Sage Journals: Discover world-class research

Abstract

Recent research has suggested that low-level psychoacoustic parameters such as loudness and spectral brightness are correlated with musical chills, a subjective emotional experience accompanied by goosebumps, shivers, and tingling sensations. These relationships may be explained by a vigilance theory of chills, through the process of auditory looming; however, these correlations or theories have never been causally tested. In the current study, participants (N = 40) listened to five variations (original, low loudness, high loudness, low brightness, high brightness) of an experimental and control piece of chills music, characterized by a crescendo and guitar solo respectively; this qualitative distinction was made based on whether the underlying musical structure of the pieces was or was not capable of engaging auditory looming processes. It was predicted that increases in loudness would result in increased chills frequency across participants, indicated by button presses; brightness was included as an exploratory parameter. Results show that for the experimental piece, increases in loudness resulted in significantly more frequent chills experiences, and increases in brightness significantly reduced the frequency of chills, whereas no effects were found for the control piece. Findings are discussed in terms of vigilance and social bonding theories of chills, and the complex interactions between low-level psychoacoustic properties and higher-level musical structures.

Keywords

Chills emotion music psychoacoustics psychophysiology

People consistently engage with music for many reasons; for example, music can be used to construct and reaffirm social identities (Lonsdale & North, 2011), aid exercise activities (Laukka & Quick, 2013), and alleviate loneliness (Groarke & Hogan, 2016; Lippmann & Greenwood, 2012). Notably, a crucial motivation for listening to music is to experience and regulate emotions (Schäfer et al., 2013); accordingly, researchers have investigated the associations between music and emotion from several perspectives, including understanding the perception of emotions in music (Fritz et al., 2009; Juslin & Timmers, 2010), characterizing emotions elicited by music (Zentner et al., 2008), and postulating on the possible mechanisms through which music can elicit emotions in listeners (Juslin & Västfjäll, 2008).

Common approaches to exploring links between music and emotion include identifying relationships between characteristics of music, and perceived or induced emotions. A key direction has focused on psychoacoustic parameters of music, such as loudness or acoustic intensity, and spectral properties including brightness (proportion of high to low frequencies in a sound). Regarding the perception of emotions in music, Juslin (2000) found that expressions of anger and fear were associated with high-frequency energy; similar results were found for loudness. Eerola et al. (2012) documented linear relationships between acoustic features and perceived affect ratings for instruments, such as spectral brightness and energy arousal. Lange and Frieler (2018) critically assessed music information retrieval procedures, but importantly described loudness, tonality, and tempo as crucial predictors for several emotions including anger, peacefulness, tenderness, and happiness. Laukka et al. (2013) carried out a cross-cultural study, finding that loudness was generally linked to anger, and negatively associated with affection and peacefulness; additionally, anger was linked to timbres characterized by higher-frequency energy (see Hailstone et al., 2009). Dean et al. (2011) inverted acoustic intensity levels in musical stimuli, and found similar changes in listener judgments of arousal, reflecting previous work highlighting relationships between acoustic intensity and perceived arousal (Gabrielsson & Lindström, 2010; Leman et al., 2005). As a related index of brightness (McAdams, 2013), the spectral centroid has been linked to tension and energy arousal (McAdams et al., 2017), although less clear associations were reported with continuous ratings of valence and arousal (Schubert, 2004). From a different perspective, Eerola (2011) reported that links between emotions and psychoacoustic parameters differed depending on the music genre. Overall, the extant literature suggests that properties in music, including loudness and brightness, are consistent communicators of emotional expression in music (Balkwill & Thompson, 1999; Balkwill et al., 2004), with relationships possibly reflecting acoustic similarities between music and emotional speech prosody (Juslin & Laukka, 2003; Ilie & Thompson, 2006).

Some studies have also explored links between felt emotional arousal and psychoacoustic parameters. Sound intensity has been associated with physiological arousal, including skin conductance and respiration rate (Gomez & Danuser, 2007); relatedly, loudness and spectral centroid have been linked to subjective and physiological arousal (Coutinho & Cangelosi, 2011). In contrast, Gingras et al. (2014) controlled for sound amplitude, and reported key spectral features related to subjective arousal ratings, including spectral flux and entropy. In a cross-cultural approach, Egermann et al. (2015) assessed emotional responses in Canadian and Pygmy participants to Western or Congolese Pygmy music, finding that loudness was linked to increased subjective arousal only in Canadian listeners, whereas spectral centroid was linked to physiological arousal increases across both groups of listeners; these results emphasize the role of culture in emotion induction and psychoacoustic features.

