
==== Front
PLoS Biol
PLoS Biol
plos
PLOS Biology
1544-9173
1545-7885
Public Library of Science San Francisco, CA USA

39241107
10.1371/journal.pbio.3002774
PBIOLOGY-D-24-00402
Short Reports
Social Sciences
Linguistics
Grammar
Phonology
Syllables
Engineering and Technology
Electronics Engineering
Electronics
Biology and Life Sciences
Cell Biology
Cellular Types
Animal Cells
Neurons
Biology and Life Sciences
Neuroscience
Cellular Neuroscience
Neurons
Social Sciences
Linguistics
Phonetics
Vowels
Engineering and Technology
Signal Processing
Engineering and Technology
Signal Processing
Signal Amplification
Research and Analysis Methods
Spectrum Analysis Techniques
Gamma Spectrometry
Biology and Life Sciences
Physiology
Electrophysiology
Membrane Potential
Evoked Potentials
Biology and Life Sciences
Physiology
Electrophysiology
Neurophysiology
Evoked Potentials
Biology and Life Sciences
Neuroscience
Neurophysiology
Evoked Potentials
Syllable processing is organized in discrete subregions of the human superior temporal gyrus
Surface organization of human cortical speech processing
https://orcid.org/0000-0001-7112-4519
Cleary Daniel R. Conceptualization Data curation Formal analysis Funding acquisition Investigation Methodology Project administration Writing – original draft Writing – review & editing 1 2 *
Tchoe Youngbin Conceptualization Data curation Formal analysis Investigation Methodology 3
Bourhis Andrew Data curation Formal analysis Investigation Methodology 3
Dickey Charles W. Conceptualization Data curation Formal analysis Investigation Methodology 4
Stedelin Brittany Data curation Investigation Methodology 2
Ganji Mehran Investigation Methodology 3
Lee Sang Heon Data curation Investigation Methodology Resources 3
Lee Jihwan Data curation Investigation Methodology 3
Siler Dominic A. Data curation Investigation Methodology 2
Brown Erik C. Conceptualization Data curation Investigation Methodology 2
Rosen Burke Q. Formal analysis Methodology 5
Kaestner Erik Data curation Methodology 6
Yang Jimmy C. Data curation Investigation Methodology 7 8
Soper Daniel J. Investigation Methodology 7
Han Seunggu Jude Investigation Methodology 2
Paulk Angelique C. Conceptualization Data curation Investigation Methodology 7 9
Cash Sydney S. Conceptualization Methodology Project administration Resources Supervision 7 9
Raslan Ahmed M. Conceptualization Data curation Investigation Methodology Supervision 2
Dayeh Shadi A. Conceptualization Data curation Funding acquisition Project administration Supervision 3 10 11
Halgren Eric Conceptualization Data curation Formal analysis Funding acquisition Investigation Methodology Project administration Supervision Writing – review & editing 12 13
1 Department of Neurosurgery, University of California San Diego, La Jolla, California, United States of America
2 Department of Neurological Surgery, Oregon Health & Science University, Portland, Oregon, United States of America
3 Department of Electrical and Computer Engineering, University of California San Diego, La Jolla, California, United States of America
4 Neurosciences Graduate Program, University of California, San Diego, La Jolla, California, United States of America
5 Department of Neuroscience, Washington University School of Medicine, St. Louis, Missouri, United States of America
6 Center for Multimodal Imaging and Genetics, University of California San Diego, La Jolla, California, United States of America
7 Department of Neurology, Massachusetts General Hospital, Boston, Massachusetts, United States of America
8 Department of Neurosurgery, Massachusetts General Hospital, Boston, Massachusetts, United States of America
9 Center for Neurotechnology and Neurorecovery, Department of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, United States of America
10 Materials Science and Engineering Program, University of California San Diego, La Jolla, California, United States of America
11 Department of Bioengineering, University of California San Diego, La Jolla, California, United States of America
12 Department of Radiology, University of California San Diego, La Jolla, California, United States of America
13 Department of Neuroscience, University of California San Diego, La Jolla, California, United States of America
Poeppel David Academic Editor
New York University, UNITED STATES OF AMERICA
SD and AR have equity in a startup company (Intelecterra). SC has a financial interest in an unrelated company (Beacon Biosignals). No other authors report competing interests.

* E-mail: clearyd@ohsu.edu
6 9 2024
9 2024
6 9 2024
22 9 e30027746 2 2024
29 7 2024
https://creativecommons.org/publicdomain/zero/1.0/ This is an open access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

Modular organization at approximately 1 mm scale could be fundamental to cortical processing, but its presence in human association cortex is unknown. Using custom-built, high-density electrode arrays placed on the cortical surface of 7 patients undergoing awake craniotomy for tumor excision, we investigated receptive speech processing in the left (dominant) human posterior superior temporal gyrus. Responses to consonant-vowel syllables and noise-vocoded controls recorded with 1,024 channel micro-grids at 200 μm pitch demonstrated roughly circular domains approximately 1.7 mm in diameter, with sharp boundaries observed in 128 channel linear arrays at 50 μm pitch, possibly consistent with a columnar organization. Peak latencies to syllables in different modules were bimodally distributed centered at 252 and 386 ms. Adjacent modules were sharply delineated from each other by their distinct time courses and stimulus selectivity. We suggest that receptive language cortex may be organized in discrete processing modules.

The brain is anatomically organized in columns, but how these columns relate to function in the human association cortex is not clear. This study reveals that the human superior temporal gyrus is functionally organized in ~1.7-mm diameter modules that process language information.

NIH Brain Initiative MH120886-01 https://orcid.org/0000-0001-7112-4519
Cleary Daniel R. NIH DP2-EB029757 Dayeh Shadi A. NIH Brain Initiative R01-NS123655 Dayeh Shadi A. NIH UG3-NS123723 Dayeh Shadi A. NIH Brain Initiative R01-DA050159 Dayeh Shadi A. http://dx.doi.org/10.13039/100000001 National Science Foundation 1728497 Dayeh Shadi A. http://dx.doi.org/10.13039/100000025 National Institute of Mental Health R01-MH117155 Halgren Eric NIH K24-NS088568 Cash Sydney S. MGH Executive Committee on Research Cash Sydney S. http://dx.doi.org/10.13039/100018260 Tiny Blue Dot Foundation Cash Sydney S. This work was supported by National Institutes of Health (NIH, https://www.nih.gov), National Science Foundation (NSF, https://www.nsf.gov), and NIH High-Risk, High-Reward Research program (https://commonfund.nih.gov/highrisk), in the form of the following awards: NIH Brain Initiative, #MH120886-01, DRC; NIBIB award #DP2-EB029757, SAD; NIH BRAIN Initiative R01NS123655, SAD; NIH BRAIN Initiative UG3NS123723, SAD; NIH BRAIN Initiative R01DA050159, SAD; NSF award #1728497, SAD; NIMH R01 MH117155, EH; NIH award #K24-NS088568, SSC; MGH Executive Committee on Research, SSC; Tiny Blue Dot Foundation, SSC, ACP. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. PLOS Publication Stagevor-update-to-uncorrected-proof
Publication Update2024-09-18
Data AvailabilityData are available: https://zenodo.org/records/12792343.
Data Availability

Data are available: https://zenodo.org/records/12792343.
==== Body
pmcCortical function is often understood as composed of discrete processing modules whose outputs converge on other modules. Commonly, cortical columns are posited to perform this role, bridging local circuits and cortical parcels. In primary sensory cortices of certain species, spatially demarcated groups of neurons form discrete columns that respond to specific stimuli, distinct from neighboring columns [1–3]. Using fMRI, orientation-selective and ocular dominance columns have been identified in V1 [4], color and binocular disparity stripes in V2 and V3 [5], axis of motion columns in MT [6], and frequency columns in A1 [7]. Column diameter varies. They are approximately 500 μm in cat somatosensory [8] or macaque ocular dominance, but human ocular dominance are approximately 1.7× wider [9]. However, due to limited spatial resolution, these fMRI studies have difficulty demonstrating a mosaic organization with functional properties relatively constant within the column and changing sharply at their boundaries. Furthermore, with the partial exceptions of MT and V2/3, these findings have been restricted to primary sensory cortex. Although columnar organization is clear anatomically [10,11], it has not been associated with functional columns [12]. Thus, the status of columns, or other modularity at that scale, in human association cortex, remains unknown.

