Figure 1.
Experimental design and spectral analysis.
A. Stimulus – the subjects heard a dynamic ascending pure tone chirp, which repeated 15 times (stimulus repetition frequency 0.033 Hz). B. Each voxel's time-course was Fourier transformed. Presented here is the normalized amplitude of the spectrum of a voxel sampled from Heschl's gyrus (HG) of a representative subject. Amplitude at stimulus repetition frequency is marked with a red circle. The voxel's phase at that frequency corresponds to the preferred tone (auditory frequency) of the voxel. C. Amplitude and phase parameters were used to construct a pure cosine used as a model of the activation. The original raw time-course of two voxels, one from HG and one from the superior temporal sulcus (STS) are drawn in red; the dashed black line shows the model for each voxel. Pearson correlation coefficients were calculated to estimate the significance of the response of each voxel, and phase maps were inspected only in regions showing high and significant correlations. D. Mean correlation coefficient (Pearson's R) map of 10 subjects, presented on a partly inflated left cortical hemisphere of the standard MNI brain. Most of the auditory cortex is marked with high R values (marked red, R(299)>0.23, p<0.05 Bonf. Corr.). Within this region R values are the highest in the core area (marked in yellow, R(299)>0.3, p<0.00005 Bonf. Corr.), including HG (marked in green) and its surroundings. For a presentation of Pearson's R values in a horizontal slice view see Fig. S2C. HS – Heschl's sulcus, STG – Superior temporal gyrus, STS – Superior temporal sulcus. E. Group (Session 1, n = 10) relative frequency preference map is presented in a lateral view of the partly inflated left cortical hemispheres of the standard MNI brain. The map within the auditory-responsive region shows multiple iso-frequency bands, in addition to the mirror- symmetric cochleotopic maps in the auditory core area on the superior temporal plane. These iso-frequency bands extend in a superior-to-inferior axis along the temporal cortex to the superior temporal sulcus.
Figure 2.
Multiple mirror-symmetric cochleotopic maps in the left hemisphere of the human auditory cortex.
Group (Exp. 1, n = 10) relative frequency preference map is presented in a lateral view of the inflated left cortical hemisphere of the standard MNI brain, exposing the entire cochleotopic organization of the multiple iso-frequency bands (STG – Superior temporal gyrus, STS – Superior temporal sulcus). On the left panels, the auditory cortex region is magnified, showing Exp. 1 relative frequency preference map on the cortical surface. The estimated border between the putative mirror- symmetric cochleotopic maps is indicated (white line) in the lowest and highest frequency tones which represent the mirror-symmetry flipping lines between the homologues of A1 and R in the core auditory cortex, and between multiple additional cochleotopic fields. Numbers indicate points along the cochleotopic gradients (similar to those depicted in Fig. 6, from which raw time-courses of activation were sampled), with white arrows demonstrating the gradient direction in each filed. On the lower panels, the same gradients are depicted in volume views in horizontal (z = 2, −1), and coronal (y = −21, −32) slices, numbered similarly to the surface view (for demonstrative and orientation purposes only), to enable the identification of the same gradients in the three-dimensional based views.
Figure 3.
Multiple mirror-symmetric cochleotopic maps in the right hemisphere of the human auditory cortex.
Group (Exp. 1, n = 10) relative frequency preference map is presented in a lateral view of the inflated right cortical hemisphere of the standard MNI brain, exposing the entire cochleotopic organization of the multiple iso-frequency bands (STG – Superior temporal gyrus, STS – Superior temporal sulcus). On the left panels, the auditory cortex region is magnified, showing Exp. 1 relative frequency preference map on the cortical surface. The estimated border between the putative mirror- symmetric cochleotopic maps is indicated (white line) in the lowest and highest frequency tones which represent the mirror-symmetry flipping lines between the homologues of A1 and R (and possibly, anterior to it, RT) in the core auditory cortex, and between multiple additional cochleotopic fields. Numbers indicate points along the cochleotopic gradients (similar to those depicted in Fig. 6, from which raw time-courses of activation were sampled), with white arrows demonstrating the gradient direction in each filed. On the lower panels, the same gradients are depicted in volume views in horizontal (z = 4, 0), and coronal (y = −28, −22) slices, numbered similarly to the surface view (for demonstrative and orientation purposes only), to enable the identification of the same gradients in the three-dimensional based views.
Figure 4.
Consistency of the mirror-symmetric cochleotopic maps across experiments and analyses.