Despite this work, there remains a lack of research linking psychoacoustic features to felt, as opposed to perceived, musical emotions; additionally, arousal is often operationalized as a broad indicator of activation and energy, meaning that while these measures are open to further correlational and causal research, existing associations are difficult to explain, with underlying psychological mechanisms poorly understood. Crucially, while facets such as loudness and brightness have been associated with perceived and felt emotional arousal during music listening, they have also been linked recently to a phenomenon indicative of peak emotional arousal, labelled “musical chills” (Bannister & Eerola, 2018; Rickard, 2004); this experience serves as the focus of the current study.

Musical Chills

Musical chills have received notable attention in music and emotion research. Chills are often characterized as indicators of peak pleasure; for example, Blood and Zatorre (2001) documented correlations between chills and brain activity in reward circuits including the nucleus accumbens and ventral striatum (see also Salimpoor et al., 2011). However, chills can be experienced across different aesthetic domains and may be conceptualized in various ways; indeed, there is some debate as to whether chills reflect a unitary peak experience of pleasure, or collection of distinct responses (Levinson, 2006; Maruskin et al., 2012; Panksepp, 1995; Pelowski et al., 2017). Regardless, a common working definition utilized in previous literature describes chills as a subjective emotional experience, accompanied by goosebumps, shivers, and tingling sensations.

Musical chills have been utilized as indicators of emotional experiences in empirical settings (Grewe et al., 2007; Rickard, 2004), given a suitable synchrony between self-reported subjective feelings and physiological responses, including skin conductance (Craig, 2005) and pupil dilation (Laeng et al., 2016). Additionally, the phenomenon has been linked to various musical characteristics and features (Harrison & Loui, 2014), individual differences such as openness to experience (Nusbaum et al., 2014), approach and avoidance behaviors (Maruskin et al., 2012) or trait empathy (Bannister, 2019), and listening contexts (Egermann et al., 2011).

Most work on musical chills has focused on relationships between the response and musical features. Sloboda (1991) found that shivers were linked to sudden dynamic and textural changes, or unprepared harmonies in music. Panksepp (1995) also highlighted links between chills and musical features such as crescendos. These correlations were developed further by Grewe et al. (2007), noting that chills were linked to features such as sudden dynamic changes, and entrances of new voices or instruments. Furthermore, Guhn et al. (2007) suggested that the interaction between solo and accompaniment instrumentation in a piece was correlated with chills experiences.

Recently, a survey of various characteristics of musical chills in a representative sample noted a variety of musical features linked to the response (Bannister, 2018); these include the human voice, lyrics, perceptions of social unity and communion in music, and structural features previously reported such as crescendos and moments of contrast or change. Furthermore, in a recent first attempt at causally manipulating musical chills experiences, Bannister and Eerola (2018) removed chills sections in three pieces of music, characterized by a crescendo, famous string theme or guitar solo. Notably, by removing these sections, the frequency of chills was reduced in listeners, alongside continuous ratings of chills intensity.

Crucially, relationships are occasionally proposed between chills and psychoacoustic parameters such as loudness and spectral features. Nagel et al. (2008) noted that frequency-dependent changes in loudness (between 920 and 4,400 Hz) were common in excerpts eliciting chills across participants; interestingly, Halpern et al. (1986) suggested that frequencies just below 4,000 Hz were associated with unpleasant chills experiences (e.g., in response to nails scratching a blackboard), and anatomical work has long documented a sensitivity of the human ear to frequencies between 3,000 and 4,000 Hz (Fletcher & Munson, 1933). Guhn et al. (2007) noted that chills passages in music could often be characterized by an expansion of the frequency range; furthermore, Grewe et al. (2007) found that peaks in loudness and auditory roughness, a complex measure of perceptual dissonance of an auditory stimulus (Plomp & Levelt, 1965), were linked to the onset of chills experiences. Most recently, Bannister and Eerola (2018) investigated, using continuous measurements of chills intensity, the association between chills and various psychoacoustic parameters. Notably, for two of the three pieces utilized, strong positive correlations were reported between reports of chills intensity, and both root-mean-square energy (RMS) and spectral brightness (ratio of high frequency to low frequency energy in a signal) levels; these findings reflect the prevailing importance of loudness and brightness in music and emotion.