Direct intracranial recordings from the dominant posterior superior temporal gyrus (STG) have characterized the response correlates of neuronal populations to speech [13]. Generally, population firing is estimated from high gamma amplitude (HG; approximately 70 to 190 Hz), recorded from approximately 3 mm diameter contacts at 10 mm pitch. Using 4 mm pitch, Mesgarani and colleagues [14] found HG selectivity for groups of phonemes that shared articulatory features, and Leonard and colleagues [15] found that STG neurons in a track perpendicular to the surface often encode similar features, and their firing is reflected in the overlying surface. However, lateral spatial resolution in these studies has not been sufficient to resolve cortical columns.

Here, we measured the shape, size, and spacing of cortical modules selectively engaged by syllables using microgrids on the pial surface of STG. For the first set of experiments (4 participants), we used a 128-channel grid with 2 × 64 contacts at 50 μm pitch (Fig 1A). The second set of experiments (3 participants, one with 2 placements) used a 1,024-channel 16 × 64 grid at 200 μm pitch (Fig 1B). All participants were undergoing acute awake craniotomies for mapping eloquent cortex prior to removal of a tumor [16]. Arrays were placed under visual control on a relatively avascular cortical surface which previous clinical testing identified as eloquent (Fig 1C and 1D). Patients listened to consonant-vowel syllables and detected those that formed valid words (words and pseudo-words were combined for analysis). Noise-vocoding of the same stimuli produced controls with the same amplitude envelope in different frequency bands as the spoken stimuli (Fig 1E).

10.1371/journal.pbio.3002774.g001 Fig 1 Electrodes and stimuli used in the preliminary experiments.

Layouts of (A) 128 ch array (2 × 64, 50 μm pitch, 30 μm diameter PEDOT contacts) and (B) 1,024 ch array (16 × 64, 200 μm pitch, 30 μm diameter Pt nanorod contacts). (C) Intraoperative photo of 1,024 ch array adhering to the pial surface. (D) Approximate centers of electrode arrays on standard brain. (E) Spectrograms of an example word and its noise-vocoded control.

Six of the 7 patients showed significant modulation of local field potentials to syllables versus noise, in both lower frequencies (as the average evoked potential, Fig 2A) and as the average instantaneous high gamma amplitude (HG; Fig 2C). In one example with a linear 2 × 64 arrays, both the evoked potential and HG responses were very similar for approximately 25 channels (approximately 1.3 mm) and then abruptly transitioned to channels with distinct response properties which again were similar to each other (Fig 2B and 2D). For HG, this effect was clear not only in the average (Fig 2B) but in individual trials where the response appeared as a simultaneous burst of HG in consistent clusters of adjacent channels (Fig 2E–2G). Similarly, responding channels were clustered using non-negative matrix factorization (NNMF) of the gamma-band (70 to 190 Hz) analytic amplitude, and the responses quantified for each cluster of channels (2H). Within these cortical “modules,” gamma-band correlations plateaued at a very high level, but at the boundary between modules, correlations dropped rapidly (Fig 2I). The consistency of responses within modules and the steep transition between modules was also apparent in other measures, including the peak value of the HG to Syllables or Noise at each location (Fig 2J). These empirical curves show a transition zone between modules of 400 μm width, which corresponds to that which would be generated by a disc of dipoles 100 μm below the cortical surface (Fig 2K).

10.1371/journal.pbio.3002774.g002 Fig 2 Selective responses to syllables.

(A) Averaged evoked potential from 2 sample channels shows greater selectivity in upper channels, with (B) an abrupt change in response midway through the array. (C, D) Same as A, B but analytic amplitude high gamma (70–190 Hz) averaged across trials. Note that the upper channels have very consistent responses, as do the lower, suggesting 2 distinct functional modules. (E) As in C, but for 2 individual trials showing spatially disjoint HG burst-responses. (F) The peak responses from each trial are plotted in accordance by the spatial middle of the response. Note the highly consistent co-activation of each module across trials. (G) A histogram of the spatial distribution of gamma responses shows the bimodal distribution. (H) ANOVA finds a syllable-selective HG response in module #1 and an interaction of syllable with module (S1 Data). (I) HG-band activity in the first approximately 25 contacts (spanning approximately 1.25 mm) were highly correlated over the entire task with each other, then after a transition zone of approximately .4 mm (10%–90% levels), the next 25 contacts were again highly correlated with each other but not with the first 25. Thus, the channels were organized into 2 groups, with very high within-group, and low out-group, correlation. (J) Blue lines show the peak value of the HG average across all Syllable trials, at each contact; Red lines Noise trials. Note the roughly flat plateau in the modules and a sharp transition zone. (K) The surface potential produced by alternating 2 mm diameter discs of dipoles placed 100 μm below the surface. The modeled curves are similar to the empirical data in panel G, suggesting possible mosaic organization of functional domains in human association cortex. The gray bars are 400 μm wide and span the distance over which the curves pass from 10% to 90% of maximum, marking the transition zone. Note that the 3.2 mm of the pial surface spanned by this 2 × 64 array at 50 μ pitch would sample activity from 2 or 3 modules with approximately 1.7 mm diameter. Similar results from other patients are shown in S1–S4 Figs.

Co-active modules were also identified in the 1,024-channel arrays using NNMF of the gamma band (70 to 190 Hz) analytic amplitude recorded over the entire task period (Fig 3A and 3E). Since the algorithm has no access to spatial information, channels are spatially co-localized only because of functional similarity in signals. HG band phase-locking is very high and consistent within modules and drops rapidly at their boundaries (Fig 3C and 3D). The average diameter of modules with significant selectivity for syllables of approximately 1.7 mm was roughly consistent across the 3 patients, 4 recordings, and 27 modules (mean ± stdev height: 1.73 ± 0.64 mm; width: 1.73 ± 0.59 mm).

10.1371/journal.pbio.3002774.g003 Fig 3 Discrete functional modules evoked by CV syllables on STG.

(A) Modules of co-active electrodes at 200 μm pitch in 16 × 64 arrays identified with NNMF of HG amplitude recorded over the entire task period, in the first placement in patient 6. Module boundaries are at 2 stdev above the mean of the entire array. (B) HG amplitude recorded from column 21 of placement 6A, passing through module 7 which peaks at approximately 200 ms after stimulus onset, and module 2 which peaks at approximately 500 ms. Columns are marked with red arrows in panel A. Note that adjacent contacts belonging to different modules have completely different time courses of activation (→). (C) Average HG amplitude to syllables and noise for modules 2 and 7. (D) PLV [19] between HG at the center contact of each module and the entire array. (E–G) Same as A–C but for a traverse in column 42 of modules 1 and 5 in patient 7. Again, these modules have different latency peaks (approximately 250 and approximately 450 ms), with a sharp transition between response profiles (←). (H) Histogram of peak module response latencies shows 2 distinct peaks. Similar results from other patients are shown in S5–S7 Figs.

In some patients, all modules have very similar time courses, and thus their delineation is unlikely to reflect differences in temporal response patterns. In order to further evaluate the effect of temporal pattern, we repeated NNMF including only the time-averaged responses from 100 to 400 ms in each trial, instead of the entire recording. Despite the approximately 1,500-fold decrease in samples, similar modules were identified (S9 Fig). We conclude that while it is possible that time-course differences made some contribution to refining module boundaries, such differences are not generic auditory on-off/sustained-phasic response modulations but rather modulation within the CV syllable.

Compared to the 100 ms pre-stimulus baseline, HG from 50 to 450 ms in all 27 modules was significantly greater to the stimuli (p < 0.02, two-tailed paired t test, FDR-corrected). While 18 of 27 modules responded preferentially to syllables, 1 responded more to noise, and 8 did not significantly distinguish syllables and noise (p < 0.02, two-tailed paired t test, FDR-corrected). Examination of contacts at module boundaries revealed a sharp transition between adjacent contacts separated by 200 μm (Fig 3B and 3C). Peak response times in the modules were bimodal, with 13 peaking below 350 ms and 14 above (Fig 3F–3H). The histogram of peak latencies was found to be marginally more parsimoniously modeled by a mixture of 2 gaussians than by one (Akaike information criterion = 314.08 versus 314.93) with means of 252 and 386 ms. The first phoneme of each stimulus was a consonant and the second a vowel, and the latency between the average onsets of the consonants and vowels were similar to those of the early and late module response peaks. Additional recordings will be needed to test the hypothesis that consonants and vowels are encoded in adjacent sharply delineated modules.

