Effects of deviant probability on the 'optimal' multi-feature mismatch negativity (MMN) paradigm

The introduction of the multi-feature 'optimal' mismatch negativity (MMN) paradigm was a significant innovation that allows for the collection of MMNs from five different deviant types in a relatively short period of time. Given the very specific stimulus presentation structure of this paradigm, the deletion of deviants results in an increase in the stimulus probability of each remaining deviant while holding constant the probability of the standard stimulus. The focus of this paper is two-fold: first, to examine the effects of altering deviant probability, but not standard probability, on MMN generation and, secondly, to validate a shorter, 3-stimulus version of the 'optimal' paradigm.

Twenty-four participants were presented either with the original 5-stimulus 'optimal' multi-feature MMN paradigm (Optimal-5) or a modified, 3-stimulus version containing only duration, frequency and intensity deviants (Optimal-3). MMN amplitudes and latencies to common deviant types were compared between groups.

MMN amplitudes elicited by Optimal-3 were significantly smaller than those elicited by Optimal-5 for frequency and intensity deviants. There were no significant differences between MMN amplitudes for duration deviants, nor were there any significant MMN latency differences for any of the three deviant types.

These results support previous suggestions that MMN attenuation following increased deviant presentation probability is due to the development of separate deviant memory traces, not due to a weakening of the standard memory trace. Furthermore, while MMN amplitudes to Optimal-3 were reduced, they were still present; the Optimal-3 paradigm may be a useful tool for eliciting MMNs in populations unable to tolerate long test sessions, such as acute schizophrenia.