Auditory and visual gratings elicit distinct gamma responses

This work was supported by Wellcome Trust/DBT India Alliance 14 (Senior fellowship IA/S/18/2/504003 to S.R.) and DBT-IISc Partnership Programme. D.G. is 15 thankful for the support from the senior research fellowship awarded by the Council of 16 Scientific and Industrial Research (CSIR). 17 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint

18 Sensory stimulation is often accompanied by fluctuations at high frequencies (>30Hz) in 19 brain signals. These could be “narrowband” oscillations in the gamma band (30-70 Hz) or 20 non-oscillatory “broadband” high-gamma (70-150 Hz) activity. Narrowband gamma 21 oscillations, which are induced by presenting some visual stimuli such as gratings and have 22 been shown to weaken with healthy aging and the onset of Alzheimer’s Disease, hold 23 promise as potential biomarkers. However, since delivering visual stimuli is cumbersome as 24 it requires head stabilization for eye tracking, an equivalent auditory paradigm could be 25 useful. Although simple auditory stimuli have been shown to produce high-gamma activity, 26 whether specific auditory stimuli can also produce narrowband gamma oscillations is 27 unknown. We tested whether auditory ripple stimuli, which are considered an analogue to 28 visual gratings, could elicit narrowband oscillations in auditory areas. We recorded 64-29 channel EEG from male and female (18 each) subjects while they either passively fixated on 30 the monitor while viewing static visual gratings, or listened to stationary and moving ripples, 31 played using loudspeakers, with their eyes open or closed. We found that while visual 32 gratings induced narrowband gamma oscillations with suppression in the alpha band (8-33 12Hz), auditory ripples did not produce narrowband gamma but instead elicited very strong 34 broadband high-gamma response and suppression in the beta band (14-26Hz). Even though 35 we used equivalent stimuli in both modalities, our findings indicate that the underlying 36 neuronal circuitry may not share ubiquitous strategies for stimulus processing. 37 38 Significance statement 39 In the visual cortex, gratings can induce robust narrowband gamma oscillations (30-70Hz). 40 These visual stimulus-induced oscillations can further be used as a biomarker for diagnosing 41 neuronal disorders. However, tasks used to elicit these oscillations are challenging for elderly 42 subjects, and therefore, we tested if we could use auditory stimuli instead. We hypothesized 43 that auditory ripple stimuli, which are analogous to visual gratings, may elicit these 44 narrowband oscillations. We found that ripples induce a broadband high-gamma response 45 (70-150Hz) in human EEG, unlike visual gratings that produce robust gamma. Thus, the 46 underlying neural circuitry in the two areas may not be canonical. 47 48 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint

49 Modulations in the gamma band ( ≥ 30Hz) are associated with higher cognitive processes 50 (Tallon-Baudry et al., 1999; Ray and Maunsell, 2010) and sensory representation (Gray and 51 Singer, 1989; Brosch et al., 2002). Sensory input driven narrowband gamma oscillations 52 (~30-70Hz) have been identified in various species and neocortical areas (Adrian, 1942; Gray 53 and Singer, 1989; Gieselmann and Thiele, 2008; Muthukumaraswamy and Singh, 2013) and 54 are thought to be due to the reciprocal interaction between excitatory glutamatergic pyramidal 55 neurons and inhibitory GABAergic interneurons (Cardin et al., 2009). Therefore, only the 56 stimuli that drive the neuronal population in a time-synchronized manner can induce such 57 oscillations. Cartesian gratings are considered an archetype to induce sustained narrowband 58 gamma in the visual cortex. The magnitude and peak frequency of these oscillations depend 59 on low-level grating features such as size, contrast, orientation, and spatial and temporal 60 frequency (Gieselmann and Thiele, 2008; Ray and Maunsell, 2010, 2011; Murty et al., 2018). 61 Recent studies have shown that the amplitude of oscillations induced by gratings in 62 electroencephalography (EEG) recording decreases with healthy ageing (Murty et al., 2020) 63 and is weaker in patients with mild cognitive impairment and Alzheimer’s disease (Murty et 64 al., 2021) compared to age and gender-matched healthy control subjects. This suggests that 65 these oscillations can be potentially used as a biomarker for identifying cognitive decline. 66 However, such studies typically require eye fixation and tracking while a full-screen, high 67 luminance-contrast grating is presented on the screen. Therefore, the task becomes 68 challenging as it may lead to visual discomfort (Wilkins et al., 1984), especially for elderly 69 subjects. 70 Replacing the visual stimulus with an auditory one would resolve such challenges, as subjects 71 can sit with closed eyes and passively listen to auditory stimuli, provided that the auditory 72 stimulus induces a narrowband rhythm. In-vitro studies of rat auditory cortex have been 73 shown to elicit a narrowband oscillation (30-80Hz) in response to stimulation (Ainsworth et 74 al., 2016) and have shown to have distinct generators for lower (30-45Hz) and higher gamma 75 (50-80Hz) (Ainsworth et al., 2011). However, in in-vivo recordings with auditory stimuli 76 such as pure tones (Crone et al., 2001; Brosch et al., 2002; Edwards et al., 2005; 77 Steinschneider et al., 2008; Fujioka et al., 2009), short tone bursts (Trautner et al., 2006; 78 Vianney-Rodrigues et al., 2011), phonemes (Crone et al., 2001; Edwards et al., 2009), clicks 79 (Brugge et al., 2009), frequency sweeps (Jeschke et al., 2008; Lenz et al., 2008), noise 80 (Griffiths et al., 2010; Sedley et al., 2012), words (Canolty et al., 2007), and sentences (Billig 81 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint et al., 2019), an increase in power occurs mainly at frequencies higher than 60 Hz. These are 82 “broadband” increases and may reach up to ~200Hz, with unequal power increases across 83 frequencies (Crone et al., 2011). Hence, auditory stimulus-induced narrowband gamma 84 oscillations have not been shown unequivocally. 85 Even in the visual cortex, narrowband gamma is strongly elicited only by specific stimuli 86 such as bars (Gray and Singer, 1989), gratings (Jia et al., 2011; Murty et al., 2018) and 87 certain iso-luminant hues (Shirhatti and Ray, 2018). Therefore, we tested whether auditory 88 ripple stimuli, whose attributes match that of a visual grating in terms of feature complexity 89 (Shamma, 2001) and neural representation (deCharms et al., 1998), might induce auditory 90 narrowband gamma oscillations (see Discussion for more details). Ripples are generated by 91 superimposing multiple sinusoidally amplitude-modulated tones (Kowalski et al., 1996). 92 They are parametric, meaning they can be fully characterized using limited features that can 93 be changed independently while maintaining their ethological relevance, as their spectra 94 match that of natural vocalizations (Langers et al., 2003). We recorded 64-channel EEG from 95 human subjects who passively listened to ripple sounds from a loudspeaker with their eyes 96 open or closed or passively fixated on the computer screen during the presentation of a full-97 screen grating stimulus on the monitor and compared the gamma responses generated by 98 these stimuli. 99 100