Based on these findings, we can estimate that each typical 3 to 4 mm diameter clinical electrode would record from 4 to 8 modules, and at 10 mm spacing would miss approximately 85% of the modules. Conversely, approximately 250 μm diameter contacts at approximately 1 mm pitch would result in recordings from all modules, in most cases without significant crosstalk. Although these estimates are very rough, and assume that the modular organization exists in other association areas and are the same size, they suggest that finer sampling may reveal important clinical [17] and basic information, and provide the approximate spatial sampling necessary in language prostheses for detecting all functional units.

In conclusion, high-resolution recordings from STG provide evidence for modular organization of functional units having: (1) roughly circular shapes with approximately 1.7 mm diameter; (2) sharp edges with immediately adjacent modules having contrasting response time courses and stimulus correlates; (3) roughly uniform size; and (4) continuous mapping over the sampled cortical surface forming a mosaic. Although these characteristics are consistent with a columnar organization, convincing demonstration would require oblique laminar recordings identifying a sharp transition at depth. Modular organization has also been seen for functional units with fMRI, for example for faces, but those are typically irregular approximately 5 to 20 mm patches, and do not “pave” the cortex with multiple modules with sharp transitions [18]. It is possible to tile the entire cortical surface with approximately 128,000 1.7 mm diameter modules, each with approximately 96,000 neurons. Investigation of whether such modules exist in other association cortices, their dimensions, and information-processing activities of modules may inform multiresolution models of cortical function.

Supplementary information

Online methods

Acute intraoperative setting. When a tumor or lesion is near eloquent areas, neurosurgeons may elect to perform a resection procedure while the patient is awake, primarily using local anesthetic and systemic analgesics while minimizing sedation [16]. This technique allows the surgeon to closely monitor brain function during surgery, ensuring maximum safe resection while avoiding damage to important areas. This provides a unique opportunity for research on functional cortex while the patient is awake and interactive. Awake craniotomies, however, are relatively infrequent surgeries, and the potential time for research is limited to 20 min. Patients gave written informed consent to the microgrid recordings under a protocol approved by the local Institutional Review Board (OHSU IRB # 19099) and in accordance with the principles expressed in the Declaration of Helsinki.

Microarrays [20,21] have ultra-low impedance 30 μm diameter PEDOT:PSS (typical 1 kHz impedance magnitude of 38 kΩ) and Pt nanorod contacts (typical 1 kHz impedance magnitude of 25.5 kΩ) on a 7-μm thick parylene C substrate that conforms to the pial surface, further facilitating high-quality recordings. The fabrication and testing of these microarrays have previously been reported [20–22]. All devices in the experimental portion are powered via battery to avoid risk of ground loops or mains flow to the patient [23].

LFP preprocessing is performed in MATLAB 2019b and LFPs inspected visually using the FieldTrip toolbox [24], and 60 Hz and harmonics are notch filtered (zero-phase).

Time-frequency analyses. Average time-frequency plots of the event-related spectral power (ERSP) are generated from the broadband LFP using EEGLAB [25]. ERSP is calculated from 1 Hz to 500 Hz with 1 Hz resolution by computing and averaging fast Fourier transforms with Hanning window tapering. Each 1 Hz bin is normalized with respect to the mean power at −200 to −150 ms and masked with two-tailed bootstrapped significance with FDR correction and α = 0.05. Event-related phase amplitude coupling (ERPAC) is calculated using an open-source MATLAB package [26].

Non-negative matrix factorization (NNMF) of high gamma (70 to 190 Hz) analytic amplitude recorded over the entire task period at 1 ms accuracy was used to cluster the channels based on signal similarity without regard for spatial colocalization of channels [27]. The entire unaveraged signal from all technically correct channels over the entire task was entered into the factorization. The optimal number of factors for NNMF was selected by picking the approximate nadir using cross-validation; one by one, a single element is left out of the data set, and the prediction error is measured for each possible number of modules [28]. Once the optimal number of modules is defined, for each module the algorithm creates a map by assigning weights to each channel. Modules are defined by spatial colocalization of highly weighted channels; the threshold for channel inclusion is defined as weights 3 standard deviations above the mean for all channels for that module. Colocalized modules of channels are subsequently refined using phase-locking value, wherein the spatial center of each module was identified, and phase-locking calculated for each channel relative to the center channel. To display the boundaries of modules, a threshold value of PLV is picked that minimizes the overlap between modules.

The NNMF analysis was repeated including only the time-averaged responses from 100 to 400 ms in each trial, instead of the entire recording at 1 ms in order. The time-averaged NNMF used only approximately 200 samples for each of the approximately 1,000 channels (one for each trial), whereas the original method entered approximately 300 k samples per channel (approximately 5 min at 1,000 Hz). Inasmuch as factorial analyses of this kind are well known to require a large number of samples, it is likely that the differences between the panels are due to both the decreased data entering the time-averaged NNMF as well as the specific absence of temporal information.

We explicitly tested if the modules identified by NNMF could be due to cross-talk in the ribbon cable running from the electrode to the electronics, or in shared processing characteristics in the electronics (S8 Fig). First, we plotted a histogram of the physical distances between contacts belonging to the same module (S8A Fig). As expected, given that modules are compact spatially, these formed a roughly Gaussian distribution centered at approximately 1 mm and never exceeding approximately 2.3 mm. In contrast, the histogram of lines between contacts belonging to the same module had 2 main peaks, one at short distances mainly approximately 1 to 8 separations, and one at approximately 800 to 1,000 separations (S8B Fig). A close-up of the short distances (S8C Fig) shows a distribution completely different from S8A Fig. If the modules were due to crosstalk then the corresponding long and short distances should have been noted in S8A Fig, but they were not. Similarly, the assignment of channels to banks of electronics generally alternated between rows of contacts (S8D Fig), so if shared electronic processing characteristics caused the module clustering, individual clusters should then also have members on alternate rows, which they did not. Recall that the NNMF procedure is completely unaware of the spatial location of the contacts, thus if cross-talk in the ribbon cable, or shared amplifier bank characteristics underlay the modules, then they would have reproduced the proximity matrices of the cable or of the amplifier assignments, rather than being organized in spatially compact roughly circular domains. Finally, we also note that the same ribbon-cable and amplifier bank assignments were used in all experiments and yet quite different module locations were observed. Furthermore, in one patient, the same actual array was simply moved from one STG site to another between blocks of the task, resulting in different module locations being observed on the same array. This evidence is inconsistent with modules arising from technical sources of signal similarity.

Phase-locking value. PLV is an instantaneous measure of phase-locking [19], which unlike coherence is not affected by shared amplitude modulation. PLV time courses are computed using the analytic angle of the Hilbert transformed 70 to 190 Hz bandpassed (zero-phase shift) signals of each channel pair, and their significance calculated using FDR correction and a null distribution [29].

Statistical analyses. HG relative to behavioral events is compared to an appropriate baseline control period, and between task conditions. To determine if a module was responsive to the stimuli, the Hilbert analytic amplitude from 70 to 190 Hz was averaged on each trial in a between baseline period (100 ms before stimulus onset) and in the active period (50 to 450 ms poststimulus onset). A two-tailed paired t test was performed using pairs of values from each trial. To determine if a module was differentially responsive to syllables versus noise, a two-tailed t test was performed comparing the average value in the active period between syllable and noise trials. All statistical tests are evaluated with α = 0.02, FDR-corrected across all patients and modules [29]. Patient 7 showed responses to both syllables and noise but they were not significantly different, possibly due to residual anesthesia that was applied during the preceding craniotomy, which was observed behaviorally as slow responses and intermittent confusion. Patient 7’s five modules failed to distinguish between syllables and noise.

Model. The surface potential produced by alternating 2 mm diameter discs of dipoles (5,017 at 25 μ pitch, each 10e-6 μAmm) placed 100 μm below the surface in an infinite homogeneous isotropic medium (0.33 S/m). Estimates of number of modules and neurons per module based on total cortical surface area from ref [30] and total cortical neurons in gray matter from ref [31].

