Article Figures & Data
Figures
- Figure 1.
Illustration of spectral and temporal manipulations applied. A , Using an exemplary speech signal, the temporal carrier of the signal as typically derived in noise-band vocoding is filtered, either retaining all natural temporal perturbations in the speech signal (e.g., using a 128 Hz low-pass cutoff, bottom waveform) or in increasing extents of low-pass filtering, which smoothes the temporal carrier and removes typical temporal detail of the speech signal. B , These various filter cutoffs can be combined with the typical spectral degradation by using more or less spectral bands in which the signal is noise vocoded, yielding an orthogonal manipulation of spectral detail (number of bands for vocoding, horizontal) and temporal detail (filter cutoff by which these bands are low-pass filtered, vertical). This combination of temporal smoothing as shown in A and spectral division in varying number of bands allows for a large degree of orthogonality in degrading temporal and spectral detail, as can be observed in the various panels in B . a.u., Arbitrary units.
- Figure 2.
A , B , Results of behavioral pilot sample ( A ) and in-scanner testing ( B ). In both graphs, gray shading going from light to dark indicates increasing (more natural) temporal detail, whereas bar groups indicate increasing (more natural) spectral detail. In behavioral testing (left), a comprehension (word typing) task was used, whereas for in-scanner testing, a four-way rating of comprehensibility was applied (right). See Materials and Methods and Results for statistical details.
- Figure 3.
Display of spectral and temporal variation sensitivity. A , Group statistical results from parametric SPM analyses (see Results for details) thresholded at p < 0.05 (familywise-error correction) are shown for the spectral sensitivity regressor (top) and the temporal sensitivity regressor (bottom). For both conditions, the overlays on sagittal, coronal, and axial (from left to right) slices of a T1-weighted brain template are shown for the respective peak coordinate. L, Left; A, anterior. B , Bar graph reflecting the significant interaction of spectral-temporal sensitivity and left-right hemispheric preference, as observed in the regions of interest analysis.
- Figure 4.
Lateralization and functional hierarchy of spectral, temporal, and monaural conditions. A , Four axial slices through a T1-weighted brain template (left to right slices going from inferior to superior) show overlays of spectral (red) and temporal (yellow) sensitivity as well as the monaural unintelligible condition (green), all thresholded at p < 0.05 (familywise-error correction). L, Left. B , Result of the iterative (jack-knife) reanalysis of the group fMRI data's lateralization quotient (see Materials and Methods for detail). Single data points indicate results of iterations, with one subject left out at a time (hence, n = 15 for all analyses). Note that the mild lateralization to the right (for spectral sensitivity; red) is very stable across the sample, whereas the left lateralization (for temporal sensitivity; yellow) exhibits greater variability depending on the sample configuration. a.u., Arbitrary units.
- Figure 5.
Relationship of comprehension ratings and brain activation. A , Two axial slices approximately through the STS (left) and the middle/inferior temporal gyrus (right) display the maxima of positive correlation with comprehension scores (see Results for details), thresholded at q(FDR) > 0.01. FDR, False discovery rate; L, left. B , For the peak regions of interest for maximal temporal (left) and maximal spectral (right) sensitivity, the monotonic increase in both the percentage signal change estimates (solid lines) and the corresponding comprehension rating scores (dashed lines) for each of the five (temporal and spectral, respectively) degradation levels is plotted. Note that the steeper response curve in the spectral conditions corroborates the steeper change in comprehension rating scores. a.u., Arbitrary units.
Tables
- Table 1.
Overview of significant clusters in random-effects analysis (familywise-error corrected, p < 0.05)
Site MNI coordinates Z Extent (mm3) Spectral variation Right middle lateral STG (BA 22) 56 −18 4 6.09 2432 Left middle STG/STS (BA 22) −60 −8 −4 5.98 1664 Temporal variation Left middle STS (BA 21) −56 −12 −4 5.61 312 Right middle STG/STS (BA 22) 54 −12 −4 5.41 232 Monaural stimulation (right ear, control) Left posterior STG (BA 42) −58 −34 10 5.5 376 Correlation with comprehension rating Right transverse temporal gyrus (BA 42) 62 −12 8 4.99 3632 Left MTG −50 −36 −6 4.75 776 Left IFG (BA 9) −48 4 −22 4.57 520 Left STG (BA 22) −62 −16 0 4.57 1688 -
Specifications refer to peak voxels. IFG, Inferior frontal gyrus; MTG, middle temporal gyrus.
-
