Neural dynamics of induced vocal tract vibrations during vocal emotion recognition

26 Emotional prosody is defined as suprasegmental and segmental changes in voice and 27 related acoustic parameters that can inform the listener about the emotional state of the speaker. 28 Despite a large corpus of literature in psychological and brain mechanisms in emotional 29 prosody perception, the perspective of embodied cognition in these mechanisms have been 30 largely neglected . Here we investigated the influence of induced bodily vibrations in the 31 categorization of ambiguous emotional vocalizations in an event-related potential study (N=24). 32 The factorial design included Vocal emotion [anger and fear] and external Vibration [anger, 33 fear, and none] as fixed factors. Emotional voices were morphed between a fearful expression 34 with the speaker’s identity -matching ang ry expression, creating blends of emotions in each 35 voice. Emotional congruent and incongruent vibrations were delivered on the skin through 36 transducers placed close to the vocal cords. We hypothesized that induced bodily vibrations 37 would constitute a n interoceptive and proprioceptive feedbacks that would influence the 38 perception of emotions, especially for more ambiguous voices as ambiguity would favour the 39 processing of other available sensory information, here toward the tactile sensory modality. 40 Behavioural results revealed that induced vibrations skewed the participants’ emotional ratings 41 by biasing responses congruent with the vibration. Event-related potentials results indicated 42 that N100 and P200 components subtending the early processing of emotional prosody were 43 significantly modulated by induced vibrations in the congruent setting , which could be 44 considered as a facilitation effect for emotion recognition at early stage of processing. A 45 significant modulation of the late positive component was also observed in the incongruent 46 setting, suggesting an error processing mechanism. EEG source reconstruction highlighted 47 significant contrasts between vibration types in prefrontal, motor, somatosensory , and insular 48 cortices. Altogether, our results suggest that voice-associated vibrations would play a 49 significant role in vocal emotion perception and recognition through embodied mechanisms at 50 both behavioral and neural levels. 51 Key words: electroencephalography, emotional prosody, bodily vibrat ions, interoception, 52 embodiment 53 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint

54 Oral communication is essential for a social animal such as a human being, both in terms 55 of semantic meaning of words and non-verbal parameters. The ‘emotional prosody’ of the voice 56 refers to the changes in these acoustic parameters, and it also refers to the melody of the voice, 57 which is related, among others, to the temporal fluctuations of the pitch, envelope amplitude, 58 as well as the spectral fluctuations related to the timbre (Grandjean et al., 2006; Laukka et al., 59 2005; Risberg & Lubker, 1978; Scherer & Oshinsky, 1977). According to Scherer's adaptation 60 of the Brunswik lens model, perception of emotional prosody—also named ‘vocal emotion’—61 variations allow the listener to make subjective attribution(s) of the speaker's emotional state(s) 62 in social interactions, often influenced by context (Brunswik, 1956; Scherer, 2003, Grandjean 63 et al., 2006). 64 One of the key mechanisms allowing this accurate attribution of a speaker's emotional 65 state based on voice signals relies on the embodied ‘simulation’ of what we hear (Hawk et al., 66 2012). Emotional embodiment is defined as the motor, sensory and perceptual reexperiencing 67 or simulating of the emotion perceived in a peer by the perceiver in the framework of the 68 Simulation of Smiles model (SIMS model; Niedenthal, 2007, Niedenthal et al., 2010, p.418). 69 This process has primarily been described in the visual modality, as authors have demonstrated 70 that the perception of facial expression s in others generates facial mimicry in the observer, 71 which is stronger for emotional compared to non-emotional facial expressions (Dimberg, 1982; 72 Korb et al., 2010; Künecke et al., 2017; Moody et al., 2007; Moody & McIntosh, 2011; 73 Chartrand & Bargh, 1999). Recently, many studies have supported the idea that facial mimicry 74 represents the sensorimotor simulation of an observed emotion rather than a mere muscular 75 reproduction of an observed facial expression, and is context - and affiliation -dependent 76 (Borgomaneri et al., 2020; Hess & Fis cher, 2014; Drimalla et al., 2019; Hess, 2021; Philip et 77 al., 2018). Importantly, some studies suggest that this process may play a role in the accurate 78 decoding of facial emotional information especially in women (Hyniewska & Sato, 2015; Korb 79 et al., 2010 ; Schneider et al., 2013; Stel & van Knippenberg, 2008), although this claim is 80 contradicted by other studies (Blairy et al., 1999; Hess & Blairy, 2001; Holland et al., 2021). 81 While studies o n the contribution of embodiment in the field of emotional prosod y 82 decoding are scarcer than in the visual domain, some point to a contribution of this mechanism 83 in the auditory modality. Indeed, several studies showed that passive listening of speech signals 84 elicits activations in motor regions of the brain similar to those involved during speech 85 production (Fadiga et al., 2002; Watkins et al., 2003; Wilson et al., 2004). Another study 86 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint showed that language comprehension was facilitated when the primary motor areas controlling 87 the lips or the tongue were stimulated trans cranially, and this was true for sounds respectively 88 produced by either of these oral organs (D'Ausilio et al., 2009). One of the mechanisms which 89 may underly the effect of emotional embodiment in the emotional prosody domain relies on 90 body resonances. In fact, such resonances—vibrations, especially in the throat region close to 91 the vocal folds—may represent a series of interoceptive and proprioceptive feedbacks during 92 vocal emotion production. Indeed, during voice production, vibrations from the vocal cords are 93 transmitted through the speaker's skin and body tissues (Munger & Thompson, 2008; Švec et 94 al, 2005). The frequency of these body resonances mainly corresponds to the fundamental 95 frequency of the auditory tone that is produced (Sundberg, 1992). T he sensation of body 96 resonances would constitute for the speaker a stable and dynamic feedback on several 97 characteristics of his or her own voice (Sundberg, 1992), and these can hardly be disturbed by 98 the environment. Lylykangas and colleagues (2009) highl ighted the potential to successfully 99 use vibrotactile feedback as relevant information for speed regulation messages. Similarly, 100 Tuuri and colleagues (2010) compared audio and tactile feedback from the same speech stimuli 101 as speed regulation cues and found that the information was delivered in a similar manner across 102 modalities. However, there is still a paucity of research in this field and further research is 103 needed to gain a deeper and clearer understanding of the different types of mechanisms that 104 may be engaged in the embodi ment of emotional prosody, both at the behavioral and brain 105 levels. To investigate precise dynamic neural correlates of the embodiment of vocal emotion 106 perception—a novel topic in the literature, we turned to surface electroencephalography (EEG). 107 We will now start by introducing research on vocal emotion perception and show how this work 108 constitutes a solid ground to build up and further investigate the dynamic mechanisms of the 109 embodiment of emotion in the vocal domain. 110 Event-related potential (ERP) correlates of emotional prosody decoding have been 111 extensively studied in the last two decades. Models of emotional prosody decoding suggest a 112 processing in three -stage (Schirmer & Kotz, 2006). First, sensory processing in the auditory 113 cortex and superior temporal sulcus (STS, 0 -150ms) takes place with low -level acoustic 114 analysis. The second stage (150 -250ms) involves the integration of emotionally significant 115 acoustic cues in the bilateral superior temporal gyrus (STG) and anterior STS and the third stage 116 (>250ms) consists of emotional processing with high-level cognitive processes in frontal areas. 117 In line with this model, a wealth of ERP studies has shown the implication of several early and 118 late ERP components in the processing of emotio nal vocalizations. Early components appear 119 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint to be involved in auditory-sensory processing, attentional processing and early recognition of 120 meaning. Firstly, the N100 component is a negative deflection peaking around 100ms after 121 stimulus onset and located most of the time in the frontocentral area. It might reflect an early 122 component of low-level prosody analysis as well as emotion discrimination ( Gädeke et al., 123 2013; Paquette et al., 2020; Tarai & Srinivasan, 2019). In other studies, basic emotions are also 124 distinguished in the P200 component located mostly in the frontal, frontocentral and parietal 125 areas (Duville et al., 2022; Maltezou-Papastylianou et al., 2022; Paulmann & Uskuul, 2017; 126 Paulmann et al., 2013; Gädeke et al., 2013). The P200 component is a positive deflection around 127 200 milliseconds that is believed to reflect an increased attention to emotional stimuli so that 128 they can be processed more effectively when needed (Paulmann & Kotz, 2008; Thomas et al., 129 2007). Later ERP components are commonly ass ociated with higher -level mechanisms of 130 emotional prosody processing such as appraisal, contextual relevance and mental 131 representations (Duville et al., 2022; Pell & Kotz, 2021). Indeed, negative or positive 132 deflections around 300 and 400 milliseconds (P30 0/N400) located in frontal and frontocentral 133 areas were observed during emotional prosody processing (Carminati et al., 2018; Proverbio et 134 al., 2020; Tarai & Srinivasan, 2019) and were associated with the detection of deviant stimuli 135 in an auditory oddball paradigm (Lindı́n et al., 2004; Tsolaki et al., 2015). Finally, the Late 136 Positive Component (LPC) is a positive deflection observed at a later stage in frontal, 137 frontocentral and parietal areas that is associated with higher order cognitive processes such as 138 arousal information (Paulmann et al., 2013), expectancy violation in prosody (Astésano et al., 139 2004; Brattico et al., 2006; Paulmann et al., 2012) and emotional ‘meaning’ (Wei et al., 2022; 140 Zora et al., 2020). A study investigating the processing of emotional facial expression reported 141 that the LPC may reflect attentional processes rather than emotional processing (Ashley et al., 142 2004), which is consistent with the attentional capture effect assoc iated with violation or 143 incongruence designs (Vachon et al., 2012). Other studies showed that patterns of mismatching 144 prosodies or musical stimuli elicited an LPC that could indicate an integration process of the 145 different sources of information (Brattico et al., 2006; Paulmann et al., 2012). 146 Functional MRI studies have demonstrated that emotional prosody perception activates 147 both subcortical and cortical networks (Grandjean, 2012; Witteman et al., 2012). These 148 networks include the bilateral temporal voice areas, inferior frontal gyrus, orbitofrontal area, 149 cerebellum, and amygdala, among others. Additionally, studies have shown the involvement of 150 the motor cortex during speech perception (Fadiga et al., 2002; Watkins et al., 2003). 151 Furthermore, studies exami ning interoceptive signals have reported activations in the insular 152 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint cortex, a region responsible for the integration both internal and external physiological signals 153 and emotional processes (Adolfi et al., 2017; Casals -Gutierrez & Abbey, 2020). Other 154 structures, such as the somatosensory cortex, the supplementary motor area (SMA), the 155 prefrontal cortex and the cingulate cortex, have been found to play a role in interoceptive and 156 proprioceptive processing (Engelen et al., 2023; Gibson, 2019; Khalsa et al., 2018). 