The representation of unattended, segmented sounds: a mismatch negativity (MMN) study

The detection of an irregular, potentially relevant change (deviance) in the regular, unattended acoustic environment is ensured by the automatic deviance detection mechanism. It underlies the formation of a regularity representation and a comparison of an incoming sound with this representation. A mismatch outcome of this comparison evokes the mismatch negativity (MMN) of the event-related potential. For unattended pure tones the automatic deviance detection mechanism operates most efficiently for initial sound parts, which is why these are suggested to contribute more to sound representation than later parts. A transient that physically segments the sound can overcome this temporal constraint in sound representation. Whether the resulting individual (initial and terminal) sound segments or the joined two-segments give rise to the regularity representation is addressed here. We took advantage that the MMN attenuation to the second of two successive deviances (deviance-repetition effect) is more pronounced when the deviances belong to the same unit of representation. We measured MMN for two deviances (frequency modulations) within segmented sounds that either occurred within the initial or the terminal segment, or that were split across both segments. Unexpectedly, we did not obtain a deviance-repetition effect. Instead, we obtained a temporal distance effect: With increasing temporal distance from deviance-onset relative to segment-onset the MMN amplitude decreased. Furthermore, this effect did not depend on whether the deviance occurred in the initial or in the terminal segment. Thus, (for the current approach) we suggest that the regularity representation is based on the individual rather than joined segments.