Despite decades of research on music and emotion, a fundamental question persists: how do abstract auditory patterns give rise to specific bodily feelings? Listeners do not simply recognize that music is sad or joyful; they often report felt changes in heartbeat, breathing, and muscular tension. This gap between acoustic structure and visceral experience has motivated a range of embodied and simulation-based theories, yet the mechanisms underlying such mappings remain incompletely specified. A prominent class of theories frames music-evoked emotion in terms of embodied simulation. The Mimetic Hypothesis, for example, proposes that listeners understand music via covert motor and subvocal imitation, subtly engaging the musculature of the limbs and vocal apparatus in response to musical structure (Cox, 2016). Similarly, Juslin’s (2013) model of music and emotion highlights emotional contagion and rhythmic entrainment as key pathways: listeners mirror the emotional valence of voice-like acoustic cues and allow internal physiological rhythms—heart rate, breathing, bodily sway—to align with musical patterns. On this view, music does not simply “represent” emotion in an abstract way; rather, it recruits the same systems that underlie vocal expression, movement, and autonomic adjustment.

Neuroscientific work lends support to this embodied perspective. Molnar-Szakacs and Overy (2006) argued that a frontoparietal mirror network encodes musical structure as intentional motor sequences, while the anterior insula links these simulations to the regulation of the body’s internal milieu, thereby contributing to feeling states. Subsequent neuroimaging studies have implicated both mirror-related regions and the insula in decoding musical emotion and in tracking individual differences in empathy and aesthetic sensitivity (Sachs et al., 2018; Koelsch et al., 2021). Yet, despite this growing body of work, existing models rarely specify how particular musical features map onto specific interoceptive states (e.g., cardiac, respiratory, or pain-like sensations), leaving a gap between fine-grained embodied descriptions of musical emotion and mechanistic accounts of bodily feeling.

Interoception is commonly defined as the sensing of the body’s internal condition (Craig, 2002). Contemporary accounts emphasize that interoceptive feelings are not a simple readout of visceral afferents, but inferred states that integrate cardiovascular, respiratory, nociceptive, somatic, and exteroceptive signals with cognitive and affective context (Ceunen et al., 2016; Carvalho and Damasio, 2021). Within this view, interoception is best understood as an integrated, cross-modal percept of bodily state. At the neural level, interoceptive processing is tightly coupled to homeostasis and supported by a hierarchically organized insula–cingulate–prefrontal network. Posterior insula receives homeostatic and visceral inputs, whereas more anterior insula, together with cingulate and prefrontal regions, generates more abstract representations of bodily state and integrates them with exteroceptive information and higher-order appraisal (Carvalho and Damasio, 2021; Damasio and Damasio, 2024). This abstraction and integration imply that interoceptive experience can be shaped by patterns in the external environment. In particular, auditory events that mimic the temporal and qualitative profile of bodily sensations may be incorporated into the interoceptive construct. Consistent with this idea, exogenous simulations of heartbeats—particularly accelerated rhythms—have been shown to modulate interoceptive processing and increase subjective and autonomic arousal (Kleint et al., 2015; Tanaka et al., 2021; Vicentin et al., 2024), indicating that stylized cardiac signals are sufficient to influence emotional and bodily states. From this standpoint, musical sound becomes a candidate source of “pseudo-interoceptive” evidence—evidence that does not originate in visceral afferents but is nevertheless treated as informative about bodily state. In other words, certain acoustic patterns can function as a proxy for interoceptive signals, biasing interoceptive inference and shaping subjective feeling.

Interestingly, historical music aesthetics already anticipated this kind of link between auditory structure and bodily feeling. In the Baroque Doctrine of Affections, musical figures were explicitly theorized as tools for arousing and sustaining specific passions (Buelow, 1973). These embodied associations between musical parameters and bodily states persisted into the pre-Classical and Classical eras and well into the nineteenth century. C. P. E. Bach’s Fantasia in A major (H. 278), written during a gout attack, was characterized by Cramer as materializing “flying pain” through rapid figurations (Head, 2016), while Mozart referred to the aria “O wie ängstlich” from Die Entführung aus dem Serail as expressing a “throbbing heart,” explicitly linking its musical figuration to cardiac sensation (Bauman, 1991). Such descriptions suggest that European art music did not only aim to convey generic affect (e.g., joy or sadness), but sometimes sought to represent specific interoceptive qualia. From a psychological perspective, these historical accounts can be read as early, informal hypotheses about pseudo-interoceptive mappings—claims that specific musical figures can simulate particular patterns of bodily sensation.