157 Our study aims to explore the potential interoceptive and proprioceptive feedbacks 158 represented by induced bodily resonances —through induced vibrations , and their impact on 159 emotional prosody perception. Revealing such as link between vocal emotion pe rception and 160 induced body resonances would constitute convincing evidence for a crucial role of vibro-tactile 161 feedback as a physiological mechanism of vocal emotion embodiment. An association between 162 interoception and emotional processing was shown in nume rous studies but never when using 163 the vibrations concurrent to expressed voice signals (Couto et al., 2015; Critchley, & Garfinkel, 164 2017; Burleson & Quigley et al., 2021). For example, Mendoza -Medialdea & Ruiz -Padial 165 (2021) revealed the role of interocepti ve accuracy on exogenous attention to disgusting and 166 fearful distractors of a competing task. Several ERP studies investigating interoception in 167 emotion processing showed that there is a link between the interoceptive sensitivity as measured 168 with a heartbe at detection task and the cortical processing of emotional stimuli. Indeed, 169 significant differences in the P300 component and in a slow wave window (550 -900 ms) have 170 been reported all over the scalp between good and poor heartbeat perceivers while rating 171 emotional pictures (Herbert et al., 2007; Pollatos et al., 2005). A somatosensory-specific P300 172 component was also revealed in parietal regions in studies investigating somatosensory 173 perception (Al et al., 2020; Auksztulewicz, & Blankenburg, 2013; Has et al., 2021). Moreover, 174 Salamone et al. (2021), showed that negative facial emotion recognition was improved after 175 focusing on one’s own heartbeat and that heart-evoked potentials around 250-300 milliseconds 176 were modulated by this interoceptive priming task. To our knowledge, no study to date has 177 specifically examined the role of vibrations as interoceptive and proprioceptive feedback 178 signals in vocal emotion perception using EEG recordings. While our previous behavioral study 179 explored a related question, its design differed by focusing on perceptual rather than emotional 180 ambiguity, and its findings did not provide conclusive evidence. This highlights the need for 181 further investigations into the potential impact of bodily resonances on auditory emotion 182 perception. 183 Nevertheless, multimodal integration/multisensorial studies using audiovisual 184 emotional stimuli have been conducted in the past and they showed that multimodal information 185 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint seems to facilitate emotional processing compared to unimodal information ( Föcker & Röder, 186 2019; Kotz, Jessen & Paulmann, 2009) and that multimodal signals interact at several stages of 187 the audiovisual integration process ( Klasen et al., 2012). Moreover, it appears that ERP 188 components usually involved in emotional processing are modulated by the interaction of 189 auditory and visual information (Klasen et al., 2012). Indeed, Föcker & Röder (2019) showed 190 several interesting results: i) performance in an emotion recognition task was higher in 191 congruent bimodal trials compared to unimod al ones; ii) the earliest multisensory interactions 192 they reported were represented by a greater positivity for congruent face -voice stimuli in the 193 P200 time range (180-230ms); iii) the N400 component was more negative for incongruent than 194 for congruent emotional expressions in face-voice pairs. Furthermore, Pourtois and colleagues 195 (2000) showed in an audiovisual task of emotion recognition that the audiovisual integration 196 was probably taking place at a very early stage (110ms post -stimulus), with the amplitude of 197 the auditory N1 being enhanced for bimodal stimuli. In addition, behavioral (Brosch, Grandjean 198 et al., 2008 ) and EEG ( Brosch, Sander et al., 2008 ) studies further support the idea that 199 emotionally relevant stimuli can modulate attention and early pe rceptual processing across 200 modalities. Although the current study involves different modalities as compared to existing 201 literature (vibro-tactile ‘interoception’ and audition), we rely on these promising results to 202 investigate the impact of interoceptive and proprioceptive information on emotional auditory 203 processing. 204 To the best of our knowledge, no study has specifically investigated the contribution of 205 body vibrations to the auditory perception of emotions using EEG. While our previous 206 behavioral study addressed a related question, its design differed, and its findings were 207 inconclusive. The present study aims to further explore this mechanism. The main hypothesis 208 is that vocal cord vibrations, which are a ‘mechanical’ consequence of the production of 209 vocalized sounds, would represent interoceptive and proprioceptive feedbacks that could 210 impact the auditory perception of emotions from an embodied perception perspective: as a valid 211 feedback, vibratory signals would influence voice production for the speak er and should 212 therefore also have an impact on perception if induced in the listener. Since the SIMS model 213 specifies that the effect of embodied simulation is observed when perceiving non -prototypical 214 emotional stimuli, our study uses ambiguous emotional v ocal stimuli in order to favor the 215 triggering of such effect. Our study targets ‘emotional ambiguity’, with stimuli simultaneously 216 conveying two emotions —i.e., anger and fear —at varying percentages using an auditory 217 morphing procedure (Latinus & Belin, 2011). The vibrations induced in two-thirds of the trials 218 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint either expressed pure anger or pure fear, and were either fully congruent, partially incongruent, 219 or fully incongruent with the auditorily-presented emotional vocalizations (see Methods section 220 for details). At the behavioral level, two key hypotheses were tested for this study. First, we 221 postulated that vibrations would affect the discriminative perceptual processes of emotional 222 vocalizations. Specifi cally, it was predicted that participants would respond "anger" more 223 frequently when the vibration also expressed anger and likewise for fear trials—when compared 224 to the ‘no-vibration’ condition. Second, these effects were expected to be observed especially 225 for ambiguous stimuli (i.e., voices containing 40 -60% of each emotion) and not for 226 unambiguous stimuli (i.e., voices containing 90% of each emotion). Indeed, the more 227 emotionally ambiguous stimuli would transmit unclear input and would therefore be the 228 conditions in which information is sought elsewhere, based on the SIMS model (Niedenthal, 229 2007, Niedenthal et al., p.418)—e.g., via embodied cognition here represented and/or triggered 230 by induced vibrations. At the neural level, we first expected amplitude changes for ERP 231 components associated with emotional prosody processing—such as the N100, P200 and LPC 232 in fronto-temporal electrodes (Paquette et al., 2020; Paulmann & Kotz; Paulmann, & Uskul, 233 2017, 2013; Pourtois et al., 2004; Wei et al., 2022; Zora et al., 2020). Second, we postulated 234 the presence of vibration -specific ERPs between 200 and 300ms modulated by interoceptive 235 and proprioceptive processes (Herbert et al., 2007; Mendoza-Medialdea & Ruiz-Padial, 2021; 236 Pollatos et al., 2005) with source mod ulations in the insular, somatosensory, prefrontal, and 237 cingulate cortices as well as in the SMA or pre-SMA (Adolfi et al., 2017; Casals -Gutierrez & 238 Abbey, 2020 ; Engelen et al., 2023; Gibson, 2019; Khalsa et al., 2018 ), for anger and fear 239 vibrations compared to the conditions without vibrations. Third, as the design included in some 240 trials an incongruence between emotional vocalizations and vibrations–anger/fear vocalizations 241 mismatched with fear/anger, we expected as well the presence of early (100ms and 200-300ms) 242 and late (400-700ms) ERP components with source modulations in the above-mentioned brain 243 areas, which can be related to to error prediction and /or integration of interoceptive and 244 proprioceptive feedbacks in the context of emotional prosody information processing. 245

246 In this study, the participants had to categorize emotional voices that were morphed with 247 a continuum between anger and fear, at varying percentages. In two-thirds of the trials, bodily 248 vibrations were ind uced on their throat through a vibrating device, and these vibrations 249 expressed either pure anger or pure fear. We analyzed the influence of these vibrations on 250 behavioral responses and ERPs and we further extracted source modulations in the brain. 251 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint Behavioral responses 252 The main effect of Vibration was significant ( χ2(2) = 18.10, p<.001, Figure 1A). The 253 main effect of Emotional Voice was also significant ( χ2(1) = 55.93, p<.001, Supplementary 254 Figure 1) as well as their interaction ( χ2(2) = 9.29, p< .01, Figure 1B). The effect size of this 255 model was medium (R2marginal = .10, R2conditional = .581). 256 Further investigation of the impact of Vibration on the participants’ responses was 257 performed using planned contrasts related to our hypotheses and the results were the following 258 (Fear vibrations: m=.57, sd=.50, Anger vibrations: m=.53, sd=.50, No vibrations: m=.55, 259 sd=.50): i) [Vibration Anger > Fear] (χ2(1) = 7.25, p None] (χ2(1) = 260 4.23, p=12, and iii) [Vibration Anger > None] (χ2(1) = 2.07, p=.45; Figure 1A.). Concerning 261 the impact of the interaction between Vibration and Emotional Voice on the participants’ 262 responses, planned contrasts were also performed: [Vibration Anger > Fear] x [Ambiguous 263 (A60, A50, A40) voice stimuli] ( χ2(1) = 7.25, p None] x [Ambiguous 264 (A60, A 50, A 40) voice stimuli] ( χ2(1) = 7.78, p None] x [Ambiguous 265 (A60, A50, A40) voice stimuli] (χ2(1) = 1.31, p=.76; Figure 1B). More specific contrasts on the 266 most ambiguous condition of Emotional Voice were computed: [Vibration Anger > Fear] x 267 [Ambiguous (A 50) voice stimuli] (χ2(1) = 3.96, p =.14), [Vibration Anger > None] x 268 [Ambiguous (A50) voice stimuli] (χ2(1) = 6.64, p None] x [Ambiguous 269 (A50) voice stimuli] (χ2(1) = 1.307, p= 1; Figure 1B) . All p-values were adjusted using the 270 Bonferroni correction method (Bland & Altman, 1995). 271 1 R²marginal represents the proportion of variance explained by the fixed effects alone, while R²conditional accounts for bo th fixed and random effects, providing a measure of the total variance explained by the model (Nakagawa & Schielzeth, 2013) .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint 272 Figure 1. A. Main effect of Vibration on the probability of fear responses. On the X axis, 273 each point represents a Vibration condition (i.e., None = no vibrations, Anger, Fear). B. 274 Interaction between Emotional Voice and Vibration on the probability of fear responses . On 275 the X axis, each point represents an Emotional Voice condition (i.e., A90, A60, A50, A40, A10) 276 and each coloured curve represents a Vibration condition (i.e., None = no vibrations, Anger, 277 Fear). Error bars indicate the standard error of the mean. *** p<.001. 278 279 EEG results 280 Event-Related Potentials 281 We started by analyzing ERPs to investigate the influence of induced vibrations during 282 vocal emotion processing on brain dynamics. Significant differences between our clusters in 283 terms of response to our conditions is materialized by a significant three -way Cluster X 284 Emotional Voice X Vibration interaction along trial length (Supplementary Table 6, 285 F(24)=[3.04-23.36 min-max]; p<.001 FDR corrected). Following clustering data reduction (see 286 Methods), three clusters contained at least one 50ms timebin presenting an Emotional voice X 287 Vibration two -way interaction. In the paragraph below, we describe the results from the 288 functionally best-defined cluster, i.e., the frontocentral cluster ( cluster C4, Figure 2 G, Figure 289 3K). Moreover, the frontocentral zone is the m ost represented in the literature on emotional 290 prosody (Duville et al., 2022; Paquette et al., 2020), as well as being relevant for somatosensory 291 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint perception (Desmedt & Bourguet, 1985 ; Macerollo et al., 2018 ). Results from the two other 292 ERP clusters (occipital and temporal) are described in the supplementary material 293 (‘Supplementary EEG results ’, I.). The results presented below pertain directly to our 294 hypotheses while exhaustive results are reported in the supplementary material for ERP data 295 (‘Supplementary EEG results ’, II, and Supplementary Figures 2 -4) and for source 296 reconstruction (Supplementary Figures 4-5). 297 Effect of Emotional voice X Vibration: frontocentral cluster C4 298 The frontocentral cluster C4 (Figure 2G & Figure 3K) presented an Emotional Voice X 299 Vibration two -way interaction at the 50 -150ms, 300 -400ms, 450 -550ms, 600 -650ms, 750 -300 850ms, and 900-950ms time intervals (Supplementary Table 6, Figure 2 and Figure 3). Three 301 clearly defined peaks exceeded the baseline threshold (see Methods): a positive deflection from 302 100 to 150ms, a positive deflection from 200 to 350ms, and a negative deflection starting 500ms 303 post-stimulus onset. 