101 Human Subjects 102 We recruited 36 healthy subjects (aged 22-38 years, a mean of 26.6±3.7 years, 18 females) 103 for the study from the Indian Institute of Science community. Participants reported having 104 normal hearing levels with no abnormalities and had corrected to normal vision (except for 105 one participant with strabismus). All subjects, barring one, were right-handed. Participation 106 was voluntary, and written informed consent was obtained from all the subjects after briefing 107 them about the experimental procedure. Subjects were given monetary compensation for their 108 time and effort. Experiments were performed according to the protocol approved by the 109 Institutional Human Ethics Committee of the Indian Institute of Science, Bangalore. 110 111 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint EEG setup and data acquisition 112 Raw EEG signals were recorded from 64-channel active electrodes (actiCap) using the 113 BrainAmp DC EEG acquisition system (Bra in Products GmbH). Electrodes were placed 114 using the international 10-10 standard reference scheme with FCz as the reference electrode 115 (unipolar reference scheme). Raw signals were sampled at 1000Hz and were filtered between 116 0.016Hz (first-order filter) and 250Hz (fifth-order Butterworth filter). Signals were 117 digitalized at 16-bit resolution (0.1µV/bit). Impedance values were kept below 25k Ω for the 118 entire recording duration. Average impedances of the final set of electrodes were 6.88±3.06 119 KΩ (mean± std) for the visual task, 6.76±2.99 for the eye-close auditory task and 7.29±3.02 120 KΩ for the eye-open auditory task. 121 122 Experimental setting and task 123 All thirty-six subjects did the visual and eye-close auditory tasks; twelve subjects out of these 124 also participated in the eye-open auditory task. The first twenty-four (twelve females) 125 subjects performed the visual protocol first, followed by the auditory protocols, which 126 comprised the eye-close version and two other auditory protocols (not described in this 127 study). For these participants, auditory protocols were run in a counterbalanced order. The 128 remaining twelve subjects completed the visual, eye-close and eye-open auditory task 129 versions. These protocols were counterbalanced. 130 Visual Task 131 Each participant performed a passive fixation task. They sat in a dark room in front of a 132 gamma-corrected LCD monitor (BenQ XL2411; resolution 1280x720 pixels; refresh rate 133 100Hz; mean luminance 60 cd/m 2). The monitor was placed 58cm away from the subject’s 134 eyes, so the full-screen stimulus subtended the width and height of 49.4° and 29° of the visual 135 field. The visual stimuli, sinusoidal luminance grating, were presented using the NIMH 136 MonkeyLogic software tool on MATLAB (The MathWorks, Inc; RRID: SCR_001622). The 137 marker for stimulus onset was recorded in the EEG file by using a digital I/O card (National 138 Instruments USB 6008 or USB 6210 multifunctional I/O device). 139 Since the auditory task (see below) involved the presentation of a continuous sequence of 140 auditory stimuli with some inter-stimulus interval, we modified our visual task to be 141 comparable to the auditory task. Specifically, unlike our previous studies (Murty et al., 2018, 142 2020, 2021), we presented visual stimuli in a single long continuous sequence where each 143 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint stimulus was presented for 800ms followed by an inter-stimulus interval of 700ms. A square 144 fixation spot of 0.2° was shown in the center of the screen throughout the task duration. 145 Subjects were instructed to hold and maintain their fixation whenever the visual stimulus was 146 presented and blink or break fixation (if needed) during the inter-stimulus interval. Eye 147 position was monitored continuously (see details below), and epochs with breaks in fixation 148 were discarded offline later. Visual stimuli were achromatic luminance full contrast and full-149 screen gratings presented at one of two spatial frequencies (SF - 2 and 4 cycles per degree) 150 and one of four orientations (Ori - 0°, 45°, 90° and 135°), generating eight different kinds of 151 stimuli. Each stimulus type was repeated 25 times, yielding 200 trials. We chose these 152 stimulus parameters as they induce a robust gamma response (Murty et al., 2018). The total 153 duration was ~5 minutes. For a couple of subjects, the stimulus sequence was paused for 30-154 60 seconds because they requested a break, after which the sequence was resumed. 155 Auditory Task 156 The task had an eye-close and eye-open version. In the eye-close session, subjects sat in a 157 dark room with closed eyes and were instructed to listen passively to the sounds. They were 158 instructed to keep their eyes closed to minimize eye movement or blink artefacts. To make 159 the auditory task equivalent to the visual task, we ran an eye-open version on a subset of 160 subjects, where subjects had to fixate on the screen where a fixation spot of 0.2° was shown 161 at the center, and sounds were played in the background. The sessions were conducted in a 162 quiet room to minimize irrelevant sounds. The stimuli were played using a multidirectional 163 speaker (Marshall Kilburn II) at 75-80 dB. The sound stimuli were generated in MATLAB by 164 custom-written code. They were presented using the NIMH MonkeyLogic software tool on 165 MATLAB. The marker for stimulus onset was recorded in the EEG file by using a digital I/O 166 card (National Instruments USB 6210 multifunctional I/O device). 167 Each trial had one stimulus of 800ms followed by an inter-trial interval of 800ms. The 168 stimulus set consisted of spectro-temporally modulated ripples. Ripple stimuli have a 169 broadband carrier with a sinusoidally varying spectral envelope that drifts along the 170 logarithmic frequency axis at a constant velocity (Kowalski et al., 1996). The stimuli 171 composed of 80 tones equally spaced along the logarithmic axis, spanning 5 octaves (250-172 8000Hz, 16 tones per octave), sampled at 44100Hz, were used. The amplitude of all 173 individual tones was sinusoidally modulated on a linear scale and was modulated at 90% or 174 10dB. Each ripple stimulus had 10ms – on/off ramps. The stationary ripple stimuli were 175 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint presented at one of three spectral modulation frequencies or ripple density ( Ω - 0, 0.8 and 1.6 176 cycles/octave). A stationary ripple can be represented mathematically as – 177 /g1845 /g4666 /g2312 /g4667 /g34041/g3397∆ /g1827 s i n /g4666 2 /g2024 Ω /g2312/g3397 Φ /g4667 (1) 178 where ∆/g1827 , modulation depth, is 0.9; /g2312 is the logarithmic frequency axis (in octaves), defined 179 as /g2312/g3404/g1864 /g1867 /g1859 /g2870/g4666 /g3033 /g3033/g3290 /g4667 where /g1858 /g3042 is the lower frequency edge, i.e., 250Hz and /g1858 is the component 180 tone frequency. The starting phase of the ripple ( Φ ) was defined relative to the low low-181 frequency edge of the spectrum and kept at 0. The moving ripples were obtained by 182 temporally modulating stationary ripples, such that the envelope moved downwards towards 183 lower frequencies at one of the four velocities constantly ( /g2033 - 0, 5, 10, 20 cycles/second 184 (Hz)). The moving ripple spectro-temporal spectrum can be represented by – 185 /g1845 /g4666 /g2312 /g4667 /g3404 1 /g3397 ∆/g1827 sin /g46662/g2024/g4666 /g2033/g2308 /g3397 Ω/g2312/g4667 /g3397 Φ/g4667 (2) 186 We chose these stimulus parameters based on the previous study done with ripple stimuli in 187 humans (Langers et al., 2003). This generated 12 possible combinations, and each stimulus 188 type was repeated 40 times, totalling 480 trials. The total task duration was about ~15 189 minutes, divided into two blocks with 2–3-minute breaks in between according to the 190 subject’s comfort. 