Building on this convergence between contemporary interoception research and historical descriptions of bodily feeling in music, the present article adopts the framework of active interoceptive inference as its core theoretical lens (Seth and Friston, 2016). Within this framework, the brain continuously predicts its internal milieu, minimizes interoceptive prediction error through belief updating and allostatic regulation, and subjective feelings arise as the conscious expression of these homeostasis-oriented inferences about bodily state. Against this background, when listeners encounter music containing pain-like or cardiac-like cues, interoceptive generative models are recruited to infer the physiological–affective states of a musical persona—a “virtual body” implied by sound. On this view, emotion contagion during music listening is understood as a consequence of interoceptive generative models minimizing prediction error.

The aims of this article are threefold. First, it formalizes musical emotion as an instance of active interoceptive inference, integrating embodied approaches to music with predictive-coding accounts of perception and affect. Second, it proposes principled mappings between specific musical parameters (e.g., rhythm, dynamics, timbre, harmony, mode) and virtual bodily states, drawing on work on crossmodal correspondences and homeostatic regulation. Third, it illustrates these mappings through analyses of Western classical repertoire, focusing on cardiac- and pain-like qualia and exploring implications for emotion contagion, aesthetic experience, and music-based therapeutic mechanisms. These aims are pursued within a unified theoretical framework that links musical structure to interoceptive prediction and control.

Extending and refining previous theories, the present framework introduces several specific advances. Relative to appraisal-based and categorical models of musical emotion, and to mimetic accounts that emphasize covert motor or vocal imitation, it foregrounds how music can mimic and organize specific bodily sensations (such as cardiac- and pain-like feelings), rather than merely signaling broad emotion categories. It also reformulates the “action program” account of musical feeling proposed by Habibi and Damasio (2014). In their model, music engages evolutionarily conserved action programs for homeostasis and survival: voice-like and rhythmic cues trigger stereotyped autonomic and motor responses (e.g., changes in heart rate, respiration, and skin conductance), and these bodily changes are then mapped by somatosensory and interoceptive cortices into subjective feeling states. The present framework recasts this predominantly bottom-up view within a hierarchical generative model, assigning a portion of the explanatory work to higher levels of the interoceptive system. Finally, the framework exploits the fact that musical works do not present a fixed emotional state but instead unfold over time, proposing that composers can shape the timing and magnitude of musical changes—and the associated prediction errors—to sustain listeners’ attention and deepen their immersion in the evolving musical narrative. From this perspective, rhythmic entrainment as described by Juslin (2013) can be understood not as literal synchronization between the musical beat and the listener’s heart rate, but as the continuous updating of beliefs about the virtual body in response to musical changes that are treated as pseudo-interoceptive evidence.

Having outlined the active inference framework, I now turn to the question of why, and in what sense, specific musical features can be mapped onto interoceptive predictions. In perceptual psychology, one relevant line of work concerns crossmodal correspondences. Spence (2011) reviewed converging evidence that the human perceptual system exhibits systematic tendencies to associate features across sensory modalities. One class of explanations appeals to structural correspondences, which are thought to arise from shared or isomorphic coding principles in the nervous system. Another invokes statistical correspondences, which emerge from learned associations based on the frequent co-occurrence of physical properties in the environment. Together, these findings suggest a principled basis for stable mappings between auditory patterns and bodily sensations.

The following sections outline how specific musical features may be mapped onto interoceptive predictions, with a focus on sensations related to cardiac and pain-like states. The same logic may also extend to other kinds of bodily feeling. For clarity, a distinction is drawn between (1) features that primarily specify the physiological state of a virtual body and (2) features that help label and interpret that state as a particular kind of emotion.

Building on the proposed mappings, this section applies the framework to five works from the Western art-music repertoire, spanning the Classical to late Romantic periods. By integrating score analysis with extramusical context, I explore how pseudo-interoceptive cues might interact with textual or programmatic constraints to suggest specific somatic meanings. Adopting Spitzer’s (2010) holistic approach, these analyses treat emotions not as static “semantic labels” but as varieties of “emotional behavior” enshrined within musical structure and dynamic trajectories unfolding over time. The examples serve to illustrate the framework’s applicability across two primary interoceptive domains: cardiac sensations and pain-like experiences.