Task. A wide range of Consonant-Vowel syllables were presented to provide diverse activation of STG. Eight consonants were drawn equally from 4 categories following linguistic categories: Obstruent-Plosive (d, g), Obstruent-Fricative (s, z), Sonorant-Liquid (l, r), and Sonorant-Nasal (m, n). Nine vowels are drawn equally from 3 categories: Type 1 (u, aɪ, ɔ), Type 2 (ə, I, ɛ), and Type 3 (eɪ, æ, aʊ). Each possible CV combination is equally represented in the total stimulus set, resulting in 72 total syllables. Each was spoken once by a male voice and once by a female. Each presented sound lasts approximately 450 ms, and the SOA (stimulus onset asynchrony) was individually adjusted but usually approximately 1,100 ms. Stimuli were presented using Presentation (Neurobehavioral Systems) in approximately 6 min blocks of 288 trials, each stimulus occurring once. The task was to detect CV syllables which are syllables.

Noise-vocoding. For every CV syllable, a matching noise-vocoded stimulus was constructed taking the existing biphoneme stimuli and creating a 6-band stimulus in which white noise was multiplied by power in each of the bands to create a matched set of auditory stimuli with identical time-varying spectral acoustics [32,33]. Noise-vocoded stimuli preserve temporal envelope cues in each spectral band, providing a control for the sensory characteristics of speech, but the spectral detail within each band was degraded.

Supporting information

S1 Fig S1 through S4 Figs show recordings from the 2 × 64 channel linear array, at 50 μm pitch.

Data from only one of the 2 channels at each location in the long dimension are presented for clarity; the opposite channel was highly similar in all cases where both were technically sound. Solid, dark blue pixels indicate channels without usable signal. S1 Fig shows spectral and spatiotemporal responses for the linear array in patient #4. (A) The mean event-related spectrogram of all trials shows a change in both the high gamma-band and low frequency after the onset of stimuli. (B) Event-related phase-amplitude coupling (ERPAC) [26] between 2–8 Hz and 70–100 Hz shows a stimulus-related link in response for all channels, as measured with the z-score for the ERPAC (top). FDR-corrected p-values (bottom) shows that the time points of statistical significance for the ERPAC are in the first few hundred milliseconds after the onset of the auditory stimuli. (C) A matrix of phase-locking values (PLV) between all channels shows 2 distinct regions of high phase-locking within but not between regions. (D) On the spatiotemporal response map, the entire electrode consistently responds to auditory stimuli, but 2 distinct clusters of channels are identified that factorize together (green bar, yellow bar). Close inspection shows a late temporal response in cluster #1 that separates it from cluster #2 (arrowhead). The same channels that cluster together in (D) show a high-degree of phase-locking in the gamma band (C). (E) The mean gamma-band response in the 50–450 ms after onset of stimuli for the clustered channels from (D) shows a stronger response to words than noise stimuli, which is only significant with cluster #2 (S2 Data).

(EPS)

S2 Fig Spectral and spatiotemporal responses for linear array in patient #8, recording from non-dominant pSTG.

(A) The mean event-related spectrogram of all trials shows an event-related change in both the high gamma-band and low frequency after the onset of stimuli. (B) On the spatiotemporal response map of gamma activity, increased activity was observed in the few hundred milliseconds after onset of auditory stimuli at time zero. Two distinct spatial segments are identified using factorization to cluster the channels (green bar, yellow bar). (C) Quantifying the mean gamma-band response in the 50–450 ms after onset of stimuli for the clustered channels shows a stronger response to noise than word stimuli for both clusters of channels (S3 Data).

(EPS)

S3 Fig Spectral and spatiotemporal responses for linear array in patient #3.

(A) The mean event-related spectrogram of all trials shows a lower amplitude of increase in both the high gamma-band and low frequency than observed in other patients. (B) Similarly, on the spatiotemporal response map of gamma activity, increased activity was only observed in one segment for in the few hundred milliseconds after onset of auditory stimuli. Only one spatial segment was identified using factorization (green bar). (C) Quantifying the mean gamma-band response in the 50–450 ms after onset of stimuli for the clustered channels shows a stronger response to words than noise stimuli (S4 Data).

(EPS)

S4 Fig Spectral and spatiotemporal responses for linear array in patient #2.

(A) The mean event-related spectrogram of all trials shows a change in both the high gamma-band and in low frequencies after the onset of stimuli. (B,C) On the spatiotemporal response map of gamma activity for words, spatially segregated increase activity was observed in the few hundred milliseconds after onset of auditory stimuli at time zero. Three distinct spatial segments are identified using factorization to cluster the channels, two of which appear to preferentially respond to word stimuli (green bar, brown bar) while one segment appears to preferentially respond to noise stimuli (yellow bar). (D) Quantifying the mean gamma-band response in the 50–450 ms after onset of stimuli for the clustered channels shows a stronger response to words than noise stimuli for one of the clusters of channels (green cluster), a stronger response to noise for one cluster (yellow cluster), and an indeterminate response for one cluster (brown cluster) (S5 Data).

(EPS)

S5 Fig Spatiotemporal maps of gamma-band responses for Patient #6A, using the 1,024-channel electrode array (16 × 64 with 200 μm spacing).

(A) The mean gamma-band response between 100 ms and 500 ms after the onset shows multiple poorly demarcated regions increased signal. (B) Finer temporal discrimination is used to view gamma-band responses every 100 ms, which shows distinct spatial regions of activation. The spatial regions have distinct temporal onset and offset. Arrowheads indicate spatial regions with unique temporal activation times. (C) A spatial map of the 16 × 64 channel electrode with the results of factorization overlaid. Factorization identified 9 distinct, spatially adjacent regions. (D) For an individual cluster of channels (Module #1), the phase-locking value for each cluster shows a high degree of concordance for spatially adjacent channels with rapid drop off. (E) The mean signal of the principal components of Module #1 shows the temporal and feature preference for the underlying channels. The mean response is calculated using the average response relative to the start of the stimuli for all trials. (F) The other modules identified by factorization similarly show distinct temporal properties in terms of average onset, offset, and response features.

(EPS)

S6 Fig Spatiotemporal maps of gamma-band responses for Patient #6B, using the 1,024-channel electrode array (16 × 64 with 200 μm spacing).

(A) A spatial map of the 16 × 64 channel electrode with the results of factorization overlaid. Factorization identified 9 distinct, spatially adjacent regions. (B) For an individual cluster of channels (Module #1), the phase-locking value relative for cluster shows a high degree of concordance for spatially adjacent channels with rapid drop off. (C) The mean signal of the principal components of Module #1 shows the temporal and feature preference for the underlying channels. The mean response is calculated using the average response relative to the start of the stimuli for all trials. (D) The other modules identified by factorization similarly show distinct temporal properties in terms of average onset, offset, and response features.

(EPS)

S7 Fig Spatiotemporal maps of gamma-band responses for Patient #7, using the 1,024-channel electrode array (16 × 64 with 200 μm spacing).

(A) A spatial map of the 16 × 64 channel electrode with the results of factorization overlaid. In this patient, factorization identified 5 distinct, spatially adjacent regions. (B) For an individual cluster of channels (Module #1), the phase-locking value relative for cluster shows a high degree of concordance for spatially adjacent channels with rapid drop off. (C) The mean signal of the principal components of Module #1 shows the temporal and feature preference for the underlying channels. The mean response is calculated using the average response relative to the start of the stimuli for all trials. (D) The other modules identified by factorization similarly show distinct temporal properties in terms of average onset, offset, and response features.

(EPS)

S8 Fig Comparison of spatial organization of modules to that expected if they were to arise from cross-talk in the ribbon-cable or shared processing electronics.

We explicitly tested if the modules identified by NNMF could be due to cross-talk in the ribbon cable running from the electrode to the electronics, or in shared processing characteristics in the electronics. (A) Histogram of the physical distances between contacts belonging to the same module. Note that, as expected given that modules are compact spatially, these formed a roughly Gaussian distribution centered at approximately 1 mm and never exceeding approximately 2.3 mm. (B) Histogram of the number of conducting wires on the ribbon cable between contacts belonging to the same module. Note that this histogram has 2 main peaks, one at short distances with mainly approximately 1–8 wires between adjacent channels and another peak approximately 800–1,000 wires between adjacent channels. (C) Close-up of panel B shows a distribution clearly different than that observed for intra-module distances between contacts in Panel A. (D) Assignment of channels to banks of electronics (Analog-to-Digital Converter [ADC] and amplifiers). Note the striped/mosaic organization of the ADC, again clearly different from that of identified modules.