304 Low emotional voice ambiguity 305 At an early stage—the 100ms peak, no expected results were observed while other non-planned 306 contrasts revealed significant. These are described and reported in the supplementary material 307 (Figure S2, Supplementary Table 5). At 250ms, anger vibrations—compared to fear vibrations 308 and no-vibration—led to a congruent increase of the P200 peak amplitude for mostly-angry 309 (A90) voices (Figure 2A, Supplementary Table 5). At a later stage—at 750ms, anger vibration 310 was associated in congruent A90 voices to a significantly weaker negative deflection compared 311 to the no-vibration condition (Figure 2A). 312 High emotional voice ambiguity 313 In the high ambiguity setting, fear vibration was associated with an incongruent decrease 314 of P200 peak amplitude for A60/A50 voices, which persists in the descending phase of the P200 315 peak (Figure 3A). Additionally, in the descending phase of the P200 peak, a significant power 316 decrease was also observed for A 60 voices under pure anger vibration induction (Figure 3A). 317 Moreover, at a later stage—around 600ms, we observed a congruent significant increase in 318 power amplitude for A 40 voice for fear vibrations compared to the no -vibration condition 319 (Figure 3 F). Finally, a t 750ms, induced fear vibration was associated with a significantly 320 stronger negative deflection for congruent A40 voices (Figure 3F). Induced anger vibration was 321 associated with a significant relative power decrease compared to no -vibration in A 60 voices 322 only (Figure 3 A). Anger vibration also led to a significantly stronger negative deflection 323 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint compared to no-vibration in A60 and A40 voices (Figure 3A,F). Finally, induced fear vibration 324 was associated to a significantly stronger negative deflection compared to no -vibration when 325 perceiving A40 voices (Figure 3F). 326 Source reconstruction 327 Source reconstruction analysis revealed correlates at the brain level of the vibration 328 ERP-dissociations observed at the sensor level (Figure 2B-D,F, Figure 3B-E,G-J). 329 Low emotional voice ambiguity 330 We observe in the 300-350ms time interval a significant fear vs anger vibrations source-331 ERP contr ast in the superior posterior lobe (Figure 2B,C). At a later stage , we observe d a 332 significant incongruent fear vs no -vibrations source-ERP contrast in the medial orbitofrontal 333 cortex (OFC), insula and primary somatosensory cortex (550 -600ms, Figure S4) followed by 334 the supplementary motor area (750-800ms) and primary motor cortex (900-950ms) (see Figure 335 2D). For fearful voices, significant congruent fear vs no-vibrations contrasts were observed in 336 the supplementary motor area (700 -750ms) and superior pa rietal lobe (700 -750ms; 850 -337 900ms). See Figure 2F. 338 339 Figure 2. Contrasts between vibrations for Low Ambiguity vocalizations. Top panel: 340 Contrasts between vibrations for low ambiguity angry voices (A 90) for A: Event-related 341 potentials and B-C-D: Brain activations from source reconstruction. Bottom panel: Contrasts 342 between vibrations for low ambiguity fearful voices (F90) for E: Event-related potentials and F: 343 Brain activations from source reconstruction. G: Graphical representation of the topographical 344 location of the analyzed cluster. For ERPs graphs, line colors for each condition represent the 345 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint vibration applied (anger in red, fear in blue, no -vibration control in black). Upper horizontal 346 lines represent contrast with substantial proof of difference (BF>3, see Methods) between 347 conditions. Black lower horizontal line represents the period where the two -way Condition X 348 Vibration interaction was significant. Blue dashed horizontal line represents the 3 X base line 349 standard deviation interval . All statistics reported here have Bayes factors > 3. M1 : primary 350 motor cortex, SMA: supplementary motor area, SPL: superior parietal lobule. 351 352 High emotional voice ambiguity 353 At the 350 -400ms peak, significant source -ERP fea r vs no vibration contrasts were 354 observed in the anterior cingulate cortex and anterior insula for fearful voices (Figure 3G-H). 355 A significant anger vs fear vibration contrast was also observed at the same timing in the 356 superior parietal cortex for fearful voices (Figure 3G). At the 450-500ms peak, a significant 357 congruent anger vibration vs no-vibration contrast was observed for angry voices in the primary 358 sensory, primary motor, and premotor areas, orbitofrontal cortex, dorsolateral prefrontal cortex 359 (DLPFC) as well as in the intraparietal sulcus (IPS; Figure 3B). In the late period, a significant 360 congruent ang er vibration vs no -vibration contrast was observed for angry voices in the 361 precuneus (700-750ms, Figure 3D), as well as superior parietal lobule and IPS (900-950ms, 362 Figure 3E). A significant congruent fear vibration vs no-vibration contrast was also observed 363 for fearful voices in the primary motor cortex, IFG pars opercularis and OFC from 650 to 364 700ms while a significant fear and anger (global concatenation of emotion) vibration vs no-365 vibration contrast reached significance in the pr imary motor area at the same timing (Figure 366 3I). Finally, at 750 -800ms, an incongruent anger vibration vs no -vibration contrast was 367 observed for fearful voices in primary motor cortex and SMA (Figure 3J). 368 369 Figure 3. Contrasts between vibrations for High Ambiguity vocalizations. Top panel: 370 Contrasts between vibrations for high ambiguity angry voices (A 10) for A: Event-related 371 potentials and B-C-D-E: Brain activations from source reconstruction. Bottom panel: Contrasts 372 between vibrations for high ambiguity fearful voices (F 10) for F: Event-related potentials and 373 G-H: Brain activations from source reconstruction . K: Graphical representation of the 374 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint topographical location of the analyzed cluster. For ERPs graphs, line colors for each condition 375 represent the vibration applied (anger in red, fear in blue, no-vibration control in black). Upper 376 horizontal lines represent contrast with substantial proof of diffe rence (BF>3, see Methods) 377 between conditions . Black lower horizontal line represents the period where the two -way 378 Condition X Vibration interaction was significant. Blue dashed horizontal line represents the 3 379 X baseline standard deviation interval. All statistics reported here have Bayes factors > 3. ACC: 380 anterior cingular cortex, DLPFC: dorsolateral prefrontal cortex, IFG: inferior frontal gyrus, IPS: 381 intraparietal sulcus, M1: primary motor cortex, OFC: orbitofrontal cortex, op: pars opercularis, 382 Prec: pr ecuneus, PreM1: premotor cortex, S1: primary somatosensory cortex, SMA: 383 supplementary motor area, SPL: superior parietal lobule. 384 385

386 Research pertaining to the impact of embodied cognition on vocal emotion processing 387 is so far lacking in the literature. Our study sought to demonstrate and better understand the 388 influence of induced bodily vibrations on the perception of emotionally ambiguous voices, at 389 the behavioural and neural levels . The research question revolved around the possible role of 390 the vocal tract vibrations —here induced and systematically manipulated through a device in 391 the throat region, necessary for vocalization and triggered by the use of vibrators, and whether 392 these could represent some type of feedback —potentially interoceptive and proprioceptive 393 feedbacks—that would influence auditory emotion perception , especially in the case of 394 ambiguous voice stimuli. Both behavioural and EEG results revealed effects that support our 395 assumptions: induced vibrations triggered a modulation of behavioral responses to emotional 396 vocalizations as well as of early and late ERP components—highlighting relevant and expected 397 brain modulators through source reconstruction. 398 As mentioned above, o ur study showed a n impact of induced v ibrations on the 399 behavioural responses to morphed emotional voices, which is in line with our main hypothesis. 400 Although not all the expected contrasts were significant, it was clearly observed that our 401 participants categorized emotional vocali zations as ‘anger’ significantly more often when a 402 concomitant anger-like vibration was induced than when it was a fear-like vibration. This is 403 consistent with the study of D’Ausilio et al. (2009) showing the facilitation effect of language 404 comprehension by the transcranial stimulation of the primary motor areas controlling the lips 405 or the tongue. Indeed, in our study, the vibrations would stimulate the mechanoreceptors in the 406 throat in an emotion-specific pattern and this mechanism would then facilitate the recognition 407 of the emotion expressed auditorily and/or bias the response curve—since there are no correct 408 responses for ambiguous voices since they are a blend of fear and anger . This is also coherent 409 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint with studies investigating multimodal integration , which showed a congruence -specific 410 facilitation effect in multimodal processing of emotion (Lin & Ding, 2019; Niedenthal, Brauer, 411 Halberstadt, & Innes-Ker, 2001). 412 Following a similar reasoning, p articipants categori zed ambiguous emotional 413 vocalizations (A60, A50, and A40) as ‘anger’ more often when anger-like vibrations were induced 414 than when fear-like vibrations were induced—or when there were no vibrations induced at all. 415 This specific contrast between anger -like vibrations and no vibration is consistent with the 416 SIMS model ( Niedenthal et al., 2010), postulating that interoceptive feedback would be 417 integrated for non-prototypical emotional stimuli, when classic available emotional information 418 are not clear enough to identify these specific stimuli— here our ambiguous morphed voices. 419 At the neural level, we chose to focus on the functionally best-defined cluster as well as 420 the most relevant one in terms of location (Duville et al., 2022; Gädeke et al., 2013; Maltezou-421 Papastylianou et al., 2022; Paquette et al., 2020), i.e., the frontocentral cluster (C4). T he 422 frontocentral cluster was the locus of an interaction between voice ambiguity and induced 423 vibrations—for 50-150ms, 300-400ms, 450-550ms, 600-650ms, 750-850ms, and 900 -950ms 424 time intervals. Three clearly defined peaks exceeded the baseline threshold (see Methods): a 425 positive deflection from 100 to 150ms, a positive deflection from 200 to 350ms and a negative 426 deflection starting 500ms post -stimulus onset. Our results showed that the induction of both 427 congruent fear and anger-like vibrations led to an increase in P100 amplitude in both low- and 428 high-ambiguity situations, then including the most ambiguous voices. This positive deflection 429 around 100ms recalls the study of Mendoza -Medialdea & Ruiz -Padial (2021) in which 430 interoceptive accuracy was m easured in a heartbeat detection paradigm, for which the 431 participants had to indicate whether an audible tone was played at the same time as their 432 heartbeat or presented with a delay (500ms). In this study, interoception accuracy impacted the 433 capture of exogenous attention by emotional stimuli as early as 100ms post-stimulus onset—434 namely, the P100 component. Also —and although it wa s in the audiovisual modalities, the 435