191 Owing to their spectro-temporal response profile, ripple envelopes look like visual gratings 192 and are thus often described as acoustic analogues of visual gratings (Shamma, 2001). The 193 spectral modulation frequency can be considered equivalent to spatial frequency, which 194 determines how dense the gratings are. The modulation depth of the ripple envelope is 195 analogous to the contrast of visual gratings; similarly, temporal modulation is like the 196 temporal frequency of drifting visual gratings. Given the similarity between features, ripples 197 are referred to as auditory gratings. In the rest of the paper, we will refer to these stimuli as 198 auditory gratings. Our rationale for using auditory gratings is discussed in more detail in the 199 Discussion. 200 201 Eye position Analysis 202 Eyes signals were recorded either using Eyelink Portable Duo head-free eye tracker or 203 Eyelink 1000 (SR Research Ltd, sampled at 1000Hz) for the entire duration of the visual and 204 eye-open version of the auditory task. Before the start of each session, the eye tracker was 205 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint calibrated for pupil position and monitor distance. No online artefact rejection was done. 206 During analysis, we segmented epochs of EEG data between -848ms to 1650ms around each 207 stimulus onset. We obtained 200 and 480 such segments for the visual and auditory tasks, 208 respectively. Each segment is referred to as a “trial”, even though this task had no trial 209 structure as the stimuli were presented continuously. We rejected trials offline, which had 210 fixation breaks, defined as eye blinks or shifts in eye position outside a square window of 211 width 5° centred on the fixation spot during -500ms to 750ms of stimulus onset. This interval 212 was chosen to ensure that the “baseline” (-500ms to 0ms of stimulus onset) and “stimulus” 213 (250ms to 750ms) periods used for the calculation of the power spectral density (PSD) were 214 both free of eye-movement-related artefacts. It led to the rejection of 20.58±16.72% (mean ± 215 std) trials from the visual task, barring one subject for which all trials were labelled bad as the 216 subject had strabismus. For the auditory eye-open task, we rejected 19.46±16 % of trials. 217 218 Artefact Rejection 219 We used a fully automated artefact rejection pipeline (see (Murty and Ray, 2022), for details) 220 with one minor modification. In our previous method, the threshold for rejection was based 221 on the standard deviation (SD) in the time-series data (any trial in which any time point 222 between -0.5 to 0.75 seconds deviated by more than 6 SD was rejected), but this threshold 223 could vary depending on the outliers. Here, we chose hard cutoffs, which helped us evade this 224 issue altogether. Specifically, after rejecting all electrodes with impedance > /i3 25kΩ and bad 225 eye trials (as done in the previous study), we first calculated the root mean squared (RMS) 226 value for each trial (-0.5s to 1.5s) after passing it through a high pass filter of 1.6Hz (to 227 remove any slow drifts), and applied a fixed threshold bound (upper RMS cutoff of 35µV and 228 lower RMS cutoff of 1.5µV) and labelled any trial that lay outside that bound as bad for that 229 electrode. 230 Next, we computed multi-tapered PSD for the rest of the trials (using the Chronux toolbox; 231 (version 2.10) (Bokil et al., 2010), RRID: SCR_005547 ), available at http://chronux.org). 232 Any repeat for which the PSD deviated by six times the standard deviation from the mean at 233 any frequency point (between 0 -200Hz) was also labelled bad. After this, we listed a 234 common set of bad repeats across all 64 electrodes. We discarded the electrodes with more 235 than 30% of all repeats labelled as bad. Any trial was labelled bad if it occurred in more than 236 10% of the remaining electrodes. Next, we selected a subset of occipital, parieto-occipital 237 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint (O1, Oz, O2, P1, P2, P3, P4, PO3, POz, PO4 – set used in Murty et al., 2018, 2020) and 238 temporal electrodes (TP9, TP7, TP8, TP10), and any trial that was marked as bad in any of 239 these electrodes were also included in the set of common bad trials. This was done to ensure 240 that the electrodes used for the calculation of power for the visual and auditory conditions 241 were artefact-free. 242 These criteria led to the rejection of less than ~30% (28.5± 9.1%) of the data collected for the 243 eye-close auditory task. For the visual and eye-open auditory tasks, these criteria and offline 244 eye rejection led to the rejection of 36.73 ± 12.89% and 41.90 ± 11.06% of data, respectively. 245 Note that this relatively higher percentage of bad trials compared to our previous studies is 246 due to the continuous stimulus presentation paradigm used in this study. In previous studies, a 247 trial had 2-3 stimuli, after which the subjects could break fixation or blink their eyes. But 248 here, the stimulus presentation was in a continuous stream, and subjects were allowed to 249 blink (if needed) during the inter-stimulus interval; hence, more data segments (“trials”) had 250 to be discarded. 251 As an additional criterion to reject electrodes, we calculated the slope in the range of 56-86Hz 252 of baseline PSD (averaged across all good trials) for each electrode by fitting the PSD with a 253 power-law function (for details, refer to Murty et al., 2020). We discarded any electrode for 254 which the PSD (1 taper) slope was less than 0. Even after applying stringent conditions to 255 reject electrodes, if some electrode that had not been rejected yet was visually noisy, we 256 manually declared that electrode as bad. We removed an additional 1.7% of the total 257 electrodes by doing this. 258 We then rejected any subject from the analysis of visual or auditory tasks with less than 50% 259 of the occipital (O1, Oz, O2, PO7, PO3, POz, PO4, PO8, Iz) or temporal (FT9, FT7, T7, TP7, 260 TP9, FT10, FT8, T8, TP8, TP10) group electrodes, respectively. This way, we rejected two 261 participants from the eye-close auditory task and one from the eye-open auditory task. For the 262 visual task, 2 participants were rejected as one had strabismus, and the baseline signal was 263 extremely noisy for the other. 264 265 EEG data analysis 266 As in our previous study (Murty et al., 2020), we used both unipolar and bipolar reference 267 schemes for analysis. For the unipolar reference scheme, we considered the following 268 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint electrodes – TP9, TP7, TP8, and TP10 for auditory task analysis and P1, P2, P3, P4, PO3, 269 POz, PO4, O1, Oz, O2 for visual task analysis. For the bipolar referencing scheme, we re-270 referenced data from each electrode data to its neighbouring electrode. We thus obtained 114 271 bipolar pairs out of 64 unipolar electrodes. We have used the following bipolar combinations 272 – TP9–TP7, TP7–T7, TP7–P7, TP7–CP5, TP10–TP8, TP8–T8, TP7–P8, TP7–CP6 for 273 auditory task analysis and PO3–P1, PO3–P3, POz-PO3, PO4–P2, PO4–P4, POz-PO4, Oz-274 POz, Oz-O1 and Oz-O2 for visual task analysis. Depending upon the reference scheme, data 275 were pooled for all these electrode or electrode combinations. 