(EPS)

S9 Fig Non-negative matrix factorization weights derived from using only the mean of the poststimulus response, from patients 6A and 6B (also shown in Fig 3 and S5 Fig, respectively).

For each auditory stimulus, the analytic amplitude of the gamma-band response for each channel was averaged from 100 ms to 400 ms poststimulus, and this limited data set was fed into the NNMF algorithm. The algorithm receives no information on spatial locations of channels, but still reconstructs a similar spatial pattern of clustered channels as when the complete recording is used. (A) Example result showing a single cluster of spatially adjacent channels with high factorization weights from patient 6A when only using the average of the poststimulus gamma-band response. The lower number is the corresponding cluster number from Fig 3. (B) The other 8 clusters identified by NNMF based on the mean poststimulus response only, with corresponding cluster number from Fig 3 at bottom. (C) The clusters from A and B overlaid onto a single graph without overlap allowed between clusters. (D) For comparison, the clusters identified by using the entire recording at 1 ms resolution (previously shown in Fig 3). (E) As above, an example cluster of channels based on factorization weights using only the average of the poststimulus gamma-band response, this time from patient 6B. (F) The 8 other clusters identified by NNMF for the average response from patient 6B. (G) The clusters from E and F overlaid onto a single graph without overlap allowed between clusters. (H) For comparison, the clusters identified by using the entire recording at 1 ms resolution (previously shown in S9 Fig).

(EPS)

S1 Data The raw values for bar graphs shown in Fig 2H.

(XLSX)

S2 Data The raw values for bar graphs shown in S1E Fig.

(XLSX)

S3 Data The raw values for bar graphs shown in S2C Fig.

(XLSX)

S4 Data The raw values for bar graphs shown in S3C Fig.

(XLSX)

S5 Data The raw values for bar graphs shown in S4D Fig.

(XLSX)

Abbreviations

ERPAC event-related phase amplitude coupling

ERSP event-related spectral power

HG high gamma

NNMF non-negative matrix factorization

PLV phase-locking value

SOA stimulus onset asynchrony

STG superior temporal gyrus

10.1371/journal.pbio.3002774.r001
Decision Letter 0
Schnell, PhD Christian Senior Editor
© 2024 Christian Schnell, PhD
2024
Christian Schnell, PhD
https://creativecommons.org/licenses/by/4.0/ This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Submission Version0
8 Feb 2024

Dear Dr Cleary,

Thank you for submitting your manuscript entitled "Modular Phoneme Processing in Human Superior Temporal Gyrus" for consideration as a Short Reports by PLOS Biology.

Your manuscript has now been evaluated by the PLOS Biology editorial staff as well as by an academic editor with relevant expertise and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. After your manuscript has passed the checks it will be sent out for review. To provide the metadata for your submission, please Login to Editorial Manager (https://www.editorialmanager.com/pbiology) within two working days, i.e. by Feb 10 2024 11:59PM.

If your manuscript has been previously peer-reviewed at another journal, PLOS Biology is willing to work with those reviews in order to avoid re-starting the process. Submission of the previous reviews is entirely optional and our ability to use them effectively will depend on the willingness of the previous journal to confirm the content of the reports and share the reviewer identities. Please note that we reserve the right to invite additional reviewers if we consider that additional/independent reviewers are needed, although we aim to avoid this as far as possible. In our experience, working with previous reviews does save time.

If you would like us to consider previous reviewer reports, please edit your cover letter to let us know and include the name of the journal where the work was previously considered and the manuscript ID it was given. In addition, please upload a response to the reviews as a 'Prior Peer Review' file type, which should include the reports in full and a point-by-point reply detailing how you have or plan to address the reviewers' concerns.

During the process of completing your manuscript submission, you will be invited to opt-in to posting your pre-review manuscript as a bioRxiv preprint. Visit http://journals.plos.org/plosbiology/s/preprints for full details. If you consent to posting your current manuscript as a preprint, please upload a single Preprint PDF.

Feel free to email us at plosbiology@plos.org if you have any queries relating to your submission.

Kind regards,

Christian

Christian Schnell, PhD

Senior Editor

PLOS Biology

cschnell@plos.org

10.1371/journal.pbio.3002774.r002
Decision Letter 1
Schnell, PhD Christian Senior Editor
© 2024 Christian Schnell, PhD
2024
Christian Schnell, PhD
https://creativecommons.org/licenses/by/4.0/ This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Submission Version1
4 Apr 2024

Dear Dr Cleary,

Thank you for your patience while your manuscript "Modular Phoneme Processing in Human Superior Temporal Gyrus" was peer-reviewed at PLOS Biology. It has now been evaluated by the PLOS Biology editors, an Academic Editor with relevant expertise, and by several independent reviewers.

In light of the reviews, which you will find at the end of this email, we would like to invite you to revise the work to thoroughly address the reviewers' reports.

As you will see below, the reviewers think that the study is overall well executed and provides important insights. However, both reviewers raise concerns about missing details and the phoneme-selectivity aspect that need to be addressed.

Given the extent of revision needed, we cannot make a decision about publication until we have seen the revised manuscript and your response to the reviewers' comments. Your revised manuscript is likely to be sent for further evaluation by all or a subset of the reviewers.

We expect to receive your revised manuscript within 3 months. Please email us (plosbiology@plos.org) if you have any questions or concerns, or would like to request an extension.

At this stage, your manuscript remains formally under active consideration at our journal; please notify us by email if you do not intend to submit a revision so that we may withdraw it.

**IMPORTANT - SUBMITTING YOUR REVISION**

Your revisions should address the specific points made by each reviewer. Please submit the following files along with your revised manuscript:

1. A 'Response to Reviewers' file - this should detail your responses to the editorial requests, present a point-by-point response to all of the reviewers' comments, and indicate the changes made to the manuscript.

*NOTE: In your point-by-point response to the reviewers, please provide the full context of each review. Do not selectively quote paragraphs or sentences to reply to. The entire set of reviewer comments should be present in full and each specific point should be responded to individually, point by point.

You should also cite any additional relevant literature that has been published since the original submission and mention any additional citations in your response.

2. In addition to a clean copy of the manuscript, please also upload a 'track-changes' version of your manuscript that specifies the edits made. This should be uploaded as a "Revised Article with Changes Highlighted" file type.

*Re-submission Checklist*

When you are ready to resubmit your revised manuscript, please refer to this re-submission checklist: https://plos.io/Biology_Checklist

To submit a revised version of your manuscript, please go to https://www.editorialmanager.com/pbiology/ and log in as an Author. Click the link labelled 'Submissions Needing Revision' where you will find your submission record.

Please make sure to read the following important policies and guidelines while preparing your revision:

*Published Peer Review*

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out. Please see here for more details:

https://blogs.plos.org/plos/2019/05/plos-journals-now-open-for-published-peer-review/

*PLOS Data Policy*

Please note that as a condition of publication PLOS' data policy (http://journals.plos.org/plosbiology/s/data-availability) requires that you make available all data used to draw the conclusions arrived at in your manuscript. If you have not already done so, you must include any data used in your manuscript either in appropriate repositories, within the body of the manuscript, or as supporting information (N.B. this includes any numerical values that were used to generate graphs, histograms etc.). For an example see here: http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.1001908#s5

*Blot and Gel Data Policy*

We require the original, uncropped and minimally adjusted images supporting all blot and gel results reported in an article's figures or Supporting Information files. We will require these files before a manuscript can be accepted so please prepare them now, if you have not already uploaded them. Please carefully read our guidelines for how to prepare and upload this data: https://journals.plos.org/plosbiology/s/figures#loc-blot-and-gel-reporting-requirements

*Protocols deposition*

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for your submission to our journal. We hope that our editorial process has been constructive thus far, and we welcome your feedback at any time. Please don't hesitate to contact us if you have any questions or comments.