of Pourtois and colleagues (2000) in t heir emotion recognition study indicate d that 436 multimodal integration could occur at a very early stage of processing, around 110ms post-437 stimulus onset. Furthermore, studies investigating interoception and somatosensory stimulation 438 observed effects on the whole scalp and particularly in the parietal area—in line with our results, 439 albeit at a slightly later stage with a modulation of the P300 component (Herbert et al., 2007; 440 Pollatos et al., 2005; Al et al., 2020; Auksztulewicz, & Blankenburg , 2013; Has et al., 2021). 441 Moreover, t he fact that emotional vibration s have this effect notably for ambiguous voices 442 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint supports the assumption of Niendenthal et al ( 2010) described in their SIMS model. Indeed, 443 ambiguous voices here represent non-protoypical emotional stimuli and therefore other sources 444 of information should be s ought to better understand and interpret the expressed emotion. 445 Therefore, and according to our results, v ibrations in the throat region near the vocal cords 446 would therefore constitute a reliable source of feedback to be considered and further integrated 447 when emotional processing is initiated and ongoing. 448 In our EEG results, we also observed the presence of a P200 component in this 449 frontocentral area. This is consistent with the vast majority of studies investigating emotional 450 voice processing (Duville et al., 2022; Maltezou-Papastylianou et al., 2022; Paulmann & 451 Uskuul, 2017; Paulmann et al., 2013; Gädeke et al., 2013). Our results showed that anger-like 452 vibrations lead to a congruent increase of the P200 peak for low-ambiguity voices, while fear-453 like vibrations lead to a decrease of the P200 peak for ambiguous voices, and for incongruent 454 high-ambiguity angry voices. These results support the above -mentioned findings on 455 multimodal integration and interoception and occur at a slightly later stage . Indeed, we have 456 here congruent vibrations that seem to impact the perception of emotional voice stimuli, which 457 is in line with studies showing that multimodal infor mation seems to facilitate emotional 458 processing when compared to unimodal information (Föcker & Röder, 2019; Kotz et al., 2009). 459 The results also align with previous research on interoception and somatosensory stimulation 460 instigating a P300 component, thro ughout the scalp and specifically in the parietal region, as 461 observed in the past (Herbert et al., 2007; Pollatos et al., 2005; Al et al., 2020; Auksztulewicz, 462 & Blankenburg, 2013; Has et al., 2021). Moreover, the effect of fear -like vibrations for 463 ambiguous voice categorization only is also supported by the SIMS model (Niedenthal, 2007, 464 Niedenthal et al., 2010, p.418). Concerning the P200 decrease for incongruent fear -like 465 vibrations in high ambiguity mentioned above , this result may be due to the fact that, at high 466 ambiguity, voice stimuli are almost equally composed of congruent and incongru ent 467 information. In this case, it is possible that the 40% of fear contained in the high -ambiguity 468 angry voices may have taken over from the angry part and that would lead to this apparent and 469 subtle incongruent effect. 470 Our data finally show late, negative ERP modulations by induced vibrations —most 471 notably between 500 and 800 ms post-stimulus onset . These modulations involve both 472 congruent and incongruent deflections, especially for low -ambiguity voices and for both 473 emotions. For instance, at 500ms, anger vibrations produce an incongruent negative deflection 474 for low-ambiguity fearful voices. These results are consistent with literature on vocal emotion 475 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint processing, as the negative deflections around 500 and 600ms recalls the N400 component 476 observed in frontal, parietal and frontocentral areas during emotional voice processing (Föcker, 477 & Röder, 2019; Maltezou-Papastylianou et al., 2022; Tarai & Srinivasan, 2019). At 800ms, an 478 increase in power is observed for both emotional vibrations compared to no -vibration for 479 ambiguous voices. In the literature, at this point in time in the processing of emotional voices, 480 there is usually instead a positive deflection, called the Late Positive Component (LPC), which 481 is associated with higher -order cognitive processes such as arousal information (Paulmann et 482 al., 2013), expectancy violation in prosody (Astésano et al., 2004; Brattico et al., 2006; 483 Paulmann et al., 2012) and emotional ‘meaning’ (Wei et al., 2022; Zora et al., 2020). However, 484 in our study, although t he power increases during this period, it still remains negative. 485 Nevertheless, this echoes once again the study of Niedenthal and colleagues (2010), which 486 postulates that embodied simulation is only engaged when the emotional stimulus is non -487 prototypical. Here , the vibrations were integrated specifically for ambiguous voices , as 488 prototypical emotional voice processing would not be enough to identify the emotion. At the 489 interoception and proprioception levels, these results globally show that vibrations are 490 integrated during emotional processing at multiple stages, dynamically and throughout the time 491 course of an event, similarly to the multimodal integration shown in audiovisual studies (Föcker 492 & Röder, 2019; Klasen et al., 2012). 493 The ERP data w ere also used for source reconstruction, revealing source modulations 494 in regions involved in vibratory processing, emotional prosody perception, multimodal 495 integration, proprioception and interoception. Regarding vibratory processing, we observed 496 source modulations in the dorsolateral prefrontal cortex, involved in vibrotactile working 497 memory (Burton et al., 2010) . Other regions such as the superior parietal lobule, the inferior 498 parietal sulcus, the precuneus, the insula, the supplementary motor area, as well as the premotor, 499 and the primary somatosensory cortices are known for their involvement in vibratory 500 processing. In fact, they are crucially observed in vibration rate and/or pattern discrimination 501 (Christensen et al., 2007; Coghill et al., 1994; Golaszewski et al., 2006; Hernández et al., 2002; 502 Lenoir et al., 2017; Li Hegner et al., 2010; Macar et al., 2002; Romo et al., 2004; Woolgar & 503 Zopf, 2017; Wu et al., 2018). Concerning interoception, we observed source modulations in the 504 insula, known to be crucially involved in interoceptive processing (Adolfi et al., 2017; Casals-505 Gutierrez & Abbey, 2020) and precuneus, importantly associated with interoceptive awareness 506 (Herbet et al, 2019; Longarzo et al., 2021; Wilson-Mendenhall et al., 2019). The modulation of 507 the primary somatosensory cortex might be more related to proprioceptive information 508 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint processing. Further source modulations were observed in the orbitofrontal cortex, which has 509 also been reported in interoception studies (Singer & Klimecki, 2014; Stern et al., 2017). 510 Regarding emotional prosody perception, we observed source modulations in the dorsolateral 511 prefrontal cortex known for its implication in prosodic emotion worki ng memory (Mitchell, 512 2007) and the superior parietal lobule found in explicit judgments of emotional prosody (Bach 513 et al., 2008), as well as other regions also known to be involved in emotional prosody processing 514 such as the intraparietal sulcus and the su pplementary motor area ( Alba-Ferrara et al., 2011; 515 Belyk & Brown, 2014; Kotz et al., 2003). Interestingly, as already mentioned, we also observed 516 source modulations in the orbitofrontal cortex, associated with higher -level processing of 517 emotional prosody ( Wildgruber et al., 2006) , especially in ambiguous situations (Grandjean, 518 2021). Regarding multimodal integration, we observed source modulations in the insula, also 519 crucially involved in integrating external emotional stimuli and emotional experiences with 520 interoceptive signals (Adolfi et al., 2017; Nguyen et al., 2016; Zaki et al., 2012). As mentioned 521 above, the superior parietal lobule, intraparietal sulcus, premotor and primary somatosensory 522 cortices were also modulated, and they are regions known for their involvement in multimodal 523 integration for auditory, visual, sensorimotor and/or tactile stimuli (Baker et al., 2018; Culham 524 & Valyear, 2006; De Borst & De Gelder, 2016; Gentile et al., 2011; Kreifelts et al., 2007; Parker 525 et al., 2020). Taken together, these results using linear constraint minimum variance ( LCMV) 526 beamformer source reconstruction suggest the implication of a wide -spread network of brain 527 areas in the processing of induced vibrations and their integration with emotional vocalizations, 528 especially in the case of highly ambiguous stimuli, for which such information is most relevant 529 for improved decision making. More specifically, since the above regions are strongly involved 530 in interoception, our results suggest that the vibrations induced on our participants’ throat are 531 processed as interoceptive feedback, which would be integrated into the processing of 532 emotional vocalizations and thus impact auditory perception and decision. 533

and perspectives 534 Our study has some limitations that we need to state. First, our sample consists mainly 535 of female participants, despite studies indicating gender differences in emotional prosody 536 recognition (Imaizumi et al., 2004; Lambrecht et al., 2014) and facial mimicry (Dimberg , & 537 Lundquist, 1990; Niedenthal et al., 2012, Korb et al., 2015). Second, it would be necessary to 538 replicate the effects we observed in our study with different emotions, particularly positive 539 ones, to enable a more reliable generalization of the results. However, morphing on other 540 emotions leads to less natural blends and sometimes strange -sounding voices, which might 541 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint confuse the participants. Third, since the vibrations induced in our participants are supposed to 542 be similar to the ones initiated by the vo cal cords during vocal production, it could be more 543 accurate to use voice stimuli created by the participants themselves. Indeed, acoustic parameters 544 of the vibrations would be much closer to their own bodily vibrations created during vocal 545 production and therefore potentially amplify or solidify the observed effect. We would like also 546 to stress that the induced vibrations were based on recorded sound and the vibrations were 547 initiated on the skin. Such skin vibrations were thought to mimic the vocal chord s vibrations 548 but, of course, the se latter m ight induce subtle tissue vibrations around them, encoded by 549 different mechano -receptors, which might not be induced by the vibration delivered at the 550 surface of the skin, even if the vibrators were localized at the closest possible location related 551 to vocal chords. Further research is needed to better clarify the relationships between vibrations 552 induced by the vocal chords itself and the induced vibrations at the skin level (i.e., the ones that 553 can be systematically manipulated at the experimental level). Fourth, the correction of p-values 554 for multiple comparisons resulted in the elimination of significant trends in some of the results, 555 particularly those concerning the contrasts between vibration conditions. It may therefore be 556 related to the sample size, and with a larger number of participants or with a better balance in 557 the number of men and women participants, these trends could emerge as significant. Finally, 558 future studies would benefit from the inclusion of th e Multidimensional Assessment of 559 Interoceptive Awareness (MAIA; Mehling et al., 2012) questionnaire in the procedure. While 560 the questionnaire does not address vibrations specifically —but rather other bodily sensations, 561 it would still be relevant for such studies. Due to variability in this dimension amongst 562 individuals, using the interoceptive awareness questionnaire as a participant-level covariate 563 could improve the control o ver the data as far as interoception is concerned and it could also 564 allow for an interesting clustering of the participants on that relevant dimension. 565

566 In conclusion, our study provides fresh insights into the topics of embodied cognition 567 and interoception mechanisms in the context of vocal em otion perception. Indeed, b oth our 568 behavioral and EEG findings revealed effects that supported our hypotheses. Specifically, 569 induced vibrations exerted an impact on emotional vocalizations perceptual decisions, and also 570 influenced several early and late ERP components. Moreover, t he source reconstruction 571 analysis revealed source modulations in regions involved in vibratory processing, i.e. encoded 572 by the different kinds of mechano -receptors, emotional prosody perception, multimodal 573 integration, and interoception. Our findings suggest that vibrations do indeed constitute an 574 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint interoceptive signal that is integrated during several stages throughout the process of vocal 575 emotion perception. Further investigation is needed to explore the mechanism(s) involved more 576 deeply, as we believe we are just getting started on this journey toward a more integrated 577 understanding of the influence of somatosensory inputs on auditory perception. 578 579