276 All the data were analyzed using custom-written codes in MATLAB. Using the Chronux 277 toolbox, we computed multi-tapered PSD and time-frequency spectrograms using a single 278 taper. We chose a period between -500ms and 0 ms (0ms marks the stimulus onset) as the 279 baseline and a period between 250ms to 750ms as the stimulus period, with a frequency 280 resolution of 2Hz. For spectrograms, we used a moving window of size 250ms and a step size 281 of 25ms, thus yielding a frequency resolution of 4Hz. We calculated the change in power for 282 narrowband gamma oscillation ( /g1858 /g1488 /g4670 20 /g3398 66 /g4671 /g1834/g1878 ) and high-gamma activity ( /g1858 /g1488 /g4670 70 /g3398283 150/g4671/g1834/g1878 ) as follows: 284 ∆/g1842/g1867/g1875/g1857/g1870 /g3404 10 /g1499 /g4666/g1864/g1867/g1859 /g2869/g2868 ∑ /g1845/g1846 /g4666/g1858/g4667 ∑ /g1828/g1838/g4666/g1858/g4667 /g4667 ST and BL represent power across frequency( f) for stimulus and frequency averaged across 285 all stimuli repeats for a particular condition and electrodes. 286 Scalp maps were generated using the topoplot.m function of the EEGLAB toolbox ((Delorme 287 and Makeig, 2004) RRID: SCR_007292). The function was modified to show each electrode 288 as a coloured disc, with the colour representation change in power for a particular frequency 289 range in decibels (dB). 290 291 Statistical Analysis 292 We compared the means of the subject averaged power spectral density during stimulus and 293 baseline periods using a paired t-test test. For comparing the change in power across visual 294 and auditory protocols, unpaired t-tests with unequal variance and F-statistic were used. 295 296 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint Data and Code availability 297 Spectral analyses of the data were performed using the Chronux toolbox. Raw data will be 298 made available to readers upon reasonable request. 299 300 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint

301 We collected 64-channel EEG data from 36 subjects (18 females). Participants passively 302 listened to auditory grating stimuli or passively fixated on the screen while presenting full-303 screen sinusoidal grating stimuli on a monitor. Each stimulus was 800ms long in either 304 modality, preceded by a baseline period of 800ms for auditory and 700ms for visual stimuli. 305 306 Fig.1 shows the subject averaged time-frequency spectrograms for change in power from 307 baseline (-500 to 0ms of stimulus onset) for each electrode as positioned on the scalp for 308 visual (right) and auditory protocol (left). The spectrograms are averaged across all stimulus 309 conditions. Visual grating stimuli elicited a robust narrowband gamma response in occipital 310 electrodes (in the range of 22-64Hz) and a pronounced alpha (8-14Hz) suppression. However, 311 auditory gratings did not produce such a narrowband gamma response. In contrast, they 312 induced a prominent high-gamma activity (70-150Hz) in electrodes located near mastoids 313 and suppression in beta rhythms (14-26Hz) across all electrodes. The response elicited by the 314 auditory gratings was much weaker than that elicited by the visual grating stimuli (note the 315 difference in scale in A versus B). For visual stimuli, some frontal electrodes showed a 316 broadband response after stimulus offset, likely due to artefacts related to eye movements or 317 blinks (we removed only those trials for which eye movement occurred between -500 to 318 750ms of stimulus onset; see Methods for more details). This frontal response was also 319 observed for the auditory protocol when eyes were open, as shown later. 320 321 Next, we averaged the responses across selected electrodes (as shown in Figure 1 inset; see 322 EEG data analysis section in Methods) depending upon the stimulus modality. Fig 2A (Top 323 row, left panel) shows that visual grating stimuli elicit two distinct gamma bands, termed 324 slow gamma (~20-34Hz) and fast gamma (~36-66Hz) in spectrograms (Murty et al., 2018, 325 2020). On the other hand, auditory grating stimuli elicited a high-gamma response (70-326 150Hz), Fig 2A (Bottom row, left panel). Our previous studies showed that using a bipolar 327 referencing scheme improved the narrowband gamma response (Murty et al., 2020), so we 328 performed the same analysis using bipolar referencing as well (Figure 2A; right panel). There 329 was a marginal improvement in the visual-induced narrowband gamma response, especially 330 in the fast gamma band. However, the auditory stimulus-induced high-gamma activity 331 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint weakened with bipolar referencing (Fig. 2A bottom panels). The change in power in the 332 narrowband gamma range was about ~1dB when visual stimuli were presented (Figure 2B, 333 second row), which was missing for auditory stimuli (Figure 2B, fourth row). Conversely, the 334 change in high-gamma power was ~0.05 dB for visual stimuli and ~0.1dB for auditory 335 stimuli for the unipolar case (note the difference in scales in the plots). The topoplots in Fig 336 2C showed that visually induced gamma was localized in the occipital and parieto-occipital 337 region, while auditory stimuli induced broadband high-gamma in electrodes located near the 338 mastoids. For further analysis, we have used unipolar referencing. 339 340 Gamma and high-gamma responses were uncorrelated across subjects 341 Figure 3 shows the change in power spectrograms (for the same set of visual and auditory 342 electrodes as Figure 2) for individual subjects, sorted by decreasing auditory high-gamma 343 power. Although auditory high-gamma was much weaker than visual narrowband gamma, it 344 appeared stronger than visual high-gamma. The range of auditory stimuli induced broadband 345 high-gamma activity varied among the subjects. Further, visual and auditory stimuli had no 346 consistent effects on subjects – subjects with strong auditory high-gamma did not have 347 stronger visual narrow or broadband high-gamma, or vice-versa. 348 349 Figure 4 shows that visual narrowband gamma was significantly stronger than auditory 350 narrowband gamma, which was negligible (Visual: 0.4±0.09, Auditory: -0.06±0.02 351 (mean±sem), p = 8.6x10 -5, N = 34, unpaired t-test, F = 15.33). In contrast, broadband high-352 gamma showed the opposite trend, but the difference was not significant (Visual: 0.061±0.02, 353 Auditory: 0.12±0.03 (mean±sem), p = 0.118, N = 34, unpaired t-test, F = 0.3646, Fig 4E). 354 Since the auditory response was computed over fewer electrodes, the visual high-gamma 355 might have been high for a subset of selected visual electrodes, but the mean value could 356 have been reduced when averaged over all the selected electrodes. To rule this out, we 357 performed the same analysis after taking only a single “best” electrode for each subject with 358 the most robust response for each modality in the respective frequency range (Fig 4C and 359 4D). The trends remained similar, with the broadband auditory response now significantly 360 higher than visual (narrowband gamma:Visual: 0.80±0.13, Auditory: 0.05±0.03 (mean±sem), 361 p = 1.4x10-6, N = 34, unpaired t-test, F = 16.65; broadband high-gamma: Visual: 0.21±0.02, 362 Auditory: 0.36±0.06 (mean±sem), p = 0.0189, N = 34, unpaired t-test, F = 0.1477, Fig 4E). 363 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint Further, Pearson’s correlation between the responses to visual and auditory stimuli was 364 insignificant for any condition (p-values are shown in the plots in 4A-D). 