Auditory object segmentation amplifies prediction error signals in a complex MMN paradigm

Auditory deviance detection depends on the segmentation of sound sequences into perceptually meaningful units, which provide the basis for predictive models. Complex mismatch negativity (MMN) paradigms, including dual-rule designs, place higher demands on these processes because regularities are defined by relationships across tones rather than by single features. However, it remains unclear how facilitation of auditory segmentation influences MMN. To address this, we compared MMN elicited during a dual-rule paradigm under continuous versus temporally segmented (750 ms gap) conditions in thirty-two participants using simultaneous EEG and MEG recordings. Two 50-ms binaural pure tones were presented in pairs. Tone A was a 1 kHz stimulus biased to the right ear, and Tone B was a 1.2 kHz stimulus biased to the left ear. Within each pair, tones followed with a 330 ms Stimulus Onset Asynchrony (SOA), with standards consisting of A-B pairs (85.6%) and deviants of A-A pairs (14.3%). Two conditions were used: a No-Gap condition, where pairs followed continuously with a 330 ms SOA, and a Gap condition, where pairs were separated by a longer 750 ms SOA. Results at EEG FCz revealed dual-rule MMN amplitudes (130-230 ms) were not significantly different between No-Gap and Gap conditions (t₍₂₉₎ = 0.93; p = 0.36; d = 0.17). Conversely, MEG source-localized responses in bilateral primary auditory cortex (A1) revealed larger MMN responses in the Gap condition relative to No-Gap (F _(1,31) = 26.07, p < 0.001, ${\eta }_p^2$ = 0.457). This suggests that temporal segmentation facilitates the grouping of tone pairs into perceptual objects, allowing the auditory system to form predictions not only at the level of individual tones, but also at the level of the tone pair as a perceptual unit. Deviants in this context violate both low-level features (pitch and location) and the higher-order pair structure, producing larger prediction error signals. These results provide insight into how the auditory system integrates local and higher-order regularities, offering a mechanistic link between auditory scene analysis and hierarchical predictive coding in complex auditory environments.