580 Participants 581 Twenty-four healthy participants were recruited through p osters and via social media 582 dedicated to students from the University of Geneva (18 females and 6 males; Age range = 18-583 41; MAge=23.67; SDAge=5.11). Sample size (N=24) was decided upon using a power analysis 584 adjusted to the design of our study with an effect size of .25, a power of .8 and an alpha of .05 585 (G*Power 3; Faul et al., 2007). All participants were at least 18 years old, had normal hearing 586 and no neurologic or psychiatric history. Participant s gave informed and written consent for 587 their participation in the experiment. The study was approved by the local ethics committee in 588 accordance with ethical and data security guidelines of the University of Geneva and conducted 589 according to the declaration of Helsinki. 590

591 Participants were presented with voice stimuli morphed including both anger and fear 592 voices, presented through headphones (HD25, Sennheiser, DE) with Matlab (The Mathworks 593 Inc., Natick, MA, USA), using the Psychophysics Toolbox ext ensions (Brainard, 1997; Pelli, 594 1997; Kleiner et al., 2007). The stimuli were presented while a fixation cross was displayed on 595 the screen. The morphing was performed by mixing (STRAIGHT toolbox; Kawahara & Matsui, 596 2003) a fearful expression with the speak er’s identity-matching angry expression, resulting in 597 the five conditions (fear%-anger%): 10-90 (A90), 40-60 (A60), 50-50 (A50), 60-40 (A40), and 90-598 10 (A10). We used fifty stimuli (five morphing conditions times ten actors) and their acoustic 599 parameters w ere analyzed (Supplementary Table 1). Voice stimuli were not significantly 600 different on any parameter (amplitude, pitch, loudness, duration, spectral centroid and slope, 601 and F1 frequency). Morphing conditions and actors were randomly distributed across trials. The 602 choice of emotions to be morphed was restricted to anger and fear because (i) mixing positive 603 and negative emotions does not give consistent, human -like stimuli, (ii) including more 604 morphing types would drastically increase the duration of the study that already lasted 2 hours, 605 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint (iii) these two emotions are among the strongest, biologically-relevant ones for survival. Voice 606 and vibration stimuli included ‘Aah’s’ from the Montreal affective voices database (Belin et 607 al., 2008). 608 Bodily resonances were induced in the form of vibrations expressing either pure anger 609 or pure fear in two-thirds of the trials while no vibrations were induced in the remaining third 610 of the trials as a control measure. Three conditions of bodily resonances were therefore 611 implemented (anger, fear, no vibrations). Bodily resonances were created by transmitting 612 emotional voice stimuli to a device that mechanically converted these sound waves into 613 vibrations. The vibrator’s (BC-10, Ortofon, DK) dimensions are 13.5*29.5*18.0 millimeters, it 614 weighs 16.5 grams, has a sensitivity of 118 decibels, a total harmonic distortion of 1.5%, an 615 impedance of 15 ohm and sensibility ranging from 100 to 1000 Hz with a sampling rate of 616 1000Hz. It was positioned close to the vocal cords (Figure 4), taped on the left side to the 617 laryngeal prominence, also known as Adam’s apple, with kinesiology tape —being both soft 618 and strong at the same time. To create the vibration, the vibrator transforms the waveform of 619 an input sound into mechanical energy, here a vibratory signal. Many studies have explored the 620 location of these vibrations, and it appears that regions such as the nasal bone, zygomatic bones, 621 temples, above the upper lip and in the upper neck are the most appropriate to consider (Munger 622 & Thomson, 2008; Nolan et al., 2009; Sundberg, 1992). These vibrations came from the same 623 database as the voice stimuli, but were not morphed (100% anger or 100% fear) and were 624 matched for stimulus duration. The same speaker was used for both vibr ator and voice stimuli 625 for the sake of consistency and to avoid biases and strange intonation effects that could arise 626 when mixing actors. We therefore used twenty stimuli for the vibration (two emotions x ten 627 actors) and their acoustic parameters were ana lyzed (Supplementary Table 2). The acoustic 628 parameters of these vibrations were those of the emotional voice stimuli used, and after a 629 Bonferroni correction (Bland & Altman, 1995), no significant differences emerged. 630 Task Procedure 631 The experiment comprised six experimental blocks (two fear -induced and two anger -632 induced vibrations blocks, alternately), each including one hundred and thirteen randomly 633 presented stimuli (Figure 4). A trial consisted of the presentation of a centered fixation cross 634 screen for five hundred milliseconds, then the presentation of a voice stimulus for one second 635 while the display of the same centered fixation cross screen, and then the categorization of this 636 stimulus by the participant in a two-alternative forced choice between anger and fear using two 637 keyboard keys. Key labels were counterbalanced across participants. Participants were 638 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint instructed to answer as fast and as accurately as possible and the response screen lasted for three 639 seconds. Bodily resonances were induced in two -thirds of the trials (N VibrationTrials= 452, 640 NTotalTrials= 678 ) as mentioned above, with vibrations expressing pure anger or pure fear, 641 matching for stimuli duration and speaker identity with the auditory voice stimuli. The 642 occurrence of the vibrations —in th e vibration trials —was either emotionally congruent or 643 incongruent with the emotional vocalization they accompanied. Therefore, the final factorial 644 design includes two factors: Emotion al Voice (A90, A 60, A 50, A 40, A 10) and Vibration (No 645 Vibrations, Anger, Fear) for a 5 x 3 design. 646 647 Figure 4. Experimental design. Participants were instructed to focus on a black central 648 fixation cross displayed on a grey screen for five hundred milliseconds. Emotional voice stimuli 649 were presented through headphones for one second. Vibrations were induced at the same time 650 in two thirds of the trials and expressed pure anger or pure fear. After each voice stimulus, the 651 cross turned white to indicate that the participant had to categorize the emotion expressed 652 auditorily using the response keys, as fast as accurately as possible and wi thin the 3-seconds 653 response window. 654 Statistical Analysis 655 Behavioral analysis 656 Participants' responses were categorical, with only two possible responses coded as 657 follows: anger = “0” and fear = “1”. A Generalized Linear Mixed Model (GLMM) was 658 performed in R (R Core Team, 2020, Version 1.4.1103) to analyze the responses as a function 659 of both the Vibration (No vibrations, Anger, Fear) and Emotional Voice (A 90, A60, A50, A40, 660 A10) factors and actor gender (male, female) was controlled for as well. An LMM was ch osen 661 to analyze the reaction times with the same factors (results in Supplementary Table 1). These 662 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint models are relevant to our study for their integration of random effects (Bolker et al., 2009; 663 Venables, & Dichmont, 2004). For both models, fixed effects included the interaction between 664 Vibration and Emotional voice, the main effect of participant and speaker gender. In addition, 665 random effects included the random slope of Vibration and Emotional voice as a function of 666 participant identity (uncorrelated) and the rand om intercept of the stimulus identity. The 667 formula was the following (glmer for response, lmer for reaction times): 668 Model.response<-669 glmer(Response~Vibration*EmotionalVoice+ParticipantGender+SpeakerGender+(1+Vibrat670 ion+EmotionalVoice||ParticipantID)+(1|StimulusID), data=data, family = binomial) 671 The effect size of each model was computed and labelled according to the thresholds 672 defined by Cohen (1988; r effects: small ≥ .10, medium ≥ .30, large ≥ .50). Planned contrasts 673 have been performed according to our hypotheses, using Bonferroni correction for multiple 674 testing (Bland & Altman, 1995). 675 EEG preprocessing 676 A classical EEG was acquired with a surface EEG system with 64 amplified electrodes 677 (Biosemi ActiveThree system, BioSemi B.V., Netherlands) and preprocessing was performed 678 using Brainvision analyser 2.2 (Brain Products, Munich, Germany) and Fieldtrip (Oostenveld 679 et al., 2011). A spline interpolation was applied to artifacted channels (5.34±4.74% of channels) 680 and an average reference was applied to the data. Blink and saccade artifacts were rejected 681 using independent component analysis (ICA). Raw data were split into two seconds epochs 682 from -1s to +1s centered on voice onset. Peak-to-peak amplitude was calculated for all epochs 683 across all channels, and a trial rejection’s threshold was calculated as the median + 1.96 684 standard deviation of the peak -to-peak amplitude distribution obtained (~114 μv). For each 685 channel, epochs exceeding the rejection threshold were r ejected from any future analysis 686 (9.44±11.95% of trials rejected in average). For each participant and each channel, ERPs were 687 obtained by averaging epochs for each emotional condition and each vibration type, as well as 688 their conjunction (anger90 -fear10—A90—with anger vibration for example). Grand -average 689 event-related potentials were then obtained by averaging the resulting ERPs across patients for 690 each channel (final average number of trials per participant per condition 40.95+ -6.08, see SI 691 Table 7). 692 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint Event-related topographical clustering analysis 693 For each channel, the grand -average ERP for each condition was averaged into forty 694 25ms bins from voice onset to 1s post-stimulus onset. To increase the specificity of the 695 topographical clustering algorithm to relevant temporal regions of interest, a baseline threshold 696 was defined as ± 3 standard deviations of the grand average ERP for all conditions in the 1s 697 baseline period. A database was created with, as lines the 40 timebins * 15 conditions (3 698 Vibration * 5 Emotional Voice), and as columns the 64 EEG channels for a final matrix size of 699 600 lines * 64 columns = 38’400 cells. If the average power in the bin exceeded the baseline 700 threshold, the average power was reported in the table and if the average power i n the bin was 701 within the baseline confidence interval, a 0 was reported in the database instead. 702 Dimension reduction and topographical clustering was performed using Uniform 703 Manifold Approximation and Projection (UMAP) for Dimension Reduction and Hierarchical 704 Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) for clustering both 705 available in python (McInnes et al., 2017). Previous publications highlighted these methods as 706 the most efficient for noisy high -dimensional data (Allaoui et al., 2020; McInnes et al., 2018; 707 Weijler et al, 2022; Yang et al., 2021). UMAP algorithm was applied to the data with 40 708 neighbor parameters. The number of components was chosen using a measure of specificity—709 the average proportion of components significantly correlated (person’s test, p<.05) with each 710 parameter, and inclusivity—the total proportion of parameters significantly correlated to at least 711 one component. Using this method, the optimal number of components (highest inclusivity and 712 specificity) for this dataset was determined to be 6. 713 The UMAP components were entered into a HDBSCAN algorithm with Cityblock 714 metric. The clustering yielded to 4 homogeneous topographical clusters with 0 non -attributed 715 electrodes (see Figure 2G & Figure 3K). 716 ERP statistical analysis 717 For this analysis, ERPs for each channel, in each trial for each condition was averaged into 718 twenty 50ms bins from stimulus onset to 1s post -stimulus onset. A database was created with 719 the average power at each of the 20 timebins as for th e 3 Vibration and 5 Emotional Voice 720 conditions, for each 64 EEG channel with the 4 clusters and electrode placement (left, center, 721 right) as factors. As power followed a linear distribution, a bayesian general linear mixed model 722 approach was used to validate the result of our topographical clustering. For each timebin, data 723 was entered into a GLMM to test the 3-way interaction between Cluster X Emotional Voice (5 724 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint levels, A 90, A 60, A 50, A 40, A 10) X Vibration (3 levels, Anger, Fear, No vibrations). Random 725 factors included: the gender of the actor‘s voice —to account for gender differences and 726 participant identity—as well as EEG channels. 727 Then, to test significant effects for each cluster and potential lateralization of these effects —728 that escaped topographical clustering analysis, data for each cluster in each timebin was entered 729 into a Bayesian General Linear Mixed model testing the main effects, 2 -way interaction 730 between Emotional Voice (5 levels, A90, A60, A50, A40, A10) X Vibration (3 levels, Anger, Fear, 731 No vibrations), via the BayesFactor package in R. This approach, computing an index of the 732 relative Bayesian likelihood of an Hypothesis 1 (H1) model compared with the null hypothesis 733 (H0) model, supplements the classic GLMM approach, as the Bayes factor com puted by this 734