Recent research on musical emotion has produced a substantial body of experimental data; however, musical stimuli are still frequently described in broad categorical terms (e.g., happy, sad, fast, or slow), leaving underspecified how particular musical patterns map onto specific bodily sensations. In this article, musical emotion is treated as a special case of active interoceptive inference, a perspective that affords granular mappings between discrete musical features and virtual bodily states. Musical cues act as pseudo-interoceptive evidence, and felt emotion arises as listeners’ generative models minimize prediction error with respect to a musically implied virtual body. At this stage, the account is offered as a conceptual framework: the correspondences proposed are probabilistic tendencies, intended to guide future empirical tests.

Conceptually, this framework distinguishes itself by foregrounding top-down active inference, thereby extending accounts that emphasize bottom-up embodied mechanisms—specifically, the action-program account of Habibi and Damasio (2014) and the rhythmic entrainment and emotional contagion components of Juslin’s (2013) unified theory. In these specific mechanisms, musical structure is typically taken to drive bodily changes and affect via stimulus-driven synchronization and peripheral feedback. Here, by contrast, bodily arousal is treated as a necessary but nonspecific energetic substrate: the listener’s real body constrains what states are plausible, yet the experienced quality of the emotion depends on an inferential interpretation. Specifically, the musically implied virtual body supplies a qualitative frame through which arousal is parsed into particular affective–somatic states, shifting the explanatory emphasis from “what the music makes the body do” to “what bodily state the brain infers, given musical evidence and prior expectations.” For instance, in Symphonie fantastique, insistent pulsations may increase real physiological arousal, yet the listener need not literally entrain to the beat. The resulting conscious experience can be understood as a composite, blending proprioceptive and motoric components (e.g., subvocal rehearsal) with ongoing visceral background, while additionally incorporating simulated cardiac rhythms attributed to the musically implied virtual body.

This difference can be situated within the hierarchical organization of interoception. Lower-level brainstem pathways continue to regulate homeostasis, while insular subregions form a graded interface between bodily physiology and higher-level affective integration. In particular, the posterior insula can be construed as a relay where sensory evidence—including acoustically structured, pseudo-interoceptive cues—can be integrated with ongoing bodily signals and made available to higher-order regions such as the anterior insula. The anterior insula, in turn, supports integrative representations that yield coherent affective–interoceptive inferences about the state of the virtual body. In this way, the framework offers a distinctive mechanism-level claim that complements prior theories: music can shape emotion by selectively weighting and reinterpreting bodily evidence within an interoceptive generative model, thereby generating a controlled, perceptual-affective “hallucination” of bodily change without necessarily perturbing the physiological integrity of the real body.

A longstanding issue in music–emotion research concerns how to characterize the affective states that music evokes, and whether they are “genuine” emotions continuous with everyday life or distinctively “aesthetic” forms of feeling. Juslin and Västfjäll (2008) argued for continuity, whereas Kivy (1990) argued that our response to purely musical structure constitutes a distinctive form of “being moved” that should not be conflated with garden-variety emotions. The present account offers a mechanistic middle ground: physiological arousal in the listener’s real body can be genuine and consequential, while its qualitative interpretation is shaped by inference, because musical structure supplies pseudo-interoceptive evidence that supports a musically implied virtual body—a bodily hypothesis that frames how arousal is experienced. Music-induced affect can therefore be both real (in energetic and autonomic terms) and virtual (in the inferred bodily hypothesis that organizes subjective feeling), clarifying how as-if experience can arise from ordinary interoceptive mechanisms under sensory constraint. This perspective also reframes how listeners can enjoy negative emotions in music (Levinson, 1982) through aesthetic distancing (Menninghaus et al., 2017). The virtual body provides a mechanistic instantiation of “distance,” enabling salient affective–interoceptive qualia without obligatorily recruiting the full set of real-world action policies or corresponding physiological perturbations. In this way, long-standing aesthetic questions could be recast in active-inference terms of policy selection, precision control, and bodily hypotheses.