A. (left hemisphere) and D. (right hemisphere) display, on the left column, auditory cortex relative frequency preference map magnification as seen in Figs. 2 (LH) and 3 (RH), showing the mirror- symmetric cochleotopic maps inspected using spectral analysis. In the middle column cross-correlation analysis for the averaged single-subject time-course is displayed, showing remarkably similar trends to that of the spectral analysis. On the right column, the continuous auditory stimulation was analyzed by dividing it in a random effect general linear model (RFX-GLM) into low, medium and high frequency tone conditions. The GLM map displays the contrast of each frequency band with the other conditions. B. (left hemisphere) and E. (right hemisphere) display cross-correlation (middle) and GLM (right) analyses for Exp. 2 (n = 5) in which the chirp was reversed (i.e. from high to low frequencies). Spectral analysis (left) is the averaged phase map of Exp. 2 with Exp. 1, thus fully controlling for the hemodynamic delay of both experiments ([52], [53]; for the spectral maps of Exp. 2 alone see Fig. S5). The consistency of these results with the main experiment show that the auditory fields and cochleotopic gradients displayed for Exp. 1 do not result from the frequency modulation direction. C. (left hemisphere) and F. (right hemisphere) display spectral (left), cross-correlation (middle) and GLM (right) analyses for Exp. 3, in which a subgroup (n = 4) of subjects was scanned again one month after the original scan, revealing similar patterns of iso-frequency bands as the original (first scan) map, demonstrating the high test-retest reliability of the auditory fields and their locations. See also Fig. S8 for further single subject analysis of this experiment. G. Similarity alignment histograms are presented for three main contrasts, between the main experiment (Exp. 1) and the two control experiments (Exps. 2 and 3) and between the spectral and cross-correlation analyses in Experiment 1, for both hemispheres. The distribution of each comparison's alignment indexes (between 0 and 1 in each comparison) show a sharp peak towards 1, demonstrating their significance, and differ significantly (p<0.00001 in all comparisons presented) from random maps indexes (marked on the histograms with red line for comparison). Therefore, the similarity indexes of the correspondence between the relative frequency preference maps across analyses and experiments support the high replicability of the cochleotopic maps.
Figure 5.
Multiple cochleotopic maps in single subjects.
Anatomical structures of the horizontal views of each subject in the magnified area of the auditory cortex, unsmoothed spectral analysis relative frequency preference maps (individual R>0.18, df = 299, p<0.05, corrected for multiple comparisons) and cross-correlation maps (p<0.05, corrected for multiple comparisons) and shown for four different subjects (these and similar maps from additional horizontal slices are presented in Fig. S6). Overlaid on the spectral analysis maps are numbers (3, 6) representing the low-frequency peaks corresponding to those presented in the group results. Point 3 corresponds to the border between A1 and R and point 6 represents the low frequency band on the lateral STG possibly corresponding to a homologue of area CL of the belt. Single subject maps show cochleotopic maps that extend beyond the auditory core to the superior temporal gyrus and superior temporal sulcus. HG - Heschl's gyrus, STS – Superior temporal sulcus.
Figure 6.
Cochleotopic maps in the human auditory cortex verified by RFX-GLM raw time-course analysis.
A. Auditory cortex relative frequency preference map magnification is the same as in Figs. 2 and 3, with points (1–5, see Talairach coordinates at Table 1) along the auditory core gradients that were used to sample individual time-courses and compute random effect GLM time-course analysis. The approximate location of the same sampling points is also presented in a volume view of sagittal and horizontal slices. B. Time-courses of activation and response averages were sampled from points (1–5) along the anterior-posterior cochleotopic gradient (of the core areas), in both cortical hemispheres. Response averages were calculated for Exp. 1 group (n = 10), Exp. 2 group (falling chirp, n = 5) and for Exp. 3 group (scan repetition, n = 4) from the same locations. The continuous auditory stimulation was analyzed by dividing it in a random effect general linear model (RFX-GLM) into low, medium and high frequency tone conditions. Error bars denote standard error of the mean (SEM). Tone preference examined using this complementary analysis was consistent with relative frequency preference maps revealed by spectral analysis. C. Auditory cortex relative frequency preference map magnification is the same as in Figs. 2 and 3, with points (6–9) marking the lowest and highest frequency tones which represent the mirror-symmetry flipping points between the extra-core cochleotopic maps (see Talairach coordinates at Table 1). These points were used to sample individual time-courses and compute random effect GLM time-course analysis, similarly to A-B. The approximate location of the same sampling points is also presented in a volume view of sagittal and horizontal slices, with reference to the sampling points of the core. D. Similar to B, response averages of activation were sampled from points 6–9 in the left hemisphere, and 6–8 in the right hemisphere, along the superior-inferior cochleotopic gradient in both scan sessions, validating the tone preference of the iso-frequency bands in the extra-core areas of auditory cortex.
Table 1.
Talairach coordinates of mirror-symmetry flipping points between the cochleotopic maps in the temporal lobe.