allows H0 to be assessed. A Bayes factor > 3 constitutes proof of H1, while a Bayes 735 factor 100 or < 1/100). Finally, the 737 comparative nature of Bayesian GLMM makes it unsusceptible to multiple comparison. The 738 gender of the actor’s voice was entered as random factor to account for gender differences and 739 participant identity and as well as EEG channel. If a significant two -way Emotional Voice X 740 Vibration interaction was found for a cluster in a timebin, contrast analysis was performed to 741 test for each significant vibration differences between Emotional Voice conditions, and for each 742 Emotional Voice condition, significant differences between Vibration conditions. Contrast 743 analysis was performed using two-conditions Bayesian GLMM using the same random factor 744 structure as the main analysis in R. Bayes factors being unsusceptible to multiple comparisons, 745 no correction was used. 746 Significant results were considered when ERPs for at least one condition exceeded the average 747 baseline ±3 standard deviations as threshold and for at least the whole 50ms bin. 748 Source reconstruction and analysis 749 Source reconstruction was performed using the MNE python from preprocessed , filtered , 750 artifacts cleaned and epoched data (see EEG preprocessing section above). After forward 751 solution computation using the fsaverage template 752 (https://surfer.nmr.mgh.harvard.edu/fswiki/FsAverage), a LCMV spatial filter was computed 753 using unit-noise-gain as weight normalization and applied to the epoched data. The average 754 timecourse was computed for each epoch and each condition in each label of the HCP-MMP1 755 atlas (mean-flip parameter; Glasser et al., 2016). Then, as performed in the sensor domain (see 756 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint above), source-level ERP for each HCP-MMP1 atlas label, in each trial for each condition was 757 averaged into twenty non-overlapping 50ms bins from stimulus onset to 1s post-stimulus onset. 758 A database was created with the average power at each of the 20 timebins as for the 3 Vibration 759 and 5 Emotional Voice conditions . Significant differences between conditions were 760 investigated using two-conditions Bayesian GLMM with the Participant ID as random factor. 761 762 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint

763 Adolfi, F., Couto, B., Richter, F., Decety, J., Lopez, J., Sigman , M., ... & Ibáñez, A. (2017). 764 Convergence of interoception, emotion, and social cognition: a twofold fMRI meta -765 analysis and lesion approach. Cortex, 88, 124-142. 766 Al, E., Iliopoulos, F., Forschack, N., Nierhaus, T., Grund, M., Motyka, P., ... & Villringer, A. 767 (2020). Heart–brain interactions shape somatosensory perception and evoked potentials. 768 Proceedings of the National Academy of Sciences, 117(19), 10575-10584. 769 Alba-Ferrara, L., Hausmann, M., Mitchell, R. L., & Weis, S. (2011). The Neural Correlates of 770 Emotional Prosody Comprehension: Disentangling Simple from Complex Emotion. 771 PLoS ONE, 6(12), e28701. https://doi.org/10.1371/journal.pone.0028701 772 Astésano, C., Besson, M., & Alter, K. (2004). Brain potentials during semantic and prosodic 773 processing in French. Cognitive Brain Research, 18(2), 172-184. 774 Ashley, V., Vuilleumier, P., & Swick, D. (2004). Time course and specificity of event -related 775 potentials to emotional expressions. Neuroreport, 15(1), 211-216. 776 Auksztulewicz, R., & Blankenburg, F. (2013). Subjective rating of weak tactile stimuli is 777 parametrically encoded in event -related potentials. Journal of Neuroscience, 33 (29), 778 11878-11887. 779 Bach, D. R., Grandjean, D., Sander, D., Herdener, M., Strik, W. K., & Seifritz, E. (2008). The 780 effect of appraisal level on processing of emotional prosody in meaningless speech. 781 NeuroImage, 42(2), 919‑927. https://doi.org/10.1016/j.neuroimage.2008.05.034 782 Baker, C. M., Burks, J. D., Briggs, R. G., Conner, A. K., Glenn, C. A., Taylor, K. N., Sali, G., 783 McCoy, T. M., Battiste, J. D., O’Donoghue, D. L., & Sughrue, M. E. (2018). A 784 Connectomic Atlas of the Human Cerebrum —Chapter 7: The Lateral Parietal Lobe. 785 Operative Neurosurgery, 15(suppl_1), S295‑S349. https://doi.org/10.1093/ons/opy261 786 Banse, R., & Scherer, K. R. (1996). Acoustic profiles in vocal emotion expression. Journal of 787 personality and social psychology, 70(3), 614. 788 Bänziger, T., Grandjean, D., & Scherer, K. R. (2009). Emotion recognition from expressions in 789 face, voice, and body: the Multimodal Emotion Recognition Test (MERT). Emotion, 790 9(5), 691. 791 Bänziger, T., & Scherer, K. R. (2010). Introducing the geneva multimodal emotion portrayal 792 (gemep) corpus. Blueprint for affective computing: A sourcebook, 2010, 271-94. 793 Barsalou, L. W., Niedenthal, P. M., Barbey, A. K., & Ruppert, J. A. (2003). Social 794 embodiment.Psychology of learning and motivation, 43, 43-92. 795 Beilock, S. L., & Goldin-Meadow, S. (2010). Gesture changes thought by grounding it in action. 796 Psychological science, 21(11), 1605-1610. 797 Belin, P., Fillion -Bilodeau, S., & Gosselin, F. (2008). The Montreal Affective Voices: A 798 validated set of nonverbal affect bursts for research on auditory affective processing. 799 Behavior research methods, 40(2), 531-539. 800 Belyk, M., & Brown, S. (2014). Perception of affective and linguistic prosody: An ALE meta-801 analysis of neuroimaging studies. Social Cognitive and Affective Neuroscience, 9 (9), 802 1395‑1403. https://doi.org/10.1093/scan/nst124 803 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint Blairy, S., Herrera, P., & Hess, U. (1999). Mimicry and the judgment of emotional facial 804 expressions. Journal of Nonverbal behavior, 23, 5-41. 805 Bland, J. M., & Altman, D. G. (1995). Multiple significance tests: the Bonferroni method. Bmj, 806 310(6973), 170. 807 Bolker, B. M., Brooks, M. E., Clark, C. J., Geange, S. W., Poulsen, J. R., Stevens, M. H. H., & 808 White, J. S. S. (2009). Generalized linear mixed models: a practical guide for ecology 809 and evolution. Trends in ecology & evolution, 24(3), 127-135. 810 Borgomaneri, S., Bolloni, C., Sessa, P., & Avenanti, A. (2020). Blocking facial mimicry affects 811 recognition of facial and body expressions. PloS one, 15(2), e0229364. 812 Brainard, D. H. (1997). The psychophysics toolbox. Spatial Vision, 10 (4), 4 33–436. 813 https://doi.org/10.1163/156856897X00357 814 Brattico, E., Tervaniemi, M., Näätänen, R., & Peretz, I. (2006). Musical scale properties are 815 automatically processed in the human auditory cortex. Brain research, 1117 (1), 162-816 174. 817 Brosch, T., Grandjean, D., Sander, D., & Scherer, K. R. (2008). Behold the voice of wrath: 818 Cross-modal modulation of visual attention by anger prosody. Cognition, 106(3), 1497-819 1503. 820 Brosch, T., Sander, D., Pourtois, G., & Scherer, K. R. (2008). Beyond fear: Rapid spatial 821 orienting toward positive emotional stimuli. Psychological science, 19(4), 362-370. 822 Brunswik, E. (1956). Perception and the representative design of psychological experiments. 823 Univ of California Press. 824 Burleson, M. H., & Quigley, K. S. (2021). Social interoception and social allostasis through 825 touch: legacy of the somatovisceral afference model of emotion. Social neuroscience, 826 16(1), 92-102. 827 Burton, H., Sinclair, R. J., & Dixit, S. (2010). Working memory for vibrotactile frequencies: 828 Comparison of cortical activity in blind and sighted individuals. Human Brain Mapping, 829 31(11), 1686‑1701. https://doi.org/10.1002/hbm.20966 830 Caldwell, D. G., Lawther, S., & Wardle, A. (1996). Tactile perception and its application to the 831 design of multi -modal cutaneou s feedback systems. In Proceedings of IEEE 832 International Conference on Robotics and Automation, 4, 3215-3221. 833 Campanella, S., & Belin, P. (2007). Integrating face and voice in person perception. Trends in 834 cognitive sciences, 11(12), 535-543. 835 Carminati, M. , Fiori -Duharcourt, N., & Isel, F. (2018). Neurophysiological differentiation 836 between preattentive and attentive processing of emotional expressions on French 837 vowels. Biological psychology, 132, 55-63. 838 Casals-Gutierrez, S., & Abbey, H. (2020). Interocepti on, mindfulness and touch: A meta -839 review of functional MRI studies. International Journal of Osteopathic Medicine, 35 , 840 22-33. 841 Cespedes-Guevara, J., & Dibben, N. (2022). The role of embodied simulation and visual 842 imagery in emotional contagion with music. Music & Science, 5, 20592043221093836. 843 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint Chartrand, T. L., & Bargh, J. A. (1999). The chameleon effect: The perception –behavior link 844 and social interaction. Journal of personality and social psychology, 76(6), 893. 845 Christensen, M. S., Lundbye -Jensen, J., Gee rtsen, S. S., Petersen, T. H., Paulson, O. B., & 846 Nielsen, J. B. (2007). Premotor cortex modulates somatosensory cortex during 847 voluntary movements without proprioceptive feedback. Nature Neuroscience, 10 (4), 848 417‑419. https://doi.org/10.1038/nn1873 849 Coghill, R., Talbot, J., Evans, A., Meyer, E., Gjedde, A., Bushnell, M., & Duncan, G. (1994). 850 Distributed processing of pain and vibration by the human brain. The Journal of 851 Neuroscience, 14 (7), 4095‑4108. https://doi.org/10.1523/JNEUROSCI.14-07-852 04095.1994 853 Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences. Hillsdle. 854 Collignon, O., Girard, S., Gosselin, F., Roy, S., Saint -Amour, D., Lassonde, M., & Lepore, F. 855 (2008). Audio-visual integration of emotion expression. Brain research, 1242, 126-135. 856 Couto, B., Adolfi, F., Velasquez, M., Mesow, M., Feinstein, J., Canales -Johnson, A., ... & 857 Ibanez, A. (2015). Heart evoked potential triggers brain responses to natural affective 858 scenes: a preliminary study. Autonomic Neuroscience, 193, 132-137. 859 Critchley, H. D., & Garfinkel, S. N. (2017). Interoception and emotion. Current opinion in 860 psychology, 17, 7-14. 861 Critchley, H. D., Wiens, S., Rotshtein, P., Öhman, A., & Dolan, R. J. (2004). Neural systems 862 supporting interoceptive awareness. Nature Neuroscience, 7 (2), 189‑195. 863 https://doi.org/10.1038/nn1176 864 Culham, J. C., & Valyear, K. F. (2006). Human parietal cortex in action. Current Opinion in 865 Neurobiology, 16(2), 205‑212. https://doi.org/10.1016/j.conb.2006.03.005 866 D'Ausilio, A., Pulvermüller, F., Salmas, P., Bufalari, I., Begliomini, C., & Fadiga, L. (2009). 867 The motor somatotopy of speech perception. Current Biology, 19(5), 381-385. 868 de Borst, A. W., & de Gelder, B. (2017). fMRI -based multivariate pattern analyses reveal 869 imagery modality and imagery content specific representations in primary 870 somatosensory, motor and auditory cortices. Cerebral cortex, 27 (8), 3994 -4009. 871 https://doi.org/10.1093/cercor/bhw211 872 Desmedt, J. E., & Bourguet, M. (1985). Color imaging of parietal and frontal somatosensory 873 potential fields evoked by stimulation of median or posterior tibial nerve in man. 874 Electroencephalography and Clinical Neurophysiology/Evoke d Potentials Section, 875 62(1), 1-17. 876 Dijkstra, K., Kaschak, M. P., & Zwaan, R. A. (2007). Body posture facilitates retrieval of 877 autobiographical memories. Cognition, 102(1), 139-149. 878 Dimberg, U. (1982). Facial reactions to facial expressions. Psychophysiology, 19(6), 643-647. 879 Dimberg, U., & Lundquist, L. O. (1990). Gender differences in facial reactions to facial 880 expressions. Biological psychology, 30(2), 151-159. 881 Duville, M. M., Alonso-Valerdi, L. M., & Ibarra-Zarate, D. I. (2022). Neuronal and behavioral 882 affective perceptions of human and naturalness-reduced emotional prosodies. Frontiers 883 in Computational Neuroscience, 16, 1022787. 884 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint Engelen, T., Solcà, M., & Tallon-Baudry, C. (2023). Interoceptive rhythms in the brain. Nature 885 Neuroscience, 26(10), 1670-1684. 886 Fadiga, L., Craighero, L., Buccino, G., & Rizzolatti, G. (2002). Speech listening specifically 887 modulates the excitability of tongue muscles: a TMS study. European journal of 888 Neuroscience, 15(2), 399-402. 889 Faul, F., Erdfelder, E., Lang, A. G., & Buchner, A. (2007). G* Power 3: A flexible statistical 890 power analysis program for the social, behavioral, and biomedical sciences. Behavior 891 research methods, 39(2), 175-191. 