365 366 No tuning characteristics or stimulus selectivity could be observed in EEG 367 The previous results were the average responses across eight and twelve different stimuli for 368 visual and auditory conditions, respectively (see Methods for details). To test whether 369 gamma/high-gamma was selective for some stimuli, we plotted the change in time-frequency 370 power spectra for different stimulus conditions (Figure 5). We did not observe gamma 371 activity tuned to any particular stimulus in either of the modalities. To eliminate the 372 possibility that individual subjects were tuned to different stimulus features, we determined 373 their coefficient of variation (CV) of power across stimulus conditions. Specifically, during 374 the stimulus period (250ms to 750ms), we calculated a ratio of average power in narrowband 375 gamma and broadband high-gamma frequency ranges across stimulus conditions to its 376 standard deviation for each selected electrode, depending on the task for each subject. The 377 values were then averaged across electrodes. The values were generally small (Narrowband: 378 Visual: 0.09±0.008, Auditory: 0.071±0.005 (mean±sem); p = 0.0475, N=34, unpaired t-test, F 379 = 2.12; Broadband: Visual: 0.05±0.003, Auditory: 0.066±0.004 (mean±sem); p = 0.0475, 380 N=34, unpaired t-test, F = 0.6047) indicating poor selectivity towards any stimulus feature. 381 These results are consistent with our previous study where we found strong orientation 382 selectivity for narrowband gamma in local field potential (LFP) data from monkeys, which 383 was absent in human and monkey EEG (see Figure 2E of Murty et al., 2018). 384 385 Broadband High-gamma activity remains the same for the eye-close and eye-open auditory 386 tasks 387 Sensory-driven gamma oscillations are known to be modulated by arousal (Vinck et al., 388 2015). Since subjects had their eyes closed for the auditory task, whereas they were instructed 389 to passively fixate on the monitor for the visual task, we ran an eye-open version of the 390 auditory task where subjects passively fixated on the screen while passively listening to the 391 sounds to ensure a similar arousal level. We fo und a similar increase in high-gamma activity 392 and suppression in the beta band for the auditory eye-open task as well (Figure 6). We also 393 saw increased broadband activity after stimulus offset attributed to eye artefacts, similar to 394 those observed in the visual task. The high-gamma power was comparable for eyes-open and 395 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint eyes-closed cases: (for high-priority electrodes: Eyes-open: 0.093±0.05 and Eyes-closed: 396 0.116±0.11 (mean±sem), p = 0.7145, N = 11, unpaired t-test, F = 1.93; for best electrode: 397 Eyes-open: 0.297±0.03 and Eyes-closed: 0.346±0.11 (mean±sem), p = 0.7592, N = 11, 398 unpaired t-test, F = 0.88). The power values were significantly correlated for the best 399 electrode, for which maximum change in power was calculated. (Pearson’s linear coefficient: 400 r = 0.75, p = 0.0079). 401 402

403 We tested whether narrowband gamma oscillations can be induced by auditory grating 404 stimuli analogous to that induced by visual sinusoidal luminance gratings in the visual areas 405 in human EEG during passive stimulus presentation tasks. We found that the auditory 406 gratings induced an extremely focal broadband high-gamma response (~70-150 Hz) in 407 temporal electrodes and a widespread decrease in beta rhythm (~14-26Hz). In contrast, visual 408 grating stimuli induced narrowband gamma oscillations (~20-70Hz) in the occipital and 409 parietooccipital electrodes accompanied by suppression in alpha oscillations (~8-14Hz) in the 410 same subjects. The auditory grating-induced broadband activity was weaker than narrowband 411 oscillations elicited by the visual gratings but was still robust than their broadband response. 412 Subjects which showed an induced response to either stimulus did not respond similarly to 413 the other stimulus modality, indicating that the networks responsible for the induced 414 responses work independently in each modality. We observed poor tuning of these responses 415 towards specific stimuli of either modality, consistent with poor stimulus selectivity for 416 visual gamma observed earlier in human EEG (Murty et al., 2018). 417 418 Comparison with previous studies 419 Several studies have reported auditory broadband high-gamma activity in response to a 420 multitude of stimuli and various aspects of auditory cortical processing in primates. When 421 simple stimuli such as pure tones (Crone et al., 2001; Edwards et al., 2005; Steinschneider et 422 al., 2008; Fujioka et al., 2009), tone bursts (Trautner et al., 2006), and clicks (Brugge et al., 423 2009) were used, only a transient increase in broadband high-gamma was reported. However, 424 our study obtained sustained high-gamma responses throughout the stimulus presentation; 425 this might have been because the neurons in the auditory cortex respond better to complex 426 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint stimuli than pure tones (Tian and Rauschecker, 2004) and are capable of firing in a sustained 427 fashion only when optimal stimuli are used (Wang et al., 2005). Thus, auditory gratings are 428 better suited to drive neurons in auditory areas. These results also align with the results of 429 other studies which used complex stimuli, such as frequency sweeps (Lenz et al., 2008), 430 phonemes (Edwards et al., 2009), words (Canolty et al., 2007), and sentences (Billig et al., 431 2019). A recent study (Ross et al., 2020) also reported increased evoked high-gamma activity 432 (90-150Hz) in response to a syllable presentation in human EEG. The polarity of response at 433 electrodes TP9 and TP10 electrodes was opposite to the results that we obtained. However, 434 these differences could be accounted for by the fact that the location of the reference 435 electrode affects EEG measurements (Nunez and Srinivasan, 2006); in our study, it was 436 located at FCz and was located on the nose in their study. 437 438 Narrowband gamma oscillations to auditory stimuli have only been reported in studies 439 performed on rodents. In-vitro studies done on rat A1 neocortical slices have shown the 440 presence of narrowband gamma in the range of 30-80 Hz in isolation (Ainsworth et al., 441 2016), i.e. without any broadband activity and further showed that two distinct local networks 442 can give rise to such oscillations in this frequency range (Ainsworth et al., 2011). However, 443 studies in awake rats and Mongolian gerbils have shown narrowband gamma response (~30-444 70 Hz) along a broadband increase in high-gamma activity till 150 Hz (Jeschke et al., 2008; 445 Vianney-Rodrigues et al., 2011). Lenz and colleagues (2008) repeated the study conducted in 446 gerbils using the same task and stimulus set in human EEG, and they observed an increase in 447 frequencies from 100Hz up to 250Hz without any narrowband gamma oscillation at lower 448 frequencies. The authors indicated that this may reflect species-specific differences and that 449 the underlying neuronal generators may differ. So, we must be careful while assuming the 450 generalizability of results across species. 451 452 The studies that involved the presentation of speech stimuli (Canolty et al., 2007; Edwards et 453 al., 2009; Billig et al., 2019) also reported beta desynchronization, similar to the decrease that 454 we observed. We also note that beta desynchronization was observed when participants 455 performed target detection tasks with simpler auditory stimuli, which was a widespread 456 signal across many electrodes in the EEG (Mazaheri and Picton, 2005), similar to our results. 