Beyond explaining how music evokes affect-laden bodily simulations, this framework also extends previous predictive-coding accounts of groove. Vuust et al. (2018) argued that, in the context of syncopation, bodily movement serves to reinforce the internal pulse and resolve rhythmic uncertainty. By contrast, this article foregrounds affective mimicry as a mechanism for minimizing interoceptive prediction error, shifting the analytic focus to the semantic significance of specific pseudo-interoceptive cues. This principle of simulating specific bodily states aligns closely with the predictive coding model of groove proposed by Tsai (in press). Focusing on low-frequency, low-complexity rhythmic patterns, this model argues that the kick drum and bass approximate the multisensory consequences of locomotion—providing footfall-like impact vibrations, vestibular fluctuations, and weight-shifting patterns. When the listener remains physically still, a mismatch arises between this auditory simulation of locomotion and sensory evidence that the body is at rest. The system reduces this prediction error by engaging in covert motor simulation—as-if actions experienced subjectively as the sensation of groove. Crucially, both the present framework and my groove model (Tsai, in press) posit that active inference is anchored in specific internal bodily sensations. Furthermore, both highlight the critical role of timbre—particularly low-frequency energy—in conveying somatic meaning. When a rhythmic pattern delivered via such timbre appears in a musical work, the simulation of locomotion and affective mimicry likely occur simultaneously, fusing into a composite experience of an animated, feeling virtual body.

Building on “mimetic” (Molnar-Szakacs and Overy, 2006; Cox, 2016) and “emotion contagion” (Juslin, 2013) accounts of music, this article advances the field in two key respects. First, it broadens the scope of internal simulation beyond the commonly discussed domain of laryngeal proprioception—specifically, the imagined vocal-fold tension required to produce pitch—to include cardiovascular and pain-like sensations. This perspective underscores the critical role of the accompaniment. While high-register principal melodies typically command attention and invite subvocal mimicry, the underlying, relatively unobtrusive accompaniment covertly shapes the emotional landscape, particularly by simulating the cardiac and motor rhythms of the musical persona.

Second, this article offers a computational rationale for embodied simulation accounts of music: we do not mimic simply for the sake of imitation, but because covert simulation enables the brain to interpret sensory signals more efficiently. From an evolutionary perspective, the neural mechanisms underlying such internal mimicry can be situated within the framework of exaptation—the reuse or “repurposing” of traits that originally evolved for one function but were later co-opted for another (Gould and Vrba, 1982). The sensorimotor system serves as a canonical example: in primates, premotor and parietal circuits likely evolved to support fine-grained sensorimotor control and were later “reused” for action understanding and imitation (Kilner et al., 2007; Hurley, 2008). A similar exaptive logic can be applied to the interoceptive network, which operates as a control system for the internal milieu and has arguably been repurposed from basic homeostatic regulation to support empathy and theory of mind (Ondobaka et al., 2017). I propose that musical experience constitutes a further reuse of this architecture: music recruits this same circuitry anew to track the state of a virtual body implied by sound.

Although the case studies in this article focus on works with text or programmatic descriptions, the proposed mechanism of active interoceptive inference is likely not limited to such contexts. Even in the absence of explicit extramusical information, listeners may still recruit generative models to interpret “absolute” music. For instance, listeners familiar with the classical style can readily associate eighteenth-century “sigh” figures with specific respiratory patterns, using this schematic knowledge to infer specific affective states. Similarly, the Sturm und Drang style—characterized by rapid, low-frequency pulsations in the minor mode, as seen in the opening themes of Mozart’s Symphony No. 25 and Haydn’s Symphony No. 45—can directly simulate the cardio-respiratory signatures of anxiety or fear. Indeed, Spitzer’s (2010) analysis of fearful feelings in Schubert’s Unfinished Symphony demonstrates how pseudo-interoceptive cues can drive affective interpretation purely through musical structure, without reliance on extramusical framing. Furthermore, research indicates that instrumental music frequently evokes visual imagery (Juslin, 2013; Presicce and Bailes, 2019), which likely entails an embodied, interoceptive dimension. Thus, it is plausible to posit that active interoceptive inference remains a primary engine of emotion even in purely instrumental contexts.

