Topographic and temporal indices of vowel spectral envelope extraction in the human auditory cortex

The auditory-evoked neuromagnetic field elicited by single vowel formants and two-formant vowels was recorded under active listening conditions using a 37-channel magnetometer. There were three single formants with formant frequencies of 200, 400, and 800 Hz, another single formant with a formant frequency of 2600 Hz, and three vowels that were constructed by linear superimposition of the high- onto one of the low-frequency formants. P50 m and N100 m latency values were inversely correlated with the formant frequency of single formants. A strong effect of formant frequency on source location was obtained along the postero-anterior axis, which is orthogonal to the well-established latero-medial tonotopic gradient. Regardless of whether single formants or first formants of vowels were considered, N100 m sources were more anterior and sustained field sources were more posterior for higher-frequency than for lower-frequency formants. The velocity of the apparent posterior-to-anterior movement across cortical surface of N100 m sources first reported by Rogers et al. [Rogers, R. L., Papanicolaou, A. C., Baumann, S. B., Saydjari, C., & Eisenberg, H. M. (1990). Neuromagnetic evidence of a dynamic excitation pattern generating the N100 auditory response. Electroencephalography and Clinical Neurophysiology,77, 237-240] decreased as a function of latency. The amount of deceleration was positively correlated with formant frequency. Responses to the vowels were superadditive, indicating that the processes elicited by the constituents of composite stimuli interact at one or more stages of the afferent auditory pathway. Such interaction may account for the absence of a lateral-to-medial tonotopic mapping of first formant frequency. The source topography found may reflect activity in auditory fields adjacent to AI with the strength of the contribution varying with formant frequency. Alternatively, it may reflect sharpness-of-tuning and inhibitory response-area asymmetry gradients along isofrequency stripes within AI. Either alternative may be interpreted in terms of a spectral blurring mechanism that abstracts spectral envelope information from the details of spectral composition, an important step towards the formation of invariant phonetic percepts.