457 Given that the spectrotemporal envelope of auditory gratings is similar to that of speech and 458 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint can mimic formant transitions between the vowels and accommodate pitch changes (Langers 459 et al., 2003), a decrease in beta might point towards the activation of circuits involved in 460 complex stimuli and/or task processing. 461 462 Mechanisms of gamma oscillations and high-gamma activity 463 Periodic activation of neuronal assemblies at an optimum time window gives rise to gamma 464 cycles (Buzsáki and Wang, 2012). Previous studies have pointed out a key role of 465 interneurons in the generation of narrowband gamma oscillations in the sensory cortex 466 (Whittington et al., 1995). Synaptic inhibition comes from fast-spiking parvalbumin-467 expressing and regular-spiking somatostatin-expressing GABAergic interneurons, which 468 generate gamma oscillations in different frequency ranges (Chen et al., 2017). On the other 469 hand, high-gamma activity is thought to reflect population firing near the microelectrode for 470 invasive recordings and synchronous firing for macrosignals such as electrocorticogram (Ray 471 et al., 2008). It, therefore, has distinct origins compared to narrowband gamma (Ray and 472 Maunsell, 2011). Broadband responses are a robust indicator of neuronal firing in the 473 auditory cortex as well (Manning et al., 2009). 474 475 Auditory grating as an effective stimulus to produce narrowband gamma 476 As discussed in the Methods, auditory gratings have several properties that make them 477 analogous to visual gratings. In addition, we thought that they would be good candidates for 478 generating narrowband gamma for two reasons. First, they drive the auditory neurons very 479 strongly (deCharms et al., 1998). Visual stimuli that generate narrowband gamma in the 480 visual cortex, such as bars and gratings, also drive the neurons in the visual cortex strongly, 481 and in fact, the ideal filters for primary visual cortex (V1) neurons have oriented Gabor-like 482 features (Olshausen and Field, 1996). It is possible that when the local orientation of the 483 grating of a particular spatial frequency matches the preference of the local cortical neurons, 484 the strong excitatory drive can induce narrowband gamma. Similarly, early processing of 485 sounds and images share equivalent filter characteristics, and the sensory code in the auditory 486 system prefers broadband sounds with smooth edges (Lewicki, 2002). The tuning of the 487 primary auditory cortex for spectro-temporal frequencies of auditory gratings matches the 488 tuning found for spatial frequencies in the primary visual cortex (Kowalski et al., 1996). 489 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint Second, narrowband gamma in the visual cortex becomes stronger when the stimulus size 490 increases because of local inhibition generated by surround suppression (Gieselmann and 491 Thiele, 2008). In the auditory cortex, frequency information is spatially coded in tonotopic 492 maps; therefore, we speculated that broadband auditory gratings, which have power over five 493 octaves (250 – 8000Hz), to excite a large neuronal population in the auditory cortex, 494 replicating the effect of large-sized visual grating. Furthermore, auditory cortical neurons can 495 lock to the amplitude modulations of the envelope of the auditory gratings and thus produce 496 sustained responses (Elhilali et al., 2004). Further, the inhibitory circuitry in visual and 497 auditory cortices has similar properties. For example, PV+ interneurons in visual and auditory 498 areas are narrowly tuned for spatial frequency and orientation (Cardin et al., 2007) and 499 frequency (Moore and Wehr, 2013). Similarly, lateral inhibition is mediated by SOM cells in 500 the primary auditory cortex as well (Kato et al., 2017). Based on this, we speculated that 501 auditory gratings would elicit narrowband oscillations. But, contrary to our expectations, 502 auditory gratings induced a broadband response. A recent study has pointed out that the 503 reduced PV+ inhibition might enhance broadband high-gamma power due to asynchronous 504 activity (Guyon et al., 2021). So, perhaps auditory gratings activated the cortical neurons but 505 failed to do so synchronously. Differences could also be due to species, as rodent cortex has 506 been shown to generate narrowband gamma (as discussed above). Finally, it could be due to 507 the recording modality. Since the primary auditory cortex is buried in the Heschl’s gyrus, its 508 contribution to the EEG signal may be relatively minor and masked by brain regions on the 509 surface, which may have more complex responses to these auditory gratings. 510 511 Stimulus tuning to gamma and high-gamma responses 512 We observed a weak gamma tuning in EEG for visual and auditory gratings, though gamma 513 shows a strong tuning in the LFP (Ray and Maunsell, 2010; Jia et al., 2011). MEG studies 514 with visual stimuli have also shown a relatively stronger tuning of gamma oscillations 515 (Adjamian et al., 2004; Koelewijn et al., 2011). Even fMRI studies using auditory gratings 516 demonstrated that voxel activation and transfer functions show tuning towards specific 517 spectro-temporal features of these sounds (Langers et al., 2003; Schönwiesner and Zatorre, 518 2009). Weaker tuning in EEG could be due to volume conduction effects, which affect other 519 recording modalities (such as MEG) to a lesser degree. We showed strong stimulus tuning in 520 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint LFP and weak tuning in EEG simultaneously recorded from the same monkeys, which rules 521 out other confounds, such as task and species differences (Murty et al., 2018). 522 523 Although we failed to get narrowband gamma using auditory gratings, we found that these 524 stimuli produce very strong and focal broadband high-gamma in EEG. Broadband high-525 gamma may also prove to be a useful biomarker for diagnosing disorders (Bragin et al., 2010) 526 and a valuable tool for building brain-computer interfaces (Bouchard and Chang, 2014). In 527 addition, differences in the responses between visual and auditory cortices to stimuli that 528 share many similarities may help understand potential differences in their neural circuitry. 529 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint

530 Adjamian P, Holliday IE, Barnes GR, Hillebrand A, Hadjipapas A, Singh KD (2004) Induced 531 visual illusions and gamma oscillations in human primary visual cortex. Eur J 532 Neurosci 20:587–592. 533 Adrian ED (1942) Olfactory reactions in the brain of the hedgehog. J Physiol 100:459–473. 534 Ainsworth M, Lee S, Cunningham MO, Roopun AK, Traub RD, Kopell NJ, Whittington MA 535 (2011) Dual Gamma Rhythm Generators Control Interlaminar Synchrony in Auditory 536 Cortex. J Neurosci 31:17040–17051. 537 Ainsworth M, Lee S, Kaiser M, Simonotto J, Kopell NJ, Whittington MA (2016) GABAB 538 receptor-mediated, layer-specific synaptic plasticity reorganizes gamma-frequency 539 neocortical response to stimulation. Proc Natl Acad Sci 113:E2721–E2729. 540 Billig AJ, Herrmann B, Rhone AE, Gander PE, Nourski KV, Snoad BF, Kovach CK, 541 Kawasaki H, Howard MA, Johnsrude IS (2019) A Sound-Sensitive Source of Alpha 542 Oscillations in Human Non-Primary Auditory Cortex. J Neurosci Off J Soc Neurosci 543 39:8679–8689. 544 Bokil H, Andrews P, Kulkarni JE, Mehta S, Mitra PP (2010) Chronux: A platform for 545 analyzing neural signals. J Neurosci Methods 192:146–151. 546 Bouchard KE, Chang EF (2014) Neural decoding of spoken vowels from human sensory-547 motor cortex with high-density electrocorticography. Annu Int Conf IEEE Eng Med 548 Biol Soc IEEE Eng Med Biol Soc Annu Int Conf 2014:6782–6785. 