“What” and “when” predictions modulate auditory processing in a mutually congruent manner

Extracting regularities from ongoing stimulus streams to form predictions is crucialfor adaptive behavior. Such regularities exist in terms of the content of the stimuliand their timing, both of which are known to interactively modulate sensoryprocessing. In real-world stimulus streams such as music, regularities can occurat multiple levels, both in terms of contents (e.g., predictions relating to individualnotes vs. their more complex groups) and timing (e.g., pertaining to timingbetween intervals vs. the overall beat of a musical phrase). However, it is unknownwhether the brain integrates predictions in a manner that is mutually congruent(e.g., if “beat” timing predictions selectively interact with “what” predictions fallingon pulses which define the beat), and whether integrating predictions in differenttiming conditions relies on dissociable neural correlates. To address thesequestions, our study manipulated “what” and “when” predictions at different levels– (local) interval-defining and (global) beat-defining – within the same stimulusstream, while neural activity was recorded using electroencephalogram (EEG) inparticipants (N = 20) performing a repetition detection task. Our results revealthat temporal predictions based on beat or interval timing modulated mismatchresponses to violations of “what” predictions happening at the predicted timepoints, and that these modulations were shared between types of temporalpredictions in terms of the spatiotemporal distribution of EEG signals. Effectiveconnectivity analysis using dynamic causal modeling showed that the integrationof “what” and “when” predictions selectively increased connectivity at relativelylate cortical processing stages, between the superior temporal gyrus and thefronto-parietal network. Taken together, these results suggest that the brainintegrates different predictions with a high degree of mutual congruence, but in ashared and distributed cortical network. This finding contrasts with recent studiesindicating separable mechanisms for beat-based and memory-based predictiveprocessing.