TTTth LANGUAGE Functional Parcellation of Human Auditory Cortex: Sensitivity to Intensity and Spectral Pattern Information J.R. Binder, J.N. Kaufman, E.T. Possing, E. Liebenthal, R. Tong, B.D. Ward Depurtment of Neurology and Biophysics Institute; Medical College of Wisconsin; Milwaukee, WI, USA Studies in nonhuman primates suggest a hierarchical subdivision of auditory cortex into core, belt, and parabelt systems [I]. Core areas in humans show typical mammalian tcnotopicity, but the overall functional organization of human auditory cortex has not been described. Our goal was to begin a parcellation of this system using functional magnetic resonance imaging. Sensitivity to sound intensity was used as a marker for tonotopically organized cortex, based on the priniciple of level-dependent frequency tuning [2]. Given the importance of spectral analysis to human speech perception, we hypothesized that areas within auditory association cortex would show sensitivity to the presence or absence of spectral pattern information. Methods: FMRI was conducted at 3 Tesla. Clustered acquisition echoplanar imaging was performed at IO-second intervals to avoid masking of the stimuli or contamination of the activation data by scanner acoustic noise. Subjects performed a simple detection response at the onset of each stimulus train. Intensity Experimenr. 14 subjects were studied. The stimulus was a sinusoidal frequency-modulated tone. Intensity of the tone was systematically varied from 30 to 70 dB SPL. Regression analysis was used to identify voxels in which BOLD signal increased as a function of stimulus intensity. Spectral Partern Experiment. 12 subjects were studied. Stimuli were distorted nonword speech sounds containing variable amounts of spectral information. Following the method of Shannon et al. [4], these were created by extracting amplitude envelopes of the stimuli over spectral regions of variable width, which were then used to modulate the amplitude of bandpassed noise limited to the same spectral regions. Finally, these AM noise bands were combined to produce stimuli with 1, 2, 3, or 4 bands of spectral resolution. Regression analysis was used to identify voxels in which BOLD signal increased as a function of stimulus spectral resolution. Average activation maps for both experiments were created by thresh- olding the individual regression maps at a voxelwise p < ,005 and measuring the degree of overlap of the thresholded maps in standard stereotaxic space. Results: All stimuli activated extensive regions of auditory cortex bilaterally in all subjects. Voxels demonstrating specific sensitivity to sound intensity were observed in the transverse temporal sulcus (TTS), caudomedial Heschl’s gyrus (HG), and caudo- medial planum temporale. The area of maximal overlap of these intensity-sensi- tive regions was in the caudomedial aspect of the TTS and HG (black areas in Figure), with a peak at stereotaxic coordinate -44;29,lO. Voxels showing specific sensitivity to spectral pattern information were observed in the rostrolateral ‘ITS, HG, and planum temporale, with maximal overlap in the rostrolateral aspect of ‘ITS and HG (white areas in Figure), and peaks at -53,-l&6 and -62,-17,6. These areas demonstrating intensity sensitivity and sensitivity to spectral information were adjacent on the dorsal temporal surface in clearly distinct, non-overlapping locations (Figure). Discussion: Auditory areas sensitive to intensity (black) and spectral information (white). Auditory areas sensitive to intensity (black) and spectral information (white). Core auditory cortex occupie> the caudomedial aspect of HG in humans 131. Neurons in the auditory core of primates exhibit broad, Irvel-dependent frequency runing: thus. stimulation JI higher level\ activate5 a larger extent of these tonotopic fields [2]. As predicted from these facts. activation of caudomrdial HG and adjacent TI’S was dependent on sound intensity. In contrast. areas in the rostrolateral aspect of HG and TTS were unaffected by the intensity manipulation. Unlike the core regions, these areas showed sensitivity to the amount of information present in the srimulus within the spectral domain. The double dissociation in response patterns exhibited by these two cortical regions suggests that they are functionally distinct components of human auditory cortex. References I. Kaas JH, Hackett TA (1998). Audio1 Neurootol 3: 73-85. 2. Phillips DP, Semple MN, Calford MB, Kitzes LM (1994). Exp Brain Res 102: 210-226. 3. Rademacher J, Caviness VS, Steinmetz H, Galaburda AM (1993). Cereb Cortex 3: 313-329 4. Shannon RV, Zeng F-G, Kamath V, Wygonski J, Ekelid M (1995). Science 270: 303-304. S296