Background: The event-related brain response mismatch negativity (MMN) registers changes in auditory stimulation with temporal lobe sources reflecting short-term echoic memory and frontal sources a deviance-induced switch in processing. Impairment, controversially present at the onset of schizophrenia, develops rapidly and can remain independent of clinical improvement. We examined the characteristics of the scalp-recorded MMN and related these to tests of short-term memory and set-shifting. We assessed whether the equivalent dipole sources are affected already at illness-onset in adolescence and how these features differ after a 14-year course following an adolescent onset. The strength, latency, orientation and location of frontal and temporal lobe sources of MMN activity early and late in the course of adolescent-onset schizophrenia are analysed and illustrated.

Methods: MMN, a measure of auditory change-detection, was elicited by short deviant tones in a 3-tone oddballpresentation and recorded from 32 scalp electrodes. Four dipole sources were placed following hypothesis-led
calculations using brain electrical source analysis on brain atlas and MR-images. A short neuropsychological test battery was administered. We compared 28 adolescent patients with a first episode of schizophrenia and 18 patients 14 years after diagnosis in adolescence with two age-matched control groups from the community (n = 22 and 18, respectively).

Results: MMN peaked earlier in the younger than the older subjects. The amplitude was reduced in patients, especially the younger group, and was here associated with negative symptoms and slow set-shifting. In first-episode patients the temporal lobe sources were more ventral than in controls, while the left cingular and right inferior-mid frontal sources
were more caudal. In the older patients the left temporal locus remained ventral (developmental stasis), the right temporal locus extended more antero-laterally (illness progression), and the right frontal source moved antero-laterally (normalised).

Conclusion: From the start of the illness there were differences in the dipole-model between healthy and patient groups. Separate characteristics of the sources of the activity differences showed an improvement, stasis or deterioration with illness-duration. The precise nature of the changes in the sources of MMN activity and their relationship to selective information processing and storage depend on the specific psychopathology and heterogeneous course of the illness.