A key boundary condition concerns the cultural specificity of the proposed music–interoceptive mappings. Some dimensions of musical structure are likely to be shaped to a greater extent by learned conventions and stylistic enculturation. Harmonic syntax and major–minor tonality are a paradigmatic case: their affective connotations plausibly depend on historically contingent compositional norms and culturally transmitted listening schemata. By contrast, other dimensions—most notably rhythm, tempo, periodicity, and intensity dynamics—may be comparatively more constrained by general perceptual and sensorimotor priors, and thus more likely to support cross-cultural mappings to arousal and action-readiness. A useful illustration comes from Chinese xiqu, where a practice known as jinla manchang (also jinda manchang; “tight accompaniment, slow/free singing”) is widely used—particularly in Yue opera—to depict heightened agitation or emotional escalation while preserving a stretched, quasi-recitative vocal delivery. In this texture, a fast, regular instrumental/percussive pulse (often organized at one or two beats per bar) coexists with a freer, elongated vocal line, yielding two concurrent temporal streams. Within the present framework, the “tight” accompaniment can be construed as pseudo-interoceptive evidence with cardiac-like signatures, whereas the freer vocal layer sustains higher-level narrative and evaluative structure that constrains how arousal is interpreted. This contrast underscores an empirical agenda: mappings grounded in tonal-harmonic conventions should show stronger dependence on cultural familiarity, whereas mappings grounded in rhythmic/tempo cues should generalize more broadly, primarily modulated by context and attentional set.

For musical cues to function as pseudo-interoceptive evidence, they must engage neural circuitry that links auditory representations to interoceptive and affective processing—most notably, pathways between the auditory cortex and the insula. This structural and functional coupling is well documented in neuroimaging research. Diffusion-weighted imaging shows that the structural integrity of white-matter tracts connecting auditory cortex and insula predicts individual differences in musical reward and aesthetic sensitivity (Sachs et al., 2016). Converging fMRI findings likewise suggest that these pathways are crucial for turning acoustic structure into embodied, affect-laden experience: during listening to joyful versus fearful music, emotion-specific functional connectivity emerges between primary and secondary auditory cortices and granular regions of the insula (Koelsch et al., 2021). The insula’s role as a hub of the salience network (Menon and Uddin, 2010) provides an additional layer of explanation for why certain musical events feel particularly gripping. During music listening, functional coupling between auditory cortex and the insula is thought to support the selection of acoustically and emotionally salient events (Putkinen et al., 2021).

Long-term musical experience appears to enhance the efficiency of “sound–body–emotion” integration. Professional musicians show increased anterior insula connectivity with networks supporting empathy, attention, and sensorimotor integration (Zamorano et al., 2017), and this connectivity correlates with higher empathic ability and affective sensitivity (Gujing et al., 2019). Such plasticity is not limited to professional musicians. Older adults with musical experience show stronger dorsal anterior insula–sensorimotor coupling, consistent with reinforced somatic awareness (Ai et al., 2022), and singing training selectively enhances bilateral anterior insula connectivity with speech–sensorimotor networks (Zamorano et al., 2023). Moreover, He et al. (2017) found that a longitudinal music intervention in patients with schizophrenia increased functional connectivity between anterior insula and anterior cingulate cortex, as well as between posterior insula and sensorimotor cortices. Together, these findings suggest that music can shape, and in some cases partially repair, the neural circuitry that integrates internal bodily states with external sensory cues.

Music therapy provides both a testing ground for, and an application of, the proposed mappings between specific musical parameters and interoceptive sensations. For instance, the Therapeutic Function of Music framework systematically links acoustic features such as tempo, contour, and dynamics to targeted levels of physiological arousal and emotion regulation, demonstrating that carefully structured musical elements can modulate bodily state via bottom-up mechanisms (Sena Moore and Hanson-Abromeit, 2015). Extending this logic, adult music listening can be understood as a form of self-administered interoceptive training, in which listeners repeatedly experience musical trajectories from tension or crisis toward resolution and recovery. Specifically, this engagement may enhance interoceptive sensibility—the subjective tendency to focus on and appraise somatic states—and sharpen interoceptive awareness by continuously recalibrating the generative models that underpin our conscious feeling of the body. Significantly, in contemporary media environments, problematic smartphone use and addiction-like social media engagement have been associated with reduced insula gray-matter volume and altered salience-network connectivity (Turel et al., 2018; Lee et al., 2024). In this context, sustained engagement with musical narratives—where interoceptive states unfold over extended timeframes—offers a vital alternative mode of experience. By continuously exercising and recalibrating interoceptive predictive models to resolve high pwPEs, such listening habits may potentially counteract these deficits and strengthen capacities for interoceptive awareness and regulation.