Importantly, musical chills research is limited by a lack of causal investigations, which is significant for two related reasons: Firstly, without causal inferences, existing theories of musical chills cannot be tested; secondly, by understanding the causal underpinnings of musical chills, researchers may identify and test psychological processes underlying emotions with music more broadly, such as those between psychoacoustics and emotional arousal. Given that chills are an indicator of peak emotional arousal, but are also a more specific construct compared to arousal, the phenomenon represents a unique opportunity to test and develop ideas regarding both how chills are elicited by music, and what psychological mechanisms underlie associations between music, psychoacoustics and emotional arousal.

Considering the need for further causal investigations into musical chills, correlations between the phenomenon and psychoacoustic parameters such as loudness and spectral brightness (Bannister & Eerola, 2018; Grewe et al., 2007; Guhn et al., 2007), and an underrepresentation of psychoacoustic parameters in research on emotions induced by music, affecting chills responses through manipulating loudness and brightness provided the central focus of this study.

Vigilance Theory of Musical Chills

Most structural features in music linked to chills can be understood as violating or delaying musical expectations developed by the listener. These expectations have long been considered to underlie musically induced emotions (Meyer, 1956; Steinbeis et al., 2006), and reflect implicit predictions as to what will come next in a musical progression; there is growing evidence that these predictions are constructed, developed, and refined through processes of enculturation and statistical learning (Pearce et al., 2010). Notably, Huron (2006) describes how emotions may be elicited through expectations: Listeners construct predictions about how the music will progress; if these expectations are violated, an immediate “worst-case scenario” appraisal is carried out, potentially linked to fear, attention, and vigilance (as incorrect predictions are maladaptive). Subsequently, the slower aesthetic judgments and appraisals are engaged, and a moment of contrastive valence occurs from a negative quick appraisal, to a positive, “safe” listening context appraisal. This moment is described by Huron (2006, p. 22) as a potential behavioral and physiological mechanism resulting in peak pleasure, through the slower positive appraisals being amplified by the contrast with the preceding negative reactions, and the unexpectedness of the positive outcome; however, the idea of contrastive valence has yet to be tested in the music-listening context. Crucially, this theory may explain a proportion of musical chills experiences, by tapping into evolutionary threat-signalling and fight-or-flight functions of goosebumps (Darwin, 1872), in which goosebumps enlarge the perceived size of the host, and act as a visual signal to conspecifics regarding an incoming threat (i.e., expectancy violation) that demands attention.

These chills experiences may be characterized in terms of high attention and vigilance, although the resulting phenomenological experience may vary depending on complex interactions between the music, listener, context, and aesthetic judgments (Juslin, 2013). Interestingly, music and psychoacoustic parameters may be particularly suited to induce this vigilance chills response through auditory looming (Neuhoff, 1998, 2001), which refers to the perception of a sound source as approaching, often through the gradual increase in acoustic intensity or loudness (Ghazanfar et al., 2002). This process has been theorized to hold adaptive significance, as an approaching source may constitute a threat; if so, then listeners should display a perceptual and attentional bias for these types of sounds. Indeed, this has been confirmed through a variety of measures; sounds with increasing intensity are perceived to change more in terms of loudness compared with decreasing intensity sounds (Bach et al., 2009; Neuhoff, 2001), and are judged to be longer in duration (Grassi & Darwin, 2006; Ries et al., 2008). Participants also underestimate the “time-to-arrival” of a sound source with increasing intensity ramps (Neuhoff et al., 2009), and reaction times to photographs were quicker following sounds with increasing intensity compared with decreasing (Tajadura-Jiménez et al., 2010). Finally, sounds with increasing sound intensity have been linked to increases in phasic skin conductance, heart deceleration, and increased amygdala activity (Bach et al., 2008). In line with previous research on musical chills, there is evidence to suggest that crescendos and sudden dynamic changes place loudness in a privileged position as an elicitor of musical chills, potentially through auditory looming mechanisms.