Sincerely,

Christian

Christian Schnell, PhD

Senior Editor

PLOS Biology

cschnell@plos.org

------------------------------------

REVIEWS:

Reviewer #1: SUMMARY

The authors used microarrays with very fine-grained spatial resolution. They recorded neural responses in superior temporal gyrus from 7 patients while they listened to consonant-vowel syllables in clear and noisy (via noise vocoding) conditions. They analyzed the similarity of responses from electrodes at variable distances in order to test for and describe the modular structure that emerges. They found that responses within ~1.7mm "modules" were very similar, which transition to a different response in the neighboring modules. This was replicated when using a number of different analysis methods, including supervised and non-supervised approaches. The authors conclude that the similarity structure they observe provides evidence for discrete processing modules along STG, with a consistent size, shape and density across patient recordings.

Overall, this is important research that I think will help propel human neuroscience into a new era of spatially resolved recordings and important discoveries.

I noted some weaknesses that need to be addressed before I can fully endorse this article for publication.

MAJOR


[1] A big claim of the current paper is in identifying the size and shape of "processing modules". I would like to see some sanity checks that the design of the micro-array itself would not be biased to finding auto-correlation structure of that reported in the study. This could be achieved by analyzing the auto-correlation structure in the noise of the micro-arrays, and conducting dummy-recordings on agar or similar.

[2] Another claim here is that the authors find "phoneme-selectivity". However, the authors conduct no analyses that support this claim directly. Phase-locking is indirectly related to the fact that the consonants occurred before the vowels across all stimuli. But a delayed response to a consonant and an immediate response to a vowel could result in the same phase-locking value. If the authors would like to make a claim about phonetic encoding, I suggest that they provide an analysis of the responses based on the factorial design of their stimuli {plosive, fricative, liquid, nasal} consonants and {low, medium, high} vowels.

[3] The patients hear CV syllables, which will sometimes form valid English words (e.g., "seɪ") and sometimes will not (e.g., "zeɪ"). The task of the participants is to press a button when they recognize a word of English, but the authors analyze responses to all CV syllables, not just those that are valid words. I ask the authors to clarify in the manuscript that analyses were performed on clear syllables and noise-vocoded syllables, where some items were words and some were not. Furthermore, to change the corresponding labels on the figures, and change them from "words" to "syllables" — or, similar.

[4] There were a few sentences in the abstract that I found unclear. I request that the following are re-phrased to be clearer what is meant:

"modular organization fundamental to cortical processing" — at what spatial scale are you referring to here?

"its presence in human association cortex is unknown" — if modular organization is critical, then how can its presence be unknown? Again, what spatial scale is being referred to?

"characterized phoneme processing" — relating to point (2) above, I don't agree that this is an accurate summary of what was characterized in this study

"phoneme selectivity" — see point (2) above

MINOR

In Figure 1C, it is not clear to me what is the array, and what is the shadow of another object.

In Figure 1D, why are single electrode positions visualized, when the authors use an entire array?

In Figure 3E and elsewhere, why are there "empty" regions where no modules emerge?

Reviewer #2 (Yves Boubenec): In this work, the authors tested the putative existence of functional modules within human auditory cortex. They show that functional responses are shared within regions of a couple of mm in diameter, suggesting discrete processing modules. That is a fascinating results that echo with long-standing literature of columnar processing in visual cortex.

The study is concise and compact, which is pleasant, but makes it missing a few points for clarity. Little, if none, is said about phonetic/syllabic tuning for these cortical domains.

________________

major comments:

From figure 3, it looks like the factorization algorithm grouped channels mainly by the temporal profile of the response time-courses. This is indeed an important feature of functional responses. Nevertheless, these columns show strong speech responses, and one should not ignore the tuning of these channels.

1) Does the clustering depend on using the time-course? Or are similar results found when using the average response to each sound (ignoring the time-course)?

2) As a matter of fact, I could not find much detail in the methods about the actual data provided to the factorization algorithm. Does it simply perform on the average response time-course? On the average time-course for noise and syllable stimuli? On the average time-course for each syllable? Or on the single-trial responses?

3) A clear link with previous literature is missing. What is the phonetic tuning of these modules? Is it consistent with results of Mesgarani's and Chang's teams (cited in the introduction)? Do they reveal another level of organization that previous studies could have missed (like each module being heavily dedicated to one specific group of phonemes)?

An alternative option is that, these modules could be poorly tuned to phonemes themselves, but instead exhibit syllabic tuning. And the data of this study can address this issue.

________________

minor comments:

i) I was unclear about the exact stimulus set: do you confirm there are 24 distinct syllables in total?

ii) figures 2 IJK are not referenced. Reference to figure 2H refers in fact to figure 2K.

iii) my understanding is that no actual word was played. Instead, stimuli were single syllables. I suggest replacing 'words' in the figures by 'syllables'.

iv) acronym PLV (for phase locking) in the caption comes out of the blue.

10.1371/journal.pbio.3002774.r003
Author response to Decision Letter 1
Submission Version2
19 Apr 2024

Attachment Submitted filename: Dan PLOS Modular Response to Reviews 4.docx

10.1371/journal.pbio.3002774.r004
Decision Letter 2
Schnell, PhD Christian Senior Editor
© 2024 Christian Schnell, PhD
2024
Christian Schnell, PhD
https://creativecommons.org/licenses/by/4.0/ This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Submission Version2
26 May 2024

Dear Dr Cleary,

Thank you for your patience while we considered your revised manuscript "Modular Phoneme Processing in Human Superior Temporal Gyrus" for consideration as a Short Reports at PLOS Biology. Your revised study has now been evaluated by the PLOS Biology editors, the Academic Editor and the original reviewers.

In light of the reviews, which you will find at the end of this email, we are pleased to offer you the opportunity to address the remaining points from the reviewers in a revision that we anticipate should not take you very long. Please either conduct the suggested analysis or provide compelling argumentation why the existing analyses obviate the need for that.

We will then assess your revised manuscript and your response to the reviewers' comments with our Academic Editor aiming to avoid further rounds of peer-review, although might need to consult with the reviewers, depending on the nature of the revisions.

We expect to receive your revised manuscript within 1 month. Please email us (plosbiology@plos.org) if you have any questions or concerns, or would like to request an extension.

At this stage, your manuscript remains formally under active consideration at our journal; please notify us by email if you do not intend to submit a revision so that we withdraw the manuscript.

**IMPORTANT - SUBMITTING YOUR REVISION**

Your revisions should address the specific points made by each reviewer. Please submit the following files along with your revised manuscript:

1. A 'Response to Reviewers' file - this should detail your responses to the editorial requests, present a point-by-point response to all of the reviewers' comments, and indicate the changes made to the manuscript.

*NOTE: In your point-by-point response to the reviewers, please provide the full context of each review. Do not selectively quote paragraphs or sentences to reply to. The entire set of reviewer comments should be present in full and each specific point should be responded to individually.

You should also cite any additional relevant literature that has been published since the original submission and mention any additional citations in your response.

2. In addition to a clean copy of the manuscript, please also upload a 'track-changes' version of your manuscript that specifies the edits made. This should be uploaded as a "Revised Article with Changes Highlighted " file type.

*Resubmission Checklist*

When you are ready to resubmit your revised manuscript, please refer to this resubmission checklist: https://plos.io/Biology_Checklist

To submit a revised version of your manuscript, please go to https://www.editorialmanager.com/pbiology/ and log in as an Author. Click the link labelled 'Submissions Needing Revision' where you will find your submission record.

Please make sure to read the following important policies and guidelines while preparing your revision:

*Published Peer Review*

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out. Please see here for more details:

https://blogs.plos.org/plos/2019/05/plos-journals-now-open-for-published-peer-review/

*PLOS Data Policy*

Please note that as a condition of publication PLOS' data policy (http://journals.plos.org/plosbiology/s/data-availability) requires that you make available all data used to draw the conclusions arrived at in your manuscript. If you have not already done so, you must include any data used in your manuscript either in appropriate repositories, within the body of the manuscript, or as supporting information (N.B. this includes any numerical values that were used to generate graphs, histograms etc.). For an example see here: http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.1001908#s5

*Blot and Gel Data Policy*

We require the original, uncropped and minimally adjusted images supporting all blot and gel results reported in an article's figures or Supporting Information files. We will require these files before a manuscript can be accepted so please prepare them now, if you have not already uploaded them. Please carefully read our guidelines for how to prepare and upload this data: https://journals.plos.org/plosbiology/s/figures#loc-blot-and-gel-reporting-requirements

*Protocols deposition*

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for your submission to our journal. We hope that our editorial process has been constructive thus far, and we welcome your feedback at any time. Please don't hesitate to contact us if you have any questions or comments.