892 Föcker, J., & Röder, B. (2019). Event-related potentials reveal evidence for late integration of 893 emotional prosody and facial expression in dynamic stimuli: An ERP study. 894 Multisensory Research, 32(6), 473-497. 895 Foglia, L., & Wilson, R. A. (2013). Embodied cognition. Wiley Int erdisciplinary Reviews: 896 Cognitive Science, 4(3), 319-325. 897 Gädeke, J. C., Föcker, J., & Röder, B. (2013). Is the processing of affective prosody influenced 898 by spatial attention? an ERP study. BMC neuroscience, 14(1), 1-15. 899 Gauffin, J., & Sundberg, J. (1974) . Masking effect of one's own voice. Speech Transmission 900 Lab Q Prog Status Rep, 1, 35-41. 901 Gentile, G., Petkova, V. I., & Ehrsson, H. H. (2011). Integration of Visual and Tactile Signals 902 From the Hand in the Human Brain: An fMRI Study. Journal of Neurophysi ology, 903 105(2), 910‑922. https://doi.org/10.1152/jn.00840.2010 904 Gibson, J. (2019). Mindfulness, interoception, and the body: A contemporary perspective. 905 Frontiers in psychology, 10, 2012. 906 Glasser, M. F., Coalson, T. S., Robinson, E. C., Hacker, C. D., Harwe ll, J., Yacoub, E., et al. 907 (2016). A multi -modal parcellation of human cerebral cortex. Nature, 1 –11. 908 http://doi.org/10.1038/nature18933 909 Golaszewski, S. M., Siedentopf, C. M., Koppelstaetter, F., Fend, M., Ischebeck, A., Gonzalez-910 Felipe, V., Haala, I., Struhal, W., Mottaghy, F. M., Gallasch, E., Felber, S. R., & 911 Gerstenbrand, F. (2006). Human brain structures related to plantar vibrotactile 912 stimulation: A functional magnetic resonance imaging study. NeuroImage, 29 (3), 913 923‑929. https://doi.org/10.1016/j.neuroimage.2005.08.052 914 Grandjean, D. (2021). Brain networks of emotional prosody processing. Emotion Review, 13(1), 915 34-43. 916 Grandjean, D., Bänziger, T., & Scherer, K. R. (2006). Intonation as an interface between 917 language and affect. Progress in brain research, 156, 235-247. 918 Has, N. A., Jamaludin, M. N., Rejab, S. M., Ahmad, Z., Zainuddin, N. F., Salam, A. H. A., ... 919 & As’ari, M. A. (2021, November). P300 Somatosensory Validation of Vibrotactile 920 Haptic Feedback for Upper Limb Prosthesis. In 2021 IEEE National Biomedical 921 Engineering Conference (NBEC) (pp. 19-24). IEEE. 922 Hawk, S. T., Fischer, A. H., & Van Kleef, G. A. (2012). Face the noise: embodied responses to 923 nonverbal vocalizations of discrete emotions. Journal of personality and social 924 psychology, 102(4), 796. 925 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint Herbet, G., Lemaitre, A.-L., Moritz-Gasser, S., Cochereau, J., & Duffau, H. (2019). The antero-926 dorsal precuneal cortex supports specific aspects of bodily awareness. Brain, 142(8), 927 2207‑2214. https://doi.org/10.1093/brain/awz179 928 Herbert, B. M., Pollatos, O., & Schandry, R. (2007). Interoceptive sensitivity and emotion 929 processing: an EEG study. International Journal of Psychophysiology, 65(3), 214-227. 930 Hernández, A., Zainos, A., & Romo, R. (2002). Temporal Evolution of a Decision -Making 931 Process in Medial Premotor Cortex. Neuron, 33 (6), 959‑972. 932 https://doi.org/10.1016/S0896-6273(02)00613-X 933 Hess, U., & Blairy, S. (2001). Facial mimicry and emotional contagion to dynamic emotional 934 facial expressions and their influence on decoding accuracy . International journal of 935 psychophysiology, 40(2), 129-141. 936 Hess, U., & Fischer, A. (2014). Emotional mimicry: Why and when we mimic emotions. Social 937 and personality psychology compass, 8(2), 45-57. 938 Holland, A. C., O’Connell, G., & Dziobek, I. (2021). Facial mimicry, empathy, and emotion 939 recognition: a meta-analysis of correlations. Cognition and Emotion, 35(1), 150-168. 940 Hyniewska, S., & Sato, W. (2015). Facial feedback affects valence judgments of dynamic and 941 static emotional expressions. Frontiers in psychology, 6, 291. 942 Imaizumi, S., Homma, M., Ozawa, Y., Maruishi, M., & Muranaka, H. (2004). GENDER 943 DIFFERENCES IN EMOTIONAL PROSODY PROCESSING—AN FMRI STUDY—944 . Psychologia, 47(2), 113-124. 945 Kawahara, H., & Matsui, H. (2003, April). Auditory morphing based on an elastic p erceptual 946 distance metric in an interference -free time -frequency representation. In 2003 IEEE 947 International Conference on Acoustics, Speech, and Signal Processing, 2003. 948 Proceedings.(ICASSP'03). (Vol. 1, pp. I-I). IEEE. 949 Kaschak, M. P., & Maner, J. K. (2009 ). Embodiment, evolution, and social cognition: An 950 integrative framework. European Journal of Social Psychology, 39(7), 1236-1244. 951 Khalsa, S. S., Adolphs, R., Cameron, O. G., Critchley, H. D., Davenport, P. W., Feinstein, J. 952 S., ... & Zucker, N. (2018). In teroception and mental health: a roadmap. Biological 953 psychiatry: cognitive neuroscience and neuroimaging, 3(6), 501-513. 954 Kitamura, T., & Ohtani, K. (2015). Non -contact measurement of facial surface vibration 955 patterns during singing by scanning laser Doppler vibrometer. Frontiers in Psychology, 956 6, 1682. 957 Klasen, M., Chen, Y. H., & Mathiak, K. (2012). Multisensory emotions: perception, 958 combination and underlying neural processes. Reviews in the Neurosciences, 23(4), 381-959 392. 960 Kleiner, M., Brainard, D., & Pelli, D. (2007). What’s new in psychtoolbox-3? 961 Korb, S., Grandjean, D., & Scherer, K. R. (2010). Timing and voluntary suppression of facial 962 mimicry to smiling faces in a Go/NoGo task —An EMG study. Biological psychology, 963 85(2), 347-349. 964 Korb, S., Malsert, J., R ochas, V., Rihs, T. A., Rieger, S. W., Schwab, S., ... & Grandjean, D. 965 (2015). Gender differences in the neural network of facial mimicry of smiles–An rTMS 966 study. Cortex, 70, 101-114. 967 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint Kotz, S. A., Jessen, S., & Paulmann, S. (2009). Investigating the Multimodal Nature of Human 968 Communication. Journal of Psychophysiology, 23(2), 63-76. 969 Kotz, S. a, Meyer, M., Alter, K., Besson, M., von Cramon, D. Y., & Friederici, A. D. (2003). 970 On the lateralization of emotional prosody: An event -related functional MR 971 investigation. Brain and Language, 86 (3), 366‑376. https://doi.org/10.1016/S0093 -972 934X(02)00532-1 973 Kreifelts, B., Ethofer, T., Grodd, W., Erb, M., & Wildgruber, D. (2007). Audiovisual integration 974 of emotional signals in voice and face: An event -related fMRI study. NeuroImage, 975 37(4), 1445‑1456. https://doi.org/10.1016/j.neuroimage.2007.06.020 976 Künecke, J., Wilhelm, O., & Sommer, W. (2017). Emotion recognition in nonverbal face -to-977 face communication. Journal of Nonverbal Behavior, 41, 221-238. 978 Lambrecht, L., Kreifelts, B., & Wildgruber, D. (2014). Gender differences in emotion 979 recognition: Impact of sensory modality and emotional category. Cognition & emotion, 980 28(3), 452-469. 981 Latinus, M., & Belin, P. (2011). Human voice perception. Current Biology, 21(4), R143-R145. 982 Laukka, P., Juslin, P., & Bresin, R. (2005). A dimensional approach to vocal expression of 983 emotion. Cognition & Emotion, 19(5), 633-653. 984 Lenoir, C., Huang, G., Vandermeeren, Y., Hatem, S. M., & Mouraux, A. (2017). Human 985 primary somatosensory cortex is differentially involved in vibrotaction and nociception. 986 Journal of Neurophysiology, 118(1), 317‑330. https://doi.org/10.1152/jn.00615.2016 987 Lenth, R. V. (2021). emmeans: Estimated Marginal Means, aka Least-Squares Means. (Version 988 X.X.X) [R package]. URL: https://cran.r-project.org/package=emmeans 989 Li Hegner, Y., Lee, Y., Grodd, W., & Braun, C. (2010). Comparing Tactile Pattern and 990 Vibrotactile Frequency Discrimination: A Human fMRI Study. Journal of 991 Neurophysiology, 103(6), 3115‑3122. https://doi.org/10.1152/jn.00940.2009 992 Lin, Y., & Ding, H. (2019). Multisensory integration of emotions in a face -prosody-semantics 993 Stroop task. In NeuroManagement and Intelligent Computing Method on Multimodal 994 Interaction (pp. 1-5). 995 Lindı́n, M., Zurrón, M., & Dı́az, F. (2004). Changes in P300 amplitude during an active standard 996 auditory oddball task. Biological psychology, 66(2), 153-167. 997 Longarzo, M., Mele, G., Alfano, V., Salvatore, M., & Cavaliere, C. (2021). Gender Bra in 998 Structural Differences and Interoception. Frontiers in Neuroscience, 14 , 586860. 999 https://doi.org/10.3389/fnins.2020.586860 1000 Lylykangas, J., Surakka, V., Rantala, J., & Raisamo, R. (2013). Intuitiveness of vibrotactile 1001 speed regulation cues. ACM Transactions on Applied Perception (TAP), 10(4), 1-15. 1002 Lylykangas, J., Surakka, V., Rantala, J., Raisamo, J., Raisamo, R., & Tuulari, E. (2009). 1003 Vibrotactile information for intuitive speed regulation. People and Computers XXIII 1004 Celebrating People and Technology, 112-119. 1005 Macar, F., Lejeune, H., Bonnet, M., Ferrara, A., Pouthas, V., Vidal, F., & Maquet, P. (2002). 1006 Activation of the supplementary motor area and of attentional networks during temporal 1007 processing. Experimental Brain Research, 142(4), 475‑485. 1008 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint Macerollo, A., Brown, M. J., Kilner, J. M., & Chen, R. (2018). Neurophysiological changes 1009 measured using somatosensory evoked potentials. Trends in Neurosciences, 41(5), 294-1010 310. 1011 Maltezou-Papastylianou, C., Russo, R., Wallace, D., Harmsworth, C., & Paulmann, S. (2022). 1012 Different stages of emotional prosody processing in healthy ageing –evidence from 1013 behavioural responses, ERPs, tDCS, and tRNS. Plos one, 17(7), e0270934. 1014 Maxwell, G. M., Cook, M. W., & Burr, R. (1985). The encoding and decoding of liking from 1015 behavioral cues in both auditory and visual channels. Journal of Nonverbal Behavior, 1016 9(4), 239-263. 1017 McInnes, L., Healy, J., & Astels, S. (2017). hdbscan: Hierarchical density based clustering. J. 1018 Open Source Softw., 2(11), 205. 1019 McInnes, L., Healy, J., & M elville, J. (2018). Umap: Uniform manifold approximation and 1020 projection for dimension reduction. arXiv preprint arXiv:1802.03426. 1021 Mehling, W. E., Price, C., Daubenmier, J. J., Acree, M., Bartmess, E., & Stewart, A. (2012). 1022 The multidimensional assessment of interoceptive awareness (MAIA). PloS one, 7(11), 1023 e48230. 1024 Mendoza-Medialdea, M. T., & Ruiz-Padial, E. (2021). Understanding the capture of exogenous 1025 attention by disgusting and fearful stimuli: the role of interoceptive accuracy. 1026 International Journal of Psychophysiology, 161, 53-63. 1027 Meteyard, L., Cuadrado, S. R., Bahrami, B., & Vigliocco, G. (2012). Coming of age: A review 1028 of embodiment and the neuroscience of semantics. Cortex, 48(7), 788-804. 1029 Mills, K. A., Scholes, L., & Brown, A. (2022). Virtual Reality and Embodiment in Multimodal 1030 Meaning Making. Written Communication, 07410883221083517. 1031 Mitchell, R. L. C. (2007). fMRI delineation of working memory for emotional prosody in the 1032 brain: Commonalities with the lexico -semantic emotion network. NeuroImage, 36(3), 1033 1015‑1025. https://doi.org/10.1016/j.neuroimage.2007.03.016 1034 Moody, E. J., & McIntosh, D. N. (2011). Mimicry of dynamic emotional and motor -only 1035 stimuli. Social Psychological and Personality Science, 2(6), 679-686. 1036 Moody, E. J., McIntosh, D. N., Mann, L. J., & Weisser, K. R. (2007). More than mere mimicry? 1037 The influence of emotion on rapid facial reactions to faces. Emotion, 7(2), 447. 1038 Mukamel, R., Ekstrom, A. D., Kaplan, J., Iacoboni, M., & Fried, I. (2010). Single-neuron 1039 responses in humans during execution and observation of actions. Current biology, 1040 20(8), 750-756. 1041 Munger, J. B., & Thomson, S. L. (2008). Frequency response of the skin on the head and neck 1042 during production of selected spee ch sounds. The Journal of the Acoustical Society of 1043 America, 124(6), 4001-4012. 1044 Nakagawa, S., & Schielzeth, H. (2013). A general and simple method for obtaining R2 from 1045 generalized linear mixed‐effects models. Methods in ecology and evolution, 4 (2), 133-1046 142. 1047 Neumann, R., & Strack, F. (2000). Approach and avoidance: The influence of proprioceptive 1048 and exteroceptive cues on encoding of affective information. Journal of personality and 1049 social psychology, 79(1), 39. 1050 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint Niedenthal, P. M. 2007. “Embodying Emotion.” Science, 316(5827): 1002–5. 1051 Niedenthal, P. M. (2008). Emotion concepts. Handbook of emotions, 3, 587-600. 1052 Niedenthal, P. M., Augustinova, M., Rychlowska, M., Droit -Volet, S., Zinner, L., Knafo, A., 1053 & Brauer, M. (2012). Negative relations between pacifier use and emotional 1054 competence. Basic and Applied Social Psychology, 34(5), 387-394. 1055 Niedenthal, P. M., Brauer, M., Halberstadt, J. B., & Innes -Ker, Å. H. (2001). When did her 1056 smile drop? Facial mimicry and the influences of emotional state on the detection of 1057 change in emotional expression. Cognition & Emotion, 15(6), 853-864. 