The relationship between spectral brightness and chills is less clear in the vigilance theory context. Anecdotally, Panksepp (1995) noted that chills may be elicited by a piercing, high-pitched solo voice or instrument; indeed, chills have recently been described as occurring occasionally when performers “hit the high note” (Bannister, 2018); however, while brightness and pitch are related, they are dissociated as physical and perceptual attributes respectively. Increases in brightness may reflect similar auditory looming mechanisms: As lower frequency sound waves are absorbed less and generally travel further than those at higher frequencies, increasing ramps of higher frequency energy may act as a spectral analogy for auditory looming and decreased proximity between listener and event. Notably, Kovacevich and Huron (2018) recently proposed a vigilance account of the related autonomous sensory meridian response (ASMR) phenomenon (Barrett & Davis, 2015), often elicited by whispering sounds that imply proximity and are characterized by high brightness. However, this appears less intuitive than the link between auditory looming and loudness, and brightness has yet to receive empirical support in broader auditory looming research; an alternative possibility is that the existing correlations between chills and brightness are mediated by different, underlying psychological mechanisms. Given the background literature linking brightness to expressed and felt musical emotions, and recent correlations between chills and brightness (Bannister & Eerola, 2018), this feature was explored alongside loudness in the current work.

Rationale for the Current Study

The current study aimed to carry out a novel, causal investigation into musical chills, and further the theoretical understanding of the phenomenon. Given the link between loudness changes and musical chills, the current study aimed to provide a test of existing theories of chills, specifically the vigilance hypothesis and auditory looming mechanism (Huron, 2006). However, spectral brightness was also targeted as an exploratory factor previously correlated with chills (Bannister & Eerola, 2018), a parameter that could be linked to vigilance processes, but also other mechanisms. Therefore, it was hypothesized that increases in loudness would result in a higher frequency of chills reports across stimuli, and that the opposite would be true for decreases in loudness; given the sparsity of empirical data linking brightness to auditory looming, no specific predictions were made regarding this parameter.

Methods

Design

A listening experiment was designed using an experimental and control excerpt from two pieces of music, previously shown to be unfamiliar to participants, and effective at eliciting chills (Bannister & Eerola, 2018). For each excerpt, there were five listening conditions (10 in total), corresponding to one original version and four manipulated versions: increased or decreased loudness, and increased or decreased brightness.

The primary dependent variable was the frequency of self-reported chills. As a secondary dependent variable, the intensity of reported chills was assessed through two measures: Skin conductance amplitudes during chills reports, and the average duration of chills reports. Subjective ratings were also collected following each excerpt, regarding the overall listening experience. The experiment followed a within-subjects design, with participants split into two experiment groups, each containing five of the ten listening conditions; these represented an original, high, or low loudness, and a high or low brightness condition. In each group, participants heard two versions of one stimulus and three of another in a pseudo-randomized order defined by two criteria: (1) any version of the experimental or control stimuli could not be heard twice in succession, and (2) versions manipulating the same psychoacoustic parameter could not be heard twice in succession. This was implemented to address effects of fatigue, order effects, and habituation in responses resulting from the within-subjects design. Importantly, participants in one group all experienced the same five listening conditions, with the remaining five attributed to the other group. The pseudo-randomization procedure, and the allocation of listening conditions across the two experiment groups, is described and visualized in the supplementary material (available online).

Participants

All participants were recruited through a pre-screening process, which involved a question confirming that participants had experienced chills with music previously (yes/no). In total, 40 participants took part in the experiment. Of the sample, 24 were female, and the mean age was 28.5 years (SD = 8.36, range = 19–52). An a priori power calculation for the experiment was difficult, given the lack of effect sizes corresponding to psychoacoustic manipulations in a musical chills paradigm; consequently, the current sample size was derived from corresponding samples in previous correlational work on the topic (Blood & Zatorre, 2001, N = 10; Craig, 2005, N = 32; Egermann et al., 2011, N = 14; Grewe et al., 2007, N = 38; Rickard, 2004, N = 21; Salimpoor et al., 2009, N = 32; Salimpoor et al., 2011, N = 10). However, standardized effect sizes are reported in the current experiment to aid power calculations for future research on chills. The descriptive statistics of the sample are presented in the supplementary material (available online).

