Did you say brain or brave? Event-related potentials reveal the central role of phonological prediction in false hearing

Relying on predictions can help listeners better comprehend acoustically challenging speech. However, when the speech signal does not match what is predicted, a listener can occasionally experience false hearing (e.g., hearing the word “brain” when “brave” was presented). In the current paper, we report two event-related brain potential (ERP) experiments to examine the time-course of false hearing. Target words of three types—semantically congruent/predictable words, semantically incongruent baseline words, or semantically incongruent phonological lures—were presented either in a cue-target pair (Exp. 1) or in a lexically constraining sentence context (Exp. 2). Phonological lures and incongruent baseline words were equally semantically incongruent but phonological lures shared a similar phonological onset with congruent words. Across both experiments, participants routinely experienced false hearing, reporting the congruent word when the phonological lure was presented. In both experiments, we saw that the N400 to words that would go on to be falsely heard more closely resembled the response to congruent words than to correctly identified phonological lures. Additionally, the N400 response to correctly identified phonological lures was significantly delayed compared to the response to incongruent words that shared no phonological relation to congruent words, suggesting the listeners engaged in phonological prediction. Altogether, the findings from the current study provide evidence that listeners’ engagement in phonological prediction can, in some cases, lead to misperception.