions, we record­ed EEG from three elec­trodes inside the ear canal (‘in-Ear-EEG’) and addi­tion­al­ly from 64 elec­trodes on the scalp. In two dif­fer­ent, con­cur­rent lis­ten­ing tasks, par­tic­i­pants ( n  =  7) were fit­ted with indi­vid­u­al­ized in-Ear-EEG pieces and were either asked to attend to one of two dichot­i­cal­ly-pre­sent­ed, con­cur­rent tone streams or to one of two diot­i­cal­ly-pre­sent­ed, con­cur­rent audio­books. A for­ward encod­ing mod­el was trained to pre­dict the EEG response at sin­gle EEG chan­nels. Main results. Each indi­vid­ual par­tic­i­pants’ atten­tion­al focus could be detect­ed from sin­gle-chan­nel EEG response record­ed from short-dis­tance con­fig­u­ra­tions con­sist­ing only of a sin­gle in-Ear-EEG elec­trode and an adja­cent scalp-EEG elec­trode. The dif­fer­ences in neur­al respons­es to attend­ed and ignored stim­uli were con­sis­tent in mor­phol­o­gy (i.e. polar­i­ty and laten­cy of com­po­nents) across sub­jects. Sig­nif­i­cance. In sum, our find­ings show that the EEG response from a sin­gle-chan­nel, hear­ing-aid-com­pat­i­ble con­fig­u­ra­tion pro­vides valu­able infor­ma­tion to iden­ti­fy a listener’s focus of attention.