Materials and Measures

Self-Reports

For every stimulus that participants listened to, several rating scales were completed, administered in the same order on each occasion. After each piece, participants were asked to confirm how familiar they were with the music (Likert scale 1–5), rate their emotional experience (Likert scale 1–7; descriptors included affection, agitated, energy, enjoyment, joy, moved, nervous, nostalgia, sad, sentimental, and tender), and to report any specific bodily activity (yes/no; descriptors included cold, warm, lump in the throat, tears, smiling, and frowning). Emotion descriptors were derived from factors in the Geneva Emotional Music Scale (Zentner et al., 2008) that reflect possible positive and negative dimensions of chills such as joy, enjoyment, sadness, nervousness, and agitation (Maruskin et al., 2012), and certain emotions linked to chills including tenderness, nostalgia, energy, being moved, and affection (Bannister, 2018); bodily activity descriptors were derived from previous literature on chills (Algoe & Haidt, 2009; Bannister, 2019; Fiske et al., 2017; Maruskin et al., 2012; Wassiliwizky, Jacobsen et al., 2017). These descriptors were utilized to assess whether the psychoacoustic manipulations altered the overall listening experience, and to explore whether participants reporting chills to a stimulus demonstrate different responses compared with those who report no chills.

After participants had listened to the first two excerpts in the experiment, they completed the Behavioral Activation/Inhibition Scales (BIS/BAS; Carver & White, 1994), reflecting two proposed, general motivational systems underlying human behavior: Approach, referring to a propensity to maintain contact or move closer to a desirable stimulus or event, and avoidance, describing the tendency to separate or move further away from undesirable stimuli or events. These dimensions were assessed for three reasons: Firstly, reward seeking has been linked to more intense chills experiences and reported frequency of chills (Mori & Iwanaga, 2015), but requires replication; secondly, the approach and avoidance distinction has been utilized to categorize two kinds of chills experiences, namely goosetingles and coldshivers respectively (Maruskin et al., 2012); finally, while it is difficult to assess tendencies to react strongly with attention and vigilance to external events, it has been suggested that behavioral inhibition is associated with experiencing fear, harm-avoidance, or anxiety (Carver & White, 1994; Jorm et al., 1998), an individual difference that might predict susceptibility to auditory looming in music, and in turn, chills.

As a final ancillary measure, trait empathy data were collected using the Interpersonal-Reactivity Index (IRI, Davis, 1983); this was motivated by the finding that chills indicate states of being moved (Benedek & Kaernbach, 2011; Wassiliwizky et al., 2015), an experience that may underlie pleasurable sadness with music, mediated by empathy (Eerola et al., 2016). The inclusion of trait empathy was favoured over previous measures linked to chills, such as openness to experience, for three reasons: Firstly, empathy has rarely been considered in relation to chills (Bannister, 2019), whereas openness to experience has received repeated attention (Colver & El-Alayli, 2016; McCrae, 2007; Nusbaum & Silvia, 2011); secondly, openness to experience encapsulates broad trait characteristics, and correlates with empathy (Melchers et al., 2016); and thirdly, associations between openness to experience and chills are occasionally difficult to interpret, given that widely used personality inventories include a specific question pertaining to experiencing chills in aesthetic engagements as an indicator of the dimension (McCrae, 2007).

Stimuli

For the current experiment, an experimental stimulus and control stimulus were selected based on their unfamiliarity, and on their efficacy to elicit chills in participants from a previous study (Bannister & Eerola, 2018); this qualitative distinction between experimental and control stimuli was based on whether the underlying musical structure of the excerpts was intuitively linked to auditory looming (experimental), or not (control).

The experimental stimulus was from “Glósóli,” by Sigur Rós, a rock or post-rock piece of music, with Icelandic lyrics. An excerpt (Glósóli) taken from this piece was 154 seconds in duration (from 3:42 to 6:16 in the commercial recording), containing a dynamic and textural crescendo, resulting in a sudden increase in dynamics and instrumental layering at the climax; this was found previously to be the section in the music most effective at eliciting chills in listeners. This excerpt was chosen as an experimental stimulus as the crescendo and underlying structure were suited to engage auditory looming processes prior to psychoacoustic manipulations; manipulations of loudness were hypothesized to directly emphasize or diminish this capacity of the music.

In contrast, the control stimulus selected was from “Ancestral,” by Steven Wilson, a progressive rock piece with electronic, ambient textures and English lyrics. The excerpt (Ancestral) taken from this piece was 104 seconds in duration (from 3:31 to 5:15 in the commercial recording), and was characterized by the entrance of a virtuoso guitar solo overlaying a traditional hard rock instrumentation; this excerpt had minimal lyrical content. Like Glósóli, this excerpt from Ancestral has been found to be effective at eliciting chills across listeners, but was designated as a control stimulus because the underlying musical structure (guitar solo) did not have an intuitive capacity to elicit auditory looming processes; therefore, Ancestral served as a control comparison to assess whether psychoacoustic manipulation effects could be generalized with regards to chills, or are sensitive to the underlying musical structure.