549 Bragin A, Engel J, Staba RJ (2010) High-frequency oscillations in epileptic brain. Curr Opin 550 Neurol 23:151–156. 551 Brosch M, Budinger E, Scheich H (2002) Stimulus-Related Gamma Oscillations in Primate 552 Auditory Cortex. J Neurophysiol 87:2715–2725. 553 Brugge JF, Nourski KV, Oya H, Reale RA, Kawasaki H, Steinschneider M, Howard MA 554 (2009) Coding of Repetitive Transients by Auditory Cortex on Heschl’s Gyrus. J 555 Neurophysiol 102:2358–2374. 556 Buzsáki G, Wang X-J (2012) Mechanisms of Gamma Oscillations. Annu Rev Neurosci 557 35:203–225. 558 Canolty R, Soltani M, Dalal S, Edwards E, Dronkers N, Nagarajan S, Kirsch H, Barbaro N, 559 Knight R (2007) Spatiotemporal dynamics of word processing in the human brain. 560 Front Neurosci 1 Available at: 561 https://www.frontiersin.org/articles/10.3389/neuro.01.1.1.014.2007 [Accessed 562 December 6, 2023]. 563 Cardin JA, Carlén M, Meletis K, Knoblich U, Zhang F, Deisseroth K, Tsai L-H, Moore CI 564 (2009) Driving fast-spiking cells induces gamma rhythm and controls sensory 565 responses. Nature 459:663–667. 566 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint Cardin JA, Palmer LA, Contreras D (2007) Stimulus Feature Selectivity in Excitatory and 567 Inhibitory Neurons in Primary Visual Cortex. J Neurosci 27:10333–10344. 568 Chen G, Zhang Y, Li X, Zhao X, Ye Q, Lin Y, Tao HW, Rasch MJ, Zhang X (2017) Distinct 569 Inhibitory Circuits Orchestrate Cortical beta and gamma Band Oscillations. Neuron 570 96:1403-1418.e6. 571 Crone NE, Hao L, Hart J, Boatman D, Lesser RP, Irizarry R, Gordon B (2001) 572 Electrocorticographic gamma activity during word production in spoken and sign 573 language. Neurology 57:2045–2053. 574 Crone NE, Korzeniewska A, Franaszczuk P (2011) Cortical gamma responses: searching 575 high and low. Int J Psychophysiol Off J Int Organ Psychophysiol 79:9–15. 576 deCharms RC, Blake DT, Merzenich MM (1998) Optimizing Sound Features for Cortical 577 Neurons. Science 280:1439–1444. 578 Delorme A, Makeig S (2004) EEGLAB: an open source toolbox for analysis of single-trial 579 EEG dynamics including independent component analysis. J Neurosci Methods 580 134:9–21. 581 Edwards E, Soltani M, Deouell LY, Berger MS, Knight RT (2005) High Gamma Activity in 582 Response to Deviant Auditory Stimuli Recorded Directly From Human Cortex. J 583 Neurophysiol 94:4269–4280. 584 Edwards E, Soltani M, Kim W, Dalal SS, Nagarajan SS, Berger MS, Knight RT (2009) 585 Comparison of Time–Frequency Responses and the Event-Related Potential to 586 Auditory Speech Stimuli in Human Cortex. J Neurophysiol 102:377–386. 587 Elhilali M, Fritz JB, Klein DJ, Simon JZ, Shamma SA (2004) Dynamics of precise spike 588 timing in primary auditory cortex. J Neurosci Off J Soc Neurosci 24:1159–1172. 589 Fujioka T, Trainor LJ, Large EW, Ross B (2009) Beta and Gamma Rhythms in Human 590 Auditory Cortex during Musical Beat Processing. Ann N Y Acad Sci 1169:89–92. 591 Gieselmann MA, Thiele A (2008) Comparison of spatial integration and surround 592 suppression characteristics in spiking activity and the local field potential in macaque 593 V1. Eur J Neurosci 28:447–459. 594 Gray CM, Singer W (1989) Stimulus-specific neuronal oscillations in orientation columns of 595 cat visual cortex. Proc Natl Acad Sci 86:1698–1702. 596 Griffiths TD, Kumar S, Sedley W, Nourski KV, Kawasaki H, Oya H, Patterson RD, Brugge 597 JF, Howard MA (2010) Direct recordings of pitch responses from human auditory 598 cortex. Curr Biol CB 20:1128–1132. 599 Guyon N, Zacharias LR, Fermino de Oliveira E, Kim H, Leite JP, Lopes-Aguiar C, Carlén M 600 (2021) Network Asynchrony Underlying Increased Broadband Gamma Power. J 601 Neurosci 41:2944–2963. 602 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint Jeschke M, Lenz D, Budinger E, Herrmann CS, Ohl FW (2008) Gamma oscillations in gerbil 603 auditory cortex during a target-discrimination task reflect matches with short-term 604 memory. Brain Res 1220:70–80. 605 Jia X, Smith MA, Kohn A (2011) Stimulus Selectivity and Spatial Coherence of Gamma 606 Components of the Local Field Potential. J Neurosci 31:9390–9403. 607 Kato HK, Asinof SK, Isaacson JS (2017) Network-Level Control of Frequency Tuning in 608 Auditory Cortex. Neuron 95:412-423.e4. 609 Koelewijn L, Dumont JR, Muthukumaraswamy SD, Rich AN, Singh KD (2011) Induced and 610 evoked neural correlates of orientation selectivity in human visual cortex. 611 NeuroImage 54:2983–2993. 612 Kowalski N, Depireux DA, Shamma SA (1996) Analysis of dynamic spectra in ferret 613 primary auditory cortex. I. Characteristics of single-unit responses to moving ripple 614 spectra. J Neurophysiol 76:3503–3523. 615 Langers DRM, Backes WH, van Dijk P (2003) Spectrotemporal features of the auditory 616 cortex: the activation in response to dynamic ripples. NeuroImage 20:265–275. 617 Lenz D, Jeschke M, Schadow J, Naue N, Ohl FW, Herrmann CS (2008) Human EEG very 618 high frequency oscillations reflect the number of matches with a template in auditory 619 short-term memory. Brain Res 1220:81–92. 620 Lewicki MS (2002) Efficient coding of natural sounds. Nat Neurosci 5:356–363. 621 Manning JR, Jacobs J, Fried I, Kahana MJ (2009) Broadband Shifts in Local Field Potential 622 Power Spectra Are Correlated with Single-Neuron Spiking in Humans. J Neurosci 623 29:13613–13620. 624 Mazaheri A, Picton TW (2005) EEG spectral dynamics during discrimination of auditory and 625 visual targets. Cogn Brain Res 24:81–96. 626 Moore AK, Wehr M (2013) Parvalbumin-Expressing Inhibitory Interneurons in Auditory 627 Cortex Are Well-Tuned for Frequency. J Neurosci 33:13713–13723. 628 Murty DV, Manikandan K, Kumar WS, Ramesh RG, Purokayastha S, Nagendra B, ML A, 629 Balakrishnan A, Javali M, Rao NP, Ray S (2021) Stimulus-induced gamma rhythms 630 are weaker in human elderly with mild cognitive impairment and Alzheimer’s disease. 631 eLife 10:e61666. 632 Murty DVPS, Manikandan K, Kumar WS, Ramesh RG, Purokayastha S, Javali M, Rao NP, 633 Ray S (2020) Gamma oscillations weaken with age in healthy elderly in human EEG. 634 NeuroImage 215:116826. 635 Murty DVPS, Ray S (2022) Stimulus-induced Robust Narrow-band Gamma Oscillations in 636 Human EEG Using Cartesian Gratings. Bio-Protoc 12:e4379. 637 Murty DVPS, Shirhatti V, Ravishankar P, Ray S (2018) Large Visual Stimuli Induce Two 638 Distinct Gamma Oscillations in Primate Visual Cortex. J Neurosci 38:2730–2744. 639 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint Muthukumaraswamy SD, Singh KD (2013) Visual gamma oscillations: The effects of 640 stimulus type, visual field coverage and stimulus motion on MEG and EEG 641 recordings. NeuroImage 69:223–230. 642 Nunez PL, Srinivasan R (2006) Electric Fields of the Brain: The neurophysics of EEG. 643 Oxford University Press. Available at: 644 https://doi.org/10.1093/acprof:oso/9780195050387.001.0001 [Accessed December 6, 645 2023]. 646 Olshausen BA, Field DJ (1996) Emergence of simple-cell receptive field properties by 647 learning a sparse code for natural images. Nature 381:607–609. 648 Ray S, Crone NE, Niebur E, Franaszczuk PJ, Hsiao SS (2008) Neural Correlates of High-649 Gamma Oscillations (60–200 Hz) in Macaque Local Field Potentials and Their 650 Potential Implications in Electrocorticography. J Neurosci 28:11526–11536. 651 Ray S, Maunsell JHR (2010) Differences in Gamma Frequencies across Visual Cortex 652 Restrict Their Possible Use in Computation. Neuron 67:885–896. 653 Ray S, Maunsell JHR (2011) Different Origins of Gamma Rhythm and High-Gamma 654 Activity in Macaque Visual Cortex. PLOS Biol 9:e1000610. 