Sincerely,

Christian

Christian Schnell, PhD

Senior Editor

PLOS Biology

cschnell@plos.org

----------------------------------------------------------------

REVIEWS:

Reviewer #1: The authors have addressed my concerns - thank you.

Reviewer #2 (Yves Boubenec): I thank the authors you for their response and the revisions made to the manuscript. However, I remain concerned that my main critique has not been fully addressed. Specifically, there is still no analysis demonstrating that the NNMF leverages differences in responses across phonemes rather than primarily grouping channels based on generic temporal response patterns (e.g., onset or sustained responses).

To adequately address this point, I propose redoing the analysis using the time-averaged response (ignoring the time-course) for each sound. This would help determine whether the clustering observed is truly due to differences in phoneme-specific responses or if it is primarily driven by the temporal characteristics of the responses. Demonstrating that similar modules are identified using time-averaged responses would strengthen the argument that the NNMF is capturing phoneme-specific tuning rather than generic temporal response patterns.

Without this analysis, it remains unclear whether the observed clustering reflects phoneme-specific processing or if it is an artifact of the temporal dynamics of the neural responses. Therefore, I strongly recommend conducting this additional analysis to address this critical aspect of the study.

10.1371/journal.pbio.3002774.r005
Author response to Decision Letter 2
Submission Version3
25 Jun 2024

Attachment Submitted filename: Cleary Modular Response to Second Review v5.docx

10.1371/journal.pbio.3002774.r006
Decision Letter 3
Schnell, PhD Christian Senior Editor
© 2024 Christian Schnell, PhD
2024
Christian Schnell, PhD
https://creativecommons.org/licenses/by/4.0/ This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Submission Version3
12 Jul 2024

Dear Dr Cleary,

Thank you for your patience while we considered your revised manuscript "Modular Phoneme Processing in Human Superior Temporal Gyrus" for publication as a Short Reports at PLOS Biology. This revised version of your manuscript has been evaluated by the PLOS Biology editors, the Academic Editor and one of the original reviewers.

Based on the reviews , we are likely to accept this manuscript for publication, provided you satisfactorily address the following data and other policy-related requests.

* We would like to suggest a different title to improve the accessibility for our broad audience: "Syllable processing is organized in discrete subregions of the human superior temporal gyrus"

* Please add the links to the funding agencies in the Financial Disclosure statement in the manuscript details

* "SD and AR have equity in a startup company (Intelecterra)." is listed twice in the Competing Interests section. Please also state whether the other co-authors have any conflicts of interest.

* In the Methods section, please provide information on whether consent was given in written or oral form and whether the study was conducted according to the principles expressed in the Declaration of Helsinki.

* The abstract is very short and not very informative. Please provide a more comprehensive abstract that gives more context and specific information of what was done and its implications, as it is the only thing most readers will access.

* DATA POLICY:

You may be aware of the PLOS Data Policy, which requires that all data be made available without restriction: http://journals.plos.org/plosbiology/s/data-availability. For more information, please also see this editorial: http://dx.doi.org/10.1371/journal.pbio.1001797

Note that we do not require all raw data. Rather, we ask that all individual quantitative observations that underlie the data summarized in the figures and results of your paper be made available in one of the following forms:

1) Supplementary files (e.g., excel). Please ensure that all data files are uploaded as 'Supporting Information' and are invariably referred to (in the manuscript, figure legends, and the Description field when uploading your files) using the following format verbatim: S1 Data, S2 Data, etc. Multiple panels of a single or even several figures can be included as multiple sheets in one excel file that is saved using exactly the following convention: S1_Data.xlsx (using an underscore).

2) Deposition in a publicly available repository. Please also provide the accession code or a reviewer link so that we may view your data before publication.

Regardless of the method selected, please ensure that you provide the individual numerical values that underlie the summary data displayed in the following figure panels as they are essential for readers to assess your analysis and to reproduce it: 2H, S1E, S2C, S3C and S4D.

NOTE: the numerical data provided should include all replicates AND the way in which the plotted mean and errors were derived (it should not present only the mean/average values).

Please also ensure that figure legends in your manuscript include information on where the underlying data can be found, and ensure your supplemental data file/s has a legend.

Please ensure that your Data Statement in the submission system accurately describes where your data can be found.

* CODE POLICY

Per journal policy, if you have generated any custom code during the course of this investigation, please make it available without restrictions. Please ensure that the code is sufficiently well documented and reusable, and that your Data Statement in the Editorial Manager submission system accurately describes where your code can be found.

Please note that we cannot accept sole deposition of code in GitHub, as this could be changed after publication. However, you can archive this version of your publicly available GitHub code to Zenodo. Once you do this, it will generate a DOI number, which you will need to provide in the Data Accessibility Statement (you are welcome to also provide the GitHub access information). See the process for doing this here: https://docs.github.com/en/repositories/archiving-a-github-repository/referencing-and-citing-content

As you address these items, please take this last chance to review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the cover letter that accompanies your revised manuscript.

We expect to receive your revised manuscript within two weeks.

To submit your revision, please go to https://www.editorialmanager.com/pbiology/ and log in as an Author. Click the link labelled 'Submissions Needing Revision' to find your submission record. Your revised submission must include the following:

- a cover letter that should detail your responses to any editorial requests, if applicable, and whether changes have been made to the reference list

- a Response to Reviewers file that provides a detailed response to the reviewers' comments (if applicable, if not applicable please do not delete your existing 'Response to Reviewers' file.)

- a track-changes file indicating any changes that you have made to the manuscript.

NOTE: If Supporting Information files are included with your article, note that these are not copyedited and will be published as they are submitted. Please ensure that these files are legible and of high quality (at least 300 dpi) in an easily accessible file format. For this reason, please be aware that any references listed in an SI file will not be indexed. For more information, see our Supporting Information guidelines:

https://journals.plos.org/plosbiology/s/supporting-information

*Published Peer Review History*

Please note that you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out. Please see here for more details:

https://blogs.plos.org/plos/2019/05/plos-journals-now-open-for-published-peer-review/

*Press*

Should you, your institution's press office or the journal office choose to press release your paper, please ensure you have opted out of Early Article Posting on the submission form. We ask that you notify us as soon as possible if you or your institution is planning to press release the article.

*Protocols deposition*

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please do not hesitate to contact me should you have any questions.

Sincerely,

Christian

Christian Schnell, PhD

Senior Editor

cschnell@plos.org

PLOS Biology

------------------------------------------------------------------------

Reviewer remarks:

Reviewer #2 (Yves Boubenec): Thanks for having taken the time to address my comments in depth.

I am satisfied both by the additional analysis on the two types of electrodes, as well as you further explanations.

10.1371/journal.pbio.3002774.r007
Author response to Decision Letter 3
Submission Version4
28 Jul 2024

Attachment Submitted filename: Cleary Modular Response to Second Review v5.docx

10.1371/journal.pbio.3002774.r008
Decision Letter 4
Schnell, PhD Christian Senior Editor
© 2024 Christian Schnell, PhD
2024
Christian Schnell, PhD
https://creativecommons.org/licenses/by/4.0/ This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Submission Version4
29 Jul 2024

Dear Dr Cleary,

Thank you for the submission of your revised Short Reports "Syllable processing is organized in discrete subregions of the human superior temporal gyrus" for publication in PLOS Biology. On behalf of my colleagues and the Academic Editor, David Poeppel, I am pleased to say that we can in principle accept your manuscript for publication, provided you address any remaining formatting and reporting issues. These will be detailed in an email you should receive within 2-3 business days from our colleagues in the journal operations team; no action is required from you until then. Please note that we will not be able to formally accept your manuscript and schedule it for publication until you have completed any requested changes.

Please take a minute to log into Editorial Manager at http://www.editorialmanager.com/pbiology/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process.

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximise its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with biologypress@plos.org. If you have previously opted in to the early version process, we ask that you notify us immediately of any press plans so that we may opt out on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

Thank you again for choosing PLOS Biology for publication and supporting Open Access publishing. We look forward to publishing your study. 