The proposed mappings between specific musical parameters and interoceptive sensations also invite a rethinking of musical listening, performance, and interpretation. If musical emotion is partly grounded in listeners’ schematic knowledge of interoceptive states, affective responses to music need not be regarded as purely intuitive or fixed. Listeners and performers can, in principle, learn to associate particular musical gestures with particular interoceptive qualities. Such learning is likely facilitated by basic music-analytic and physiological knowledge, together with enriched extramusical information. On this view, sensitivity to the interoceptive dimensions of music is a trainable skill. Future studies could test this claim by comparing behavioral reports, peripheral physiological responses, and neural markers of interoceptive–auditory processing before and after targeted training in recognizing specific musical cues.

Complementary experiments could systematically manipulate cardiac-like versus pain-like cues within tightly controlled musical stimuli to determine whether—and through which pathways—these features modulate canonical markers of interoceptive inference. A recent intracranial electrophysiology study (Banks et al., 2023) derived a low-dimensional “functional geometry” of auditory cortical resting-state networks and showed that the posterior insula occupies an intermediate position between auditory cortex and limbic structures in the resulting embedding. Banks et al. (2023) further proposed that, given its robust auditory responsiveness, the posterior insula may help transform auditory cortical information into affective representations in the anterior insula, consistent with a linking role between auditory and limbic systems. While resting-state geometry cannot establish directionality, it helps sharpen a testable hypothesis: pseudo-interoceptive musical cues should reconfigure effective connectivity. Specifically, they should enhance coupling between auditory cortex and posterior insula, and strengthen posterior-insula interactions with the anterior insula and anterior cingulate cortex. This pattern would be consistent with acoustic evidence being converted into affectively salient interoceptive representations and subjective qualia. These predictions can be tested by combining connectivity models with concurrent autonomic measures. Cardiac-like cues should preferentially bias cardiovascular-related bodily hypotheses and autonomic readiness, whereas pain-like cues should more strongly engage insula–cingulate pathways linked to salience, aversive qualia, and protective bodily predictions.

The network-level organization reported in Banks et al. (2023) also offers a principled way to distinguish what is driven by musical structure from what is driven by contextual priors. Their findings place auditory cortex in close functional proximity to a limbic–semantic constellation that includes the temporal pole and medial temporal lobe structures, providing an anatomically plausible route through which semantic and mnemonic information can shape auditory inference. This aligns with task-based evidence linking the temporal pole to context-sensitive socioemotional integration (Pehrs et al., 2014; Pehrs et al., 2018; Li et al., 2019). Accordingly, an empirically tractable approach is to hold the acoustic stimulus constant while manipulating biographical knowledge, textual meaning, or programmatic narratives as contextual primes that vary the precision of higher-level priors. Mechanistically, precision weighting can be framed more simply as context-dependent gain modulation of prediction-error signaling and belief updating, which should be observable as systematic changes in directed coupling between auditory cortex, posterior insula, temporal pole/medial temporal circuitry, and anterior insula/anterior cingulate cortex. On this account, contextual primes should produce shifts in (i) the relative coupling between auditory cortex and posterior insula versus coupling between temporal pole/medial temporal circuitry and anterior insula/anterior cingulate cortex, (ii) the relative weighting of cardiac-like versus pain-like bodily hypotheses inferred from the same musical input, and (iii) the correspondence between subjective interoceptive qualia and peripheral physiology.

This article proposes an active interoceptive inference framework wherein musical cues function as pseudo-interoceptive evidence, prompting a “controlled hallucination” of bodily change. On this view, listeners recruit generative models to infer the state of a virtual body, with felt emotion arising as these models minimize prediction error. Admittedly, this framework captures only part of the multifaceted ways in which music moves us; mechanisms such as memory, reward, and social meaning undoubtedly interact with interoceptive inference in ways that warrant further investigation. Moreover, while the musical analyses presented here focus on Western art music and on cardiac- and pain-like qualia, future work should explore whether similar principles extend to other genres, cultures, and somatic domains—including respiratory rhythms, thermal sensations, muscular tension, vibrotactile roughness (e.g., the tingling, buzzing sensation often reported with rock music), and even experiences of weightlessness.