To create manipulated excerpts, stimuli were edited in terms of overall volume levels or frequency spectrum, using Logic Pro 8 software. For both pieces, a specific epoch of 8 seconds was targeted for manipulation, characterized as the moment of salient structural change linked to musical chills; this included either a sudden dynamic and textural change (Glósóli), or the entrance of a guitar solo (Ancestral). During these epochs, loudness or brightness was gradually increased or decreased, in line with auditory looming processes (see Figure 1; see Figure 2 for an example time-series of loudness and brightness values across corresponding manipulations); following these manipulations within the 8-second epoch, the modified loudness or brightness levels were maintained for the remainder of the excerpt. For loudness, the volume of the excerpts was increased or decreased by 6 dBA. For brightness, frequencies above the 2,000 Hz threshold were amplified or reduced by 6 dBA, manipulating the high-to-low frequency ratio; this is a commonly utilized frequency threshold for brightness measurements in an auditory stimulus (Juslin, 2000; Lartillot et al., 2008; Laukka et al., 2005). To control for possible interactions between loudness and brightness, measures were taken to ensure that loudness manipulations did not significantly alter brightness levels, and vice-versa; this was achieved by manually moderating both loudness and brightness in the manipulation procedure, and quantifying the acoustic properties using MIRToolbox (Lartillot et al., 2008), to verify that the manipulation of one parameter did not significantly alter the other. To further validate the manipulations, a short pilot-test was carried out before the main experiment to assess whether the pieces remained realistic, and that listeners (N = 10) could hear the manipulations; all manipulations were judged to be reasonably realistic, and were noticeable, delivering the appropriate experiential balance for the experiment. Finally, a feature-extraction procedure was carried out following the methods of Eerola (2011), to assess other descriptive relationships across psychoacoustic parameters and manipulations (see Table 1); RMS levels were associated with spectral roughness and flux, while brightness was linked to the spectral centroid and entropy. This highlighted that the loudness and brightness manipulations targeted distinct, separable psychoacoustic constructs.

Figure 1.
Visualization of the psychoacoustic manipulations procedure; loudness or brightness was gradually increased or decreased over an eight second epoch containing the main structural change and transition in the piece, and was then sustained for the remainder of the excerpt.

Figure 2.
Time-series of loudness (first row) and brightness (second row) values for Glósóli (left column) and Ancestral (right column). Loudness is presented across control and loudness manipulations, whilst brightness is presented across control and brightness manipulations. Smoothed trend lines are a result of fitting a generalized additive model to the time-series data.

Table 1.
Mean z-values for six psychoacoustic parameters across stimulus conditions.

Piece/condition RMS Brightness Centroid Entropy Roughness Flatness

Glósóli

Original −.13 .00 −.02 .001 −.20 −.11

High intensity .98 .00 −.02 .001 1.04 .82

Low intensity −.65 −.04 −.04 −.009 −.52 −.55

High brightness −.13 .19 .47 .06 −.05 −.08

Low brightness −.05 −.16 −.38 −.06 −.26 −.07

Ancestral

Original −.11 −.004 −.03 .004 −.22 −.11

High intensity 1.07 −.004 −.03 .004 1.19 1.05

Low intensity −.68 −.007 −.03 −.003 −.58 −.67

High brightness −.14 .55 .68 .07 −.11 −.09

Low brightness −.12 −.54 −.58 −.08 −.26 −.16

Notes. RMS = root mean square.

Importantly, participants were unfamiliar with the stimuli upon first listen (Glósóli: mean = 1.72, SD = 1.15; Ancestral: mean = 1.72, SD = 1.06; possible range: 1–5). An essential aspect of stimulus selection was the control of lyrics or linguistic content in the two excerpts; Glósóli contained exclusively Icelandic lyrics, which none of the English-speaking participants understood, and while Ancestral contained English lyrics, the content was highly minimal and ambiguous in affective connotation, containing only two short lines: You can try if you want to and come back if you want to. Consequently, there was a high level of control over confounding effects derived from lyrics, an important consideration for the current approach.