Sincerely, 

Christian

Christian Schnell, PhD

Senior Editor

PLOS Biology

cschnell@plos.org
==== Refs
References

1 Mountcastle VB . Modality and topographic properties of single neurons of cat’s somatic sensory cortex. J Neurophysiol. 1957;20 (4 ):408–434. doi: 10.1152/jn.1957.20.4.408 13439410
2 Hubel DH , Wiesel TN . Shape and arrangement of columns in cat’s striate cortex. J Physiol. 1963;165 (3 ):559–568. doi: 10.1113/jphysiol.1963.sp007079 13955384
3 Petersen CC . The functional organization of the barrel cortex. Neuron. 2007;56 (2 ):339–355. doi: 10.1016/j.neuron.2007.09.017 17964250
4 Yacoub E , Harel N , Ugurbil K . High-field fMRI unveils orientation columns in humans. Proc Natl Acad Sci U S A. 2008;105 (30 ):10607–10612. doi: 10.1073/pnas.0804110105 18641121
5 Nasr S , Polimeni JR , Tootell RB . Interdigitated Color- and Disparity-Selective Columns within Human Visual Cortical Areas V2 and V3. J Neurosci. 2016;36 (6 ):1841–1857. doi: 10.1523/JNEUROSCI.3518-15.2016 26865609
6 Zimmermann J , Goebel R , De Martino F , van de Moortele PF , Feinberg D , Adriany G , et al . Mapping the organization of axis of motion selective features in human area MT using high-field fMRI. PLoS ONE. 2011;6 (12 ):e28716. doi: 10.1371/journal.pone.0028716 22163328
7 De Martino F , Moerel M , Ugurbil K , Goebel R , Yacoub E , Formisano E . Frequency preference and attention effects across cortical depths in the human primary auditory cortex. Proc Natl Acad Sci U S A. 2015;112 (52 ):16036–16041. doi: 10.1073/pnas.1507552112 26668397
8 Mountcastle VB , Davies PW , Berman AL . Response properties of neurons of cat’s somatic sensory cortex to peripheral stimuli. J Neurophysiol. 1957;20 (4 ):374–407. doi: 10.1152/jn.1957.20.4.374 13439409
9 Adams DL , Sincich LC , Horton JC . Complete pattern of ocular dominance columns in human primary visual cortex. J Neurosci. 2007;27 (39 ):10391–10403. doi: 10.1523/JNEUROSCI.2923-07.2007 17898211
10 Jones EG , Rakic P . Radial columns in cortical architecture: it is the composition that counts. Cereb Cortex. 2010;20 (10 ):2261–2264. doi: 10.1093/cercor/bhq127 20667930
11 Goldman-Rakic PS , Schwartz ML . Interdigitation of contralateral and ipsilateral columnar projections to frontal association cortex in primates. Science. 1982;216 (4547 ):755–757. doi: 10.1126/science.6177037 6177037
12 Horton JC , Adams DL . The cortical column: a structure without a function. Philos Trans R Soc Lond B Biol Sci. 2005;360 (1456 ):837–862. doi: 10.1098/rstb.2005.1623 15937015
13 Chan AM , Dykstra AR , Jayaram V , Leonard MK , Travis KE , Gygi B , et al . Speech-specific tuning of neurons in human superior temporal gyrus. Cereb Cortex. 2014;24 (10 ):2679–2693. doi: 10.1093/cercor/bht127 23680841
14 Mesgarani N , Cheung C , Johnson K , Chang EF . Phonetic feature encoding in human superior temporal gyrus. Science. 2014;343 (6174 ):1006–1010. doi: 10.1126/science.1245994 24482117
15 Leonard MK , Gwilliams L , Sellers KK , et al . Large-scale single-neuron speech sound encoding across the depth of human cortex. Nature. 2024;626 (7999 ):593–602. doi: 10.1038/s41586-023-06839-2 38093008
16 Stedelin B , Cleary D , Paulk A , Bourhis A , Dayeh S , Tchoe Y , et al . Implementation of High-Resolution Non-penetrating Cortical Thin-Film Electrodes in the Awake Craniotomy for Research. Neurosurgery. 2020;67 (Supplement_1 ).
17 Duraivel S , Rahimpour S , Chiang CH , Trumpis M , Wang C , Barth K , et al . High-resolution neural recordings improve the accuracy of speech decoding. Nat Commun. 2023;14 (1 ):6938. doi: 10.1038/s41467-023-42555-1 37932250
18 Halgren E , Dale AM , Sereno MI , Tootell RB , Marinkovic K , Rosen BR . Location of human face-selective cortex with respect to retinotopic areas. Hum Brain Mapp. 1999;7 (1 ):29–37. 9882088
19 Lachaux JP , Rodriguez E , Martinerie J , Varela FJ . Measuring phase synchrony in brain signals. Hum Brain Mapp. 1999;8 (4 ):194–208. doi: 10.1002/(sici)1097-0193(1999)8:4&lt;194::aid-hbm4&gt;3.0.co;2-c 10619414
20 Ganji M , Paulk AC , Yang JC , Vahidi NW , Lee SH , Liu R , et al . Selective Formation of Porous Pt Nanorods for Highly Electrochemically Efficient Neural Electrode Interfaces. Nano Lett. 2019;19 (9 ):6244–6254. doi: 10.1021/acs.nanolett.9b02296 31369283
21 Tchoe Y , Bourhis AM , Cleary DR , et al . Human brain mapping with multithousand-channel PtNRGrids resolves spatiotemporal dynamics. Sci Transl Med. 2022;14 (628 ):eabj1441. doi: 10.1126/scitranslmed.abj1441 35044788
22 Ganji M , Kaestner E , Hermiz J , et al . Development and Translation of PEDOT: PSS Microelectrodes for Intraoperative Monitoring. Adv Funct Mater. 2018;28 (12 ):1700232.
23 Hermiz J , Rogers N , Kaestner E , et al . A clinic compatible, open source electrophysiology system. Conference proceedings: Annual International Conference of the IEEE Engineering in Medicine and Biology Society IEEE Engineering in Medicine and Biology Society Annual Conference. 2016;2016 :4511–4514. doi: 10.1109/EMBC.2016.7591730 28325009
24 Oostenveld R , Fries P , Maris E , Schoffelen JM . FieldTrip: Open source software for advanced analysis of MEG, EEG, and invasive electrophysiological data. Comput Intell Neurosci. 2011;2011 :156869. doi: 10.1155/2011/156869 21253357
25 Delorme A , Makeig S . EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods. 2004;134 (1 ):9–21. doi: 10.1016/j.jneumeth.2003.10.009 15102499
26 Voytek B D’Esposito M , Crone N , Knight RT . A method for event-related phase/amplitude coupling. Neuroimage. 2013;64 :416–424. doi: 10.1016/j.neuroimage.2012.09.023 22986076
27 Liu W , Zheng N , Li X . Nonnegative matrix factorization for EEG signal classification. Paper presented at: International Symposium on Neural Networks 2004.
28 Kim J , He Y , Park H . Algorithms for nonnegative matrix and tensor factorizations: A unified view based on block coordinate descent framework. J Glob Optim. 2014;58 :285–319.
29 Benjamini Y , Hochberg Y . Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc B Methodol. 1995;57 (1 ):289–300.
30 Rosen BQ , Halgren E . A Whole-Cortex Probabilistic Diffusion Tractography Connectome. eNeuro. 2021;8 (1 ). doi: 10.1523/ENEURO.0416-20.2020 33483325
31 Azevedo FA , Carvalho LR , Grinberg LT , et al . Equal numbers of neuronal and nonneuronal cells make the human brain an isometrically scaled-up primate brain. J Comp Neurol. 2009;513 (5 ):532–541. doi: 10.1002/cne.21974 19226510
32 Chen F , Loizou PC . Predicting the intelligibility of vocoded speech. Ear Hear. 2011;32 (3 ):331–338. doi: 10.1097/AUD.0b013e3181ff3515 21206363
33 Souza P , Rosen S . Effects of envelope bandwidth on the intelligibility of sine- and noise-vocoded speech. J Acoust Soc Am. 2009;126 (2 ):792–805. doi: 10.1121/1.3158835 19640044