655 Ross B, Tremblay KL, Alain C (2020) Simultaneous EEG and MEG recordings reveal vocal 656 pitch elicited cortical gamma oscillations in young and older adults. NeuroImage 657 204:116253. 658 Schönwiesner M, Zatorre RJ (2009) Spectro-temporal modulation transfer function of single 659 voxels in the human auditory cortex measured with high-resolution fMRI. Proc Natl 660 Acad Sci U S A 106:14611–14616. 661 Sedley W, Teki S, Kumar S, Overath T, Barnes GR, Griffiths TD (2012) Gamma band pitch 662 responses in human auditory cortex measured with magnetoencephalography. 663 NeuroImage 59:1904–1911. 664 Shamma S (2001) On the role of space and time in auditory processing. Trends Cogn Sci 665 5:340–348. 666 Shirhatti V, Ray S (2018) Long-wavelength (reddish) hues induce unusually large gamma 667 oscillations in the primate primary visual cortex. Proc Natl Acad Sci 115:4489–4494. 668 Steinschneider M, Fishman YI, Arezzo JC (20 08) Spectrotemporal Analysis of Evoked and 669 Induced Electroencephalographic Responses in Primary Auditory Cortex (A1) of the 670 Awake Monkey. Cereb Cortex 18:610–625. 671 Tallon-Baudry C, Kreiter A, Bertrand O (1999) Sustained and transient oscillatory responses 672 in the gamma and beta bands in a visual short-term memory task in humans. Vis 673 Neurosci 16:449–459. 674 Tian B, Rauschecker JP (2004) Processing of Frequency-Modulated Sounds in the Lateral 675 Auditory Belt Cortex of the Rhesus Monkey. J Neurophysiol 92:2993–3013. 676 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint Trautner P, Rosburg T, Dietl T, Fell J, Korzyukov OA, Kurthen M, Schaller C, Elger CE, 677 Boutros NN (2006) Sensory gating of auditory evoked and induced gamma band 678 activity in intracranial recordings. NeuroImage 32:790–798. 679 Vianney-Rodrigues P, Iancu OD, Welsh JP (2011) Gamma oscillations in the auditory cortex 680 of awake rats. Eur J Neurosci 33:119–129. 681 Vinck M, Batista-Brito R, Knoblich U, Cardin JA (2015) Arousal and Locomotion Make 682 Distinct Contributions to Cortical Activity Patterns and Visual Encoding. Neuron 683 86:740–754. 684 Wang X, Lu T, Snider RK, Liang L (2005) Sustained firing in auditory cortex evoked by 685 preferred stimuli. Nature 435:341–346. 686 Whittington MA, Traub RD, Jefferys JG (1995) Synchronized oscillations in interneuron 687 networks driven by metabotropic glutamate receptor activation. Nature 373:612–615. 688 Wilkins A, Nimmo-Smith I, Tait A, McManus C, Sala SD, Tilley A, Arnold K, Barrie M, 689 Scott S (1984) A neurological basis for visual discomfort. Brain 107:989–1017. 690 691 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint Figure Legends 692 Figure 1: Change in power after stimulus presentation pooled across all stimulus 693 conditions. 694 Subject-averaged time-frequency change in power spectra up to 150Hz for all 64 channels ( 695 frequency: 0-150Hz, vertical axis and time: -500-1300ms, horizontal axis). The plots are 696 arranged according to the 64-channel layout (actiCap) with FCz as the reference electrode, 697 unipolar referencing. Stimulus is presented during 0 to 800ms, indicated by dashed vertical 698 lines. Colorbar indicates log Power ratio in decibels (dB). (A) In response to the presentation 699 of the visual grating stimuli. (B) In response to the presentation of the auditory grating 700 stimuli. The star-marked electrodes in the topoplots at the bottom show the electrodes used 701 for further analysis. 702 Figure 2: Subject and electrode averaged change in power compared to baseline (-703 500ms to 0ms), pooled across all stimulus conditions with two different reference 704 schemes. 705

are shown for the unipolar reference scheme (left) and bipolar (right). The top row in 706 A and the first two in B and C are in response to visual stimuli. The bottom row in A and the 707 last two in B and C are in response to the auditory stimuli. Electrodes used for averaging in A 708 and B are highlighted with black dots in scalp maps shown in C. (A) Time-frequency change 709 in power. Dashed vertical lines (black) represent stimulus onset and offset. (B) Power spectra 710 (first and third-row panels – red trace is stimulus period (250ms to 750ms), and the green 711 trace is the baseline period (-500ms to 0ms)) and change in power spectra vs frequency 712 (second and fourth-row panels, blue traces). The solid traces represent the mean across 713 subjects, and thickness represents SEM. Dashed vertical lines represent narrowband gamma 714 (20-66Hz, blue) and broadband high-gamma (70-150Hz, cyan). Coloured squares at the 715 bottom represent the significance of differences in means (paired t-test – purple: p-values 716 between 0.01-0.05, pink: p <0.01). (C) Scalp maps of 64 unipolar electrodes (left) and 114 717 bipolar electrodes (right). The first and third rows show the change in power for frequencies 718 20-66Hz. The second and fourth rows show the change in power for frequencies 70-150Hz. 719 The colorbar represents the log power ratio in dB. 720 Figure 3: Visual and auditory responses for individual subjects. 721 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint Change in time-frequency power spectra for males (A) and females (B). The top row 722 represents the results in response to visual stimuli, and the bottom represents the results in 723 response to auditory stimuli. Subjects are ordered horizontally based on the decreasing high-724 gamma activity in 70-150Hz for the auditory condition, starting from the left. The response is 725 averaged across selected occipital and temporal electrodes (highlighted in the inset of Fig 1) 726 for respective protocols. The colorbar represents the log power ratio in dB. Empty plots refer 727 to subjects rejected from the analysis (refer to methods – artefact rejection for more details). 728 Figure 4: Auditory-induced high-gamma activity is not correlated with visually induced 729 narrowband gamma. 730 Scatter plots for change in power in response to visual stimuli vs auditory stimuli. (A) For 731 narrowband gamma power averaged across chosen electrode groups. (B) For broadband high-732 gamma power averaged across chosen electrode groups. (C) For narrowband gamma power 733 for electrode with maximum change in power. (D) For broadband high-gamma power for 734 electrode with maximum change in power. Pearson’s correlation coefficient and p-value are 735 mentioned at the bottom right in each panel. Note the difference in axis limits across the 736 panels. (E) Bar plots showing the mean change in power from baseline across subjects for 737 different gamma frequency ranges (NB refers to Narrowband gamma and BB refers to 738 Broadband high-gamma) during different tasks. The dots represent the change in the power of 739 each subject. The significance of the unpaired t-test (with unequal variance) is indicated at 740 the top of the bar plots. 741 Figure 5: No stimulus selectivity was observed for either of the induced gamma(s). 742 Change in time-frequency power spectrum from baseline for different stimulus conditions. 743 The last row and column represent averaged responses across that row or column. (A) In 744 visual modality. (B) In auditory modality. Power is averaged across chosen electrodes for 745 each modality. Dashed vertical lines (black) represent stimulus onset and offset. The colorbar 746 represents the log power ratio in dB. 747 Figure 6: Subject-averaged time-frequency change in power spectra for the (A) eye-open 748 and (B) eye-close auditory tasks—same format as Figure 1. 749 750 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted January 6, 2024. ; https://doi.org/10.1101/2024.01.05.574448doi: bioRxiv preprint