Jemel, B., Achenbach, C., Wiemer, P., Röpcke, B., & Oades,R. D., (2001). Auditory frequency- and duration-deviant detection elicit similar asymmetrical dipole sources localised in both the temporal lobe and in the frontal cortices. NeuroImage, 13, (6 part 2), 323.

Introduction: Event-related potential measures (ERPs) of auditory sensory memory (mismatch negativity, MMN, [Derivation : ERP after a standard tone subtracted from that after a deviant]) and of selective attention (negative difference, Nd, [Derivation : ERP after a non-target subtracted from that after the same tone as a target]) are well-suited for studies of automatic and controlled processing, respectively.
Yet, controversy remains over whether the use of frequency- or duration-deviants produce the more replicable results in test-retest studies and if the sources of both measures are located in the frontal lobe as well as the auditory temporal cortex.
The aim of this study was to see if frontal dipoles contribute to an account of MMN and Nd generation and examine the replicability of MMN and Nd waveforms and dipoles studied a month apart in the same subjects.

Methods: Topographic EEG recordings were made from 32 sites on the scalp of 14 healthy subjects. They were given passive (no task) and active discrimination presentations of a 3-tone oddball paradigm: standard tones, 1.0 kHz, 80 ms, p= 0.82; frequency-deviants, 1.5 kHz, 80 ms, p= 0.06; duration-deviants 1.0 kHz, 40 ms, p= 0.06, repeated after one month. (Novel tones, p=0.06, are not discussed here.)
Dipole sources were modelled using BESA software and localised in Talairach space (Garnero, Baillet and Renault).

the frequency-deviant produced a larger and earlier MMN than that recorded after the duration-deviant: MMN for the duration-deviant correlated more poorly between sessions than MMN for the frequency-deviant and was significantly smaller during the second active condition .
Secondly, Good Nd test-retest reliability was confirmed by high correlation coefficients between sessions: latency correlations between sessions, while significant were more modest.
Thirdly, two MMN dipoles proved to be in the auditory cortex and two were located slightly later in the frontal lobe (left cingulate, right inferior frontal cortex). The low residual variance remaining (ca. 0.8%) was well replicated a) in the majority of subjects, b) for the duration-deviant and c) in the second session a month later.
the Nd had separate sources in the temporal, right parieto-cingulate and left prefrontal cortices.

Conclusions: This study confirms interactions between activity generated in the frontal and auditory temporal cortices in automatic attention-like processes that differ in locus from those involved in controlled processing.
The locus of the right frontal MMN generator is broadly consistent with that attributed to generic deviance detection from a recent imaging report (Strange et al., 2000).
Finally, the lack of a strong interaction between sessions shows that the situation is suitable for pre/post-treatment measures with the larger frequency-deviant response likely to be sensitive to treatment or illness effects on fronto-temporal interactions.

For some of the figures illustrating these results, see: Oades RD, et al., (2001). Sources of prefrontal activity for auditory working memory (MMN): evidence for an impaired fronto-temporal lobe dialogue in schizophrenia. Eighth International Congress: Schizophrenia Research, [Latebreaking data], Whistler, B.C., Canada