Predictability of stimulus

The mismatch negativity (MMN), a component of event-related brain potentials (ERPs), generally reflects automatic change detection of stimulus characteristics associated with auditory sensory memory.1–2 A common way to elicit a MMN is to randomly present infrequent tonal stimuli mixed with frequent tonal stimuli. The infrequent stimuli, detected as being different from the frequent stimuli, elicit a negativity generated in supratemporal auditory cortex which is maximal at the frontocentral scalp locations and often inverts in polarity at scalp locations below the sylvian fissure.3–4 Unlike the P3 component,5 the MMN has previously been reported to be indifferent to the regular occurrence of deviant stimuli within a tonal sequence. Scherg et al.6 presented sequences of 400 ms tones at a stimulus onset asynchrony (SOA) of 1.3 s in two conditions. In the regular condition the deviant stimuli were presented systematically, after every fourth standard (i.e. SSSSDSSSSD). In the irregular condition the deviant stimuli were presented at the same probability (0.2) but were randomly dispersed among the standard stimuli. They found no difference in the MMN amplitude or latency obtained to deviant stimuli occurring either regularly or irregularly among the standard stimuli. That is, the manner of stimulus delivery did not affect the MMN. This finding suggests that the MMN is insensitive to the predictable occurrence of the deviant stimuli. Determination of what is deviant in the recent auditory input first assumes that the MMN system has determined what is invariant. Since at least two consecutive repetitions of a tone are required before a frequency MMN is elicited,7 it can be inferred that the mechanisms underlying MMN generation detect invariance of the acoustic input and record this information in a transient auditory memory. If we extend the finding of Winkler et al.7 to multi-tone patterns (i.e. two repetitions of a pattern are needed to establish the standard representation in transient memory), then in the study of Scherg et al.6 11.7 s would have elapsed for the presentation of two repetitions of SSSSD. Since it has been estimated that the duration of the memory underlying MMN generation is 10 s,8 the MMN system may not have been capable of noting the regularity of the repeating pattern since 11.7 s may extend beyond the approximate limits of the memory associated with the MMN. It is not known, however, how many repetitions are needed to elicit a MMN when the standard consists not of a single tone, but of a pattern of tones in a sequence. The purpose of the current study was to reexamine the idea that the regularity of the stimulus presentation has no effect on how the deviants are processed in terms of the MMN system. We hypothesized that if the SSSSD pattern could be detected as a regularly repeating pattern of five tones, then no MMN would be elicited by the D position tone. We therefore decreased the SOA to 100 ms so that many repetitions of the five tone (SSSSD) pattern would occur within the limits of sensory memory for the current study. Additionally, we predicted that if the D tones were randomly interspersed among the S tones (using the same probabilities and the shorter Cognitive Neuroscience