Infrequently omitting a stimulus from a repetitive tone sequence elicits MMN when the constant stimulus onset asynchrony separating successive tones is shorter than 200 ms. This phenomenon is usually explained on the basis of the temporal window of integration (TWI). It is assumed that the human auditory system integrates the sounds arriving within a single TWI period into a common trace. Omitting a sound from an isochronous sequence results in the formation of a memory trace that differs in its temporal structure from the regular one and, thus, evokes the MMN response.
However, it is also possible that omissions cause deviation from the regularity of the sound sequence by allowing the after-effect of the stimulus preceding the omission to be integrated into the sensory memory trace of this stimulus, thereby increasing its perceived loudness. Regular sounds of the sequence are perceived as being softer, because the sound following them within the TWI period prevents full integration of their after-effect.
When stimulus onset asynchrony (SOA) is randomly varied within the sound sequence (within the range of the TWI), the two explanations lead to different predictions. 1) On the temporal structure violation explanation, no MMN can be expected for infrequent omissions, because no constant integrated memory trace can be formed in the random-SOA sequence. 2) The loudness summation explanation predicts that MMN will be elicited, because the sound preceding the omission will be louder than those sounds, which are followed by another sound within the TWI period.
Significant MMN responses were observed in both experimental conditions although the MMN amplitude was significantly lower in the random-SOA condition. These results support the loudness-summation interpretation of omission MMN.