Abstract

Recent interest in loudness dependency of auditory evoked potentials (LDAEP) owes largely to its promise as a predictor of clinical response to treatment with serotonin agonists. Originally based on peak-to-peak N1/P2 differences, LDAEP has been quantified by inverse models as N1 activity attributable to primary auditory cortex (e.g., BESA tangential dipole; LORETA current density). However, reference-free, scalp-recorded current source density (CSD; surface Laplacian) and temporal PCA together allow simplification and separation of overlapping component waveforms and topographies that are obscured in ERPs by volume-conduction and superposition. Because CSD-PCA provides a conservative, model-independent approach to the problem of generator localization, we used unrestricted, Varimax-rotated temporal CSD factors (covariance matrix) to quantify scalp current source and sink patterns underlying LDAEP. Auditory ERPs were recorded from a 72-channel BioSemi ActiveTwo system in 16 healthy adults while they listened to pure (1000 Hz), 40-ms binaural tones (1600-2100 ms ISI) presented at five, equiprobable intensities (60-100 dB SPL). An N1 factor (loadings peak: 116 ms) with the expected tangentially-oriented sink/source topography consistent with activation of auditory cortex showed a robust, monotonic association with intensity. Likewise, a P2 factor (222 ms) with a midline source topography also showed intensity effects; however, these were less robust and included variable secondary sources at frontotemporal sites. In contrast, no significant association with loudness was observed for a radial temporal lobe N1 sink (167 ms) or for a temporal lobe P3 source (343 ms). Moreover, loudness dependency of N1 was not statistically improved for combined topographies (e.g., sums/differences of N1 with P2/temporal N1). Thus, CSD-PCA offers a concise and efficient, but conservative, characterization of LDAEP generators. [Supported by NIMH grants MH036295 and MH066597.]