New paper in Neu­roim­age by Fiedler et al.: Track­ing ignored speech mat­ters

Lis­ten­ing requires selec­tive neur­al pro­cess­ing of the incom­ing sound mix­ture, which in humans is borne out by a sur­pris­ing­ly clean rep­re­sen­ta­tion of attend­ed-only speech in audi­to­ry cor­tex. How this neur­al selec­tiv­i­ty is achieved even at neg­a­tive sig­nal-to-noise ratios (SNR) remains unclear. We show that, under such con­di­tions, a late cor­ti­cal rep­re­sen­ta­tion (i.e., neur­al track­ing) of the ignored acoustic sig­nal is key to suc­cess­ful sep­a­ra­tion of attend­ed and dis­tract­ing talk­ers (i.e., neur­al selec­tiv­i­ty). We record­ed and mod­eled the elec­troen­cephalo­graph­ic response of 18 par­tic­i­pants who attend­ed to one of two simul­ta­ne­ous­ly pre­sent­ed sto­ries, while the SNR between the two talk­ers var­ied dynam­i­cal­ly between +6 and −6 dB. The neur­al track­ing showed an increas­ing ear­ly-to-late atten­tion-biased selec­tiv­i­ty. Impor­tant­ly, acousti­cal­ly dom­i­nant (i.e., loud­er) ignored talk­ers were tracked neu­ral­ly by late involve­ment of fron­to-pari­etal regions, which con­tributed to enhanced neur­al selec­tiv­i­ty. This neur­al selec­tiv­i­ty, by way of rep­re­sent­ing the ignored talk­er, pos­es a mech­a­nis­tic neur­al account of atten­tion under real-life acoustic con­di­tions.

The paper is avail­able here.