These "no precedent" trials served as baseline conditions for the same- and different-speaker conditions, respectively.
Note that observing the same basic pattern of theta-power effects in the interaction as well as in the simple contrast rules out the possibility that a negative effect for the different-speaker conditions may have manifested as an overall positive effect compared with the same-speaker conditions, thus, potentially driving the interaction effect.
Thirteen participants completed the multi-speaker condition and the remaining 13 participants completed the mono-speaker condition, with condition assignment randomized across participants.
In the multi-speaker condition, scenes were pseudo-randomly mapped to one of three spatial regions across the nine speaker array, namely (1) a central region of speakers straddling 315° to 45°, (2) a leftward region straddling 270° to 0°, and (3) a rightward region straddling 0° to 90°.
Because listeners knew in advance of each test phase which speaker would be speaking, we expected stronger retrieval of the precedent in the same-speaker condition (i.e., speaker identity serving as a further retrieval cue).
Insofar as the expectation for a certain name is stronger in the same-speaker condition (because of a tighter contextual similarity), the unexpectedness of the mismatch might be most prominent in the same-speaker, mismatch condition and therefore lead to the strongest TPJ response in that condition.
This analysis revealed significant interactions between speaker and precedent that showed a strikingly similar pattern (1 spatially distributed cluster, P = 0.008; Fig. 4, Panel B; Supplementary Table S1, Panel B) to that of the simple contrast in the same-speaker condition described earlier.
A time-frequency analysis at the sensor level between −800 and 1000 ms in the range of 2 30 Hz revealed a significant cluster (P = 0.012) in the theta range (4 6 Hz) for the precedent mismatch vs. no precedent comparison within the same-speaker condition, in a time window around 350 650 ms after naming onset (Fig. 3, top row).
For each test utterance, speaker, or condition, a linear transform is estimated to adapt the test features towards the clean acoustic model.
C To indicate that the speaker's condition happens unexpectedly.
The President pro tempore of the Senate has survived (with minor injuries) and, on learning of the Speaker's condition, urges that he be sworn in as Acting President for the sake of providing leadership to a stricken nation - just as soon as they can find the Chief Justice of the Supreme Court to provide a favorable interpretation of the law.