1058 Niedenthal, P. M., Mermillod, M., Maringer, M., & Hess, U. (2010). The Simulation of Smiles 1059 (SIMS) model: Embodied simulation and the meaning of facial expression. Behavioral 1060 and brain sciences, 33(6), 417. 1061 Nguyen, V. T., Breakspear, M., Hu, X., & Guo, C. C. (2016). The integration of the internal 1062 and external milieu in the insula during dynamic emotional experiences . NeuroImage, 1063 124, 455‑463. https://doi.org/10.1016/j.neuroimage.2015.08.078 1064 Nolan, M., Madden, B., & Burke, E. (2009). Accelerometer based measurement for the 1065 mapping of neck surface vibrations during vocalized speech. Annual International 1066 Conference of the IEEE Engineering in Medicine and Biology Society, 4453-4456. 1067 Oostenveld, R., Fries, P., Maris, E., & Schoffelen, J. M. (2011). FieldTrip: open source software 1068 for advanced analysis of MEG, EEG, and invasive electrophysiological data. 1069 Computational intelligence and neuroscience, 2011, 1-9. 1070 Paquette, S., Rigoulot, S., Grunewald, K., & Lehmann, A. (2020). Temporal decoding of vocal 1071 and musical emotions: Same code, different timecourse?. Brain Research, 1741, 1072 146887. 1073 Parker G. Allan, Allan, P. G., Robert G. Briggs, Robert G. Briggs, Briggs , R. G., Andrew K. 1074 Conner, Conner, A. K., Christen M. O’Neal, O’Neal, C. M., Phillip A. Bonney, Bonney, 1075 P. A., B. David Maxwell, Maxwell, B. D., Cordell M Baker, Baker, C. M., Joshua D. 1076 Burks, Joshua D. Burks, Burks, J. D., Goksel Sali, … Sughrue, M. E. (2 020). 1077 Parcellation-based tractographic modeling of the ventral attention network . Journal of 1078 the Neurological Sciences, 408, 116548‑116548. 1079 Paulmann, S., Bleichner, M., & Kotz, S. A. (2013). Valence, arousal, and task effects in 1080 emotional prosody processing. Frontiers in psychology, 4, 345. 1081 Paulmann, S., Jessen, S., & Kotz, S. A. (2012). It's special the way you say it: An ERP 1082 investigation on the temporal dynamics of two types of prosody . Neuropsychologia, 1083 50(7), 1609-1620. 1084 Paulmann, S., & Kotz, S. A. (20 08). An ERP investigation on the temporal dynamics of 1085 emotional prosody and emotional semantics in pseudo -and lexical -sentence context. 1086 Brain and Language, 105(1), 59-69. 1087 Paulmann, S., & Uskul, A. K. (2017). Early and late brain signatures of emotional prosody 1088 among individuals with high versus low power. Psychophysiology, 54(4), 555-565. 1089 Pell, M. D., & Kotz, S. A. (2011). On the time course of vocal emotion recognition. PLoS One, 1090 6(11). 1091 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint Pell, M. D., & Kotz, S . A. (2021). Comment: The next frontier: Prosody research gets 1092 interpersonal. Emotion Review, 13(1), 51-56. 1093 Pelli, D. G. (1997). The videotoolbox software for visual psychophysics: Transforming 1094 numbers into movies. Spatial Vision, 10 (4), 437–442. https://d oi.org/ 10.1163/15685 1095 6897X00366 1096 Petrushin, V. A. (2000). Emotion recognition in speech signal: experimental study, 1097 development, and application. In Sixth international conference on spoken language 1098 processing. 1099 Philip, L., Martin, J. C., & Clavel, C. (2018 ). Rapid facial reactions in response to facial 1100 expressions of emotion displayed by real versus virtual faces . i -Perception, 9 (4), 1101 2041669518786527. 1102 Pollatos, O., Kirsch, W., & Schandry, R. (2005). On the relationship between interoceptive 1103 awareness, emotional experience, and brain processes. Cognitive brain research, 25(3), 1104 948-962. 1105 Pollatos, O., Traut -Mattausch, E., Schroeder, H., & Schandry, R. (2007). Interoceptive 1106 awareness mediates the relationship between anxiety and the intensity of unpleasant 1107 feelings. Journal of anxiety disorders, 21(7), 931-943. 1108 Pourtois, G., De Gelder, B., Vroomen, J., Rossion, B., & Crommelinck, M. (2000). The time-1109 course of intermodal binding between seeing and hearing affective information. 1110 Neuroreport, 11(6), 1329-1333. 1111 Price, S., & Jewitt, C. (2013). A multimodal approach to examining'embodiment'in tangible 1112 learning environments. In Proceedings of the 7th International Conference on Tangible, 1113 Embedded and Embodied Interaction (pp. 43-50). 1114 Proverbio, A. M., De Benedetto, F., & Guazzone, M. (2020). Shared neural mechanisms for 1115 processing emotions in music and vocalizations. European Journal of Neuroscience, 1116 51(9), 1987-2007. 1117 Risberg, A., & Lubker, J. (1978). Prosody and speechreading. Speech Transmission Laboratory 1118 Quarterly Progress Report and Status Report, 4, 1-16. 1119 Rizzolatti, G., & Craighero, L. (2004). The mirror -neuron system. Annu. Rev. Neurosci., 27 , 1120 169-192. 1121 Romo, R., Hernández, A., & Zainos, A. (2004). Neuronal Correlates of a Perceptual Decision 1122 in Ventral Premot or Cortex. Neuron, 41 (1), 165‑173. https://doi.org/10.1016/S0896 -1123 6273(03)00817-1 1124 Salamone, P. C., Legaz, A., Sedeño, L., Moguilner, S., Fraile -Vazquez, M., Campo, C. G., ... 1125 & Ibañez, A. (2021). Interoception primes emotional processing: multimodal evidence 1126 from neurodegeneration. Journal of Neuroscience, 41(19), 4276-4292. 1127 Salminen, K. (2015). Emotional Responses to Friction -based, Vibrotactile, and Thermal 1128 Stimuli. 1129 Sander, D., Grandjean, D., Pourtois, G., Schwartz, S., Seghier, M. L., Scherer, K. R., & 1130 Vuilleumier, P. (2005). Emotion and attention interactions in social cognition: brain 1131 regions involved in processing anger prosody. Neuroimage, 28(4), 848-858. 1132 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint Sauter, D. A., & Ei mer, M. (2010). Rapid detection of emotion from human vocalizations. 1133 Journal of cognitive neuroscience, 22(3), 474-481. 1134 Sauter, D. A., Eisner, F., Calder, A. J., & Scott, S. K. (2010). Perceptual cues in nonverbal vocal 1135 expressions of emotion. The quarterl y journal of experimental psychology, 63 (11), 1136 2251-2272. 1137 Scherer, K. R. (2003). Vocal communication of emotion: A review of research paradigms. 1138 Speech communication, 40(1-2), 227-256. 1139 Scherer, K. R., & Oshinsky, J. S. (1977). Cue utilization in emotion at tribution from auditory 1140 stimuli. Motivation and emotion, 1(4), 331-346. 1141 Schirmer, A., & Kotz, S. A. (2006). Beyond the right hemisphere: brain mechanisms mediating 1142 vocal emotional processing. Trends in cognitive sciences, 10(1), 24-30. 1143 Schneider, K. G., Hempel, R. J., & Lynch, T. R. (2013). That “poker face” just might lose you 1144 the game! The impact of expressive suppression and mimicry on sensitivity to facial 1145 expressions of emotion. Emotion, 13(5), 852. 1146 Singer, T., & Klimecki, O. M. (2014). Empathy and com passion. Current biology, 24 (18), 1147 R875-R878. 1148 Stel, M., & Van Knippenberg, A. (2008). The role of facial mimicry in the recognition of affect. 1149 Psychological science, 19(10), 984. 1150 Stern, E. R., Grimaldi, S. J., Muratore, A., Murrough, J., Leibu, E., Fleysher, L., ... & Burdick, 1151 K. E. (2017). Neural correlates of interoception: Effects of interoceptive focus and 1152 relationship to dimensional measures of body awareness. Human brain mapping, 1153 38(12), 6068-6082. 1154 Stevens, L. E., & Fiske, S. T. (1995). Motivation and cognition in social life: A social survival 1155 perspective. Social cognition, 13(3), 189-214. 1156 Strack, F., Martin, L. L., & Stepper, S. (1988). Inhibiting and facilitating conditions of the 1157 human smile: a nonobtrusive test of the facial feedback hypothesis. Journal of 1158 personality and social psychology, 54(5), 768. 1159 Sundberg, J. (1977). The acoustics of the singing voice. Scientific American, 236(3), 82-91. 1160 Sundberg, J. (1983). Chest wall vibrations in singers. Journal of Speech, Language, and 1161 Hearing Research, 26(3), 329-340. 1162 Sundberg, J. (1992). Phonatory vibrations in singers: A critical review. Music Perception: An 1163 Interdisciplinary Journal, 9(3), 361-381. 1164 Švec, J. G., Titze, I. R., & Popolo, P. S. (2005). Estimation of sound pressure levels of voiced 1165 speech f rom skin vibration of the neck. The Journal of the Acoustical Society of 1166 America, 117(3), 1386-1394. 1167 Tarai, S., & Srinivasan, N. (2019). Emotional prosody Stroop effect in Hindi: an event related 1168 potential study. Progress in Brain Research, 247, 193-217. 1169 Thomas, S. J., Johnstone, S. J., & Gonsalvez, C. J. (2007). Event -related potentials during an 1170 emotional Stroop task. International journal of psychophysiology, 63(3), 221-231. 1171 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint Tsolaki, A., Kosmidou, V., Hadjileontiadis, L., Kompatsiaris, I. Y., & Tsolaki, M. (2015). Brain 1172 source localization of MMN, P300 and N400: Aging and gender differences. Brain 1173 research, 1603, 32-49. 1174 Turella, L., Pierno, A. C., Tubaldi, F., & Castiello, U. (2009). Mirror neurons in humans: 1175 Consisting or confounding evidence?. Brain and language, 108(1), 10-21. 1176 Tuuri, K., Eerola, T., & Pirhonen, A. (2010, September). Leaping across modalities: Speed 1177 regulation messages in audio and tactile domains. In International Workshop on Haptic 1178 and Audio Interaction Design (pp. 10-19). Springer, Berlin, Heidelberg. 1179 Vachon, F., Hughes, R. W., & Jones, D. M. (2012). Broken expectations: violation of 1180 expectancies, not novelty, captures auditory attention. Journal of Experimental 1181 Psychology: Learning, Memory, and Cognition, 38(1), 164. 1182 Venables, W. N., & Dichmont, C. M. (2004). GLMs, GAMs and GLMMs: an overview of 1183 theory for applications in fisheries research. Fisheries research, 70(2-3), 319-337. 1184 Watkins, K. E., Strafella, A. P., & Paus, T. (2003). Seeing and hearing speech excites the motor 1185 system involved in speech production. Neuropsychologia, 41(8), 989-994. 1186 Wei, H., He, Y., Kauschke, C., Scharinger, M., & Domahs, U. (2022). An EEG-study on L2 1187 categorization of emotional prosody in German. Speech Prosody 2022, 629-633. 1188 Weijler, L., Kowarsch, F., Wödlinger, M., Reiter, M., Maurer -Granofszky, M., Schumich, A., 1189 & Dworzak, M. N. (2022). UMAP based anomaly detection for minimal residual disease 1190 quantification within acute myeloid leukemia. Cancers, 14(4), 898. 1191 Wiens, S., Mezzacappa, E. S., & Katkin, E. S. (2000). Heartbeat detection and the experience 1192 of emotions. Cognition & Emotion, 14(3), 417-427. 1193 Wildgruber, D., Ackermann, H., Kreifelts, B., & Ethofer, T. (2006). Cerebral processing of 1194 linguistic and emotional prosody: fMRI studies. In Progress in Brain Research (Vol. 1195 156, p. 249‑268). Elsevier. https://doi.org/10.1016/S0079-6123(06)56013-3 1196 Wilson, S. M., Saygin, A. P., Sereno, M. I., & Iacoboni, M. (2004). Listening to speech activates 1197 motor areas involved in speech production. Nature neuroscience, 7(7), 701-702. 1198 Wilson-Mendenhall, C. D., Henriques, A., Barsalou, L. W., & Barrett, L. F. (2019). Primary 1199 Interoceptive Cortex Activity during Simula ted Experiences of the Body. Journal of 1200 Cognitive Neuroscience, 31(2), 221‑235. https://doi.org/10.1162/jocn_a_01346 1201 Witteman, J., Van Heuven, V. J., & Schiller, N. O. (2012). Hearing feelings: a quantitative 1202 meta-analysis on the neuroimaging literature of emotional prosody perception. 1203 Neuropsychologia, 50(12), 2752-2763. 1204 Woolgar, A., & Zopf, R. (2017). Multisensory coding in the multiple -demand regions: 1205 Vibrotactile task information is coded in frontoparietal cortex. Journal of 1206 Neurophysiology, 118(2), 703‑716. https://doi.org/10.1152/jn.00559.2016 1207 Wu, Y., Uluç, I., Schmidt, T. T., Tertel, K., Kirilina, E., & Blankenburg, F. (2018). Overlapping 1208 frontoparietal networks for tactile and visual parametric working memory 1209 representations. NeuroImage, 166, 325‑334. 1210 Yang, Y., Sun, H., Zhang, Y., Zhang, T., Gong, J., Wei, Y., ... & Yu, D. (2021). Dimensionality 1211 reduction by UMAP reinforces sample heterogeneity analysis in bulk transcriptomic 1212 data. Cell reports, 36(4). 1213 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint Zaki, J., Davis, J. I., & Ochsner, K. N. (2012). Overlapping activity in anterior insula during 1214 interoception and emotional experience. NeuroImage, 62(1), 493‑499. 1215 Zora, H., Rudner, M., & Montell Magnusson, A. K. (2020). Concurrent affective and linguistic 1216 prosody with the same emotional valence elicits a late positive ERP response. European 1217 Journal of Neuroscience, 51(11), 2236-2249. 1218 .CC-BY-NC-ND 4.0 International licensemade available 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 The copyright holder for this preprintthis version posted March 24, 2025. ; https://doi.org/10.1101/2025.03.24.644945doi: bioRxiv preprint