Chills Measurement

To capture experiences of chills while listening, participants reported the onset of a perceived chills response by pushing down a button (the space bar on the experiment laptop), with the frequency of these self-reports serving as the primary dependent variable. Additionally, participants could estimate the duration of chills experiences by releasing the button to report the offset of the response, although participants were notified that should this task detract from the emotional experience, they should focus on the onsets of chills; this precaution was taken to keep the experiment simple, as performing tasks while listening to music may influence emotional and psychophysiological data recorded (Jäncke et al., 2018). To further validate chills reports, skin conductance data were collected from participants. This measure was preferred for three reasons: Firstly, skin conductance measurements are non-intrusive, which is important when attempting to capture and preserve strong emotional experiences in experimental settings; secondly, skin conductance has been utilized in existing research as a reliable indicator of chills experiences with music (Craig, 2005), and as a way of validating self-reports of the response (Benedek & Kaernbach, 2011; Grewe et al., 2007); and thirdly, several studies have shown that pressing a button during an experiment or task does not systematically produce skin conductance responses or increases in skin conductance level (Guhn et al., 2007; Rickard, 2004); however, to further accommodate this, participants were explicitly instructed to not move the hand that the electrodes were attached to. In the current paradigm, if a button press was not accompanied by significant increases in skin conductance levels compared with a baseline period, the report was omitted from the analysis. Skin conductance data were collected using two electrodes (Ag/AgCL), attached to the distal phalanx of the index and middle fingers of the non-dominant hand, with the NeXus-10 MKII and BioTrace software.

The intensity of chills was explored as a secondary dependent variable, utilizing average skin conductance amplitudes across chills reports, and the average duration of chills reports.

Procedure

Participants were tested separately, and assigned to one of the two experimental groups; the experiment was administered via a laptop, using OpenSesame software. Participants first familiarized themselves with the experimental procedure, tasks and data management statements provided via an information form; informed consent was obtained through participants completing a checklist of statements and providing confirmation that they agreed to take part and understood the procedure. Before the experiment, participants completed basic demographic questions, while electrodes were attached to the non-dominant hand to collect skin-conductance data. As an important measure, participants were played a short excerpt of music at a level of amplitude to match the maximum amplitude in the experiment; participant and experimenter confirmed the most comfortable maximum volume to listen to this excerpt, to serve as a loudness threshold for the remainder of the experiment. This served an important safety function, but also controlled for individual differences in loudness experience, by establishing a comparable perceptual threshold for all participants. This was crucial, as while the acoustic intensity of the stimuli may have differed by establishing individual loudness thresholds, it was deemed unlikely that these differences would affect chills experiences in the same way as notable perceptual differences in loudness, given the idiosyncratic embodied experience of the phenomenon (Bannister, 2018); consequently, it was judged more essential to make consistent the perceptual experience of the stimuli across participants. Stimuli were delivered through headphones.

In the first phase of the experiment, participants listened to two excerpts, pressing a button whenever chills were experienced, and completing self-reports after each stimulus. Then, the BIS/BAS scales were completed. A further two excerpts were then presented to participants, with identical tasks and procedure to the first phase. Following this, the IRI instrument was completed, before participants listened to the final excerpt of the experiment. Both the BIS/BAS and IRI instruments were placed within the experimental procedure to act as a break from music listening, and as a distractor task. This was to limit repetition or habituation effects and minimize fatigue, all important factors to consider when investigating strong emotional experiences. Furthermore, these active psychometric tasks were preferred over extended baseline periods of silence in between pieces, as a more explicit separation of emotional experiences across the listening conditions, and to control for changes in affective state possibly induced through processes including mind-wandering and visual imagery (Taruffi & Küssner, 2019).

The experiment lasted approximately 25 minutes, and was approved by the departmental Ethics committee; all data were anonymized throughout the process. Upon completion, participants could join an experiment raffle, with a chance of winning one of three £20 Amazon gift vouchers.

Data Analysis

All data were processed fully or partially in the R environment (https://cran.r-project.org). For the BIS/BAS and IRI instruments, data were aggregated in accordance with distinct factor structures identified in previous research (Carver & White, 1994; Davis, 1983). Skin-conductance data were pre-processed, decomposed into tonic and phasic components, and analyzed in accordance with methods presented by Bannister and Eerola (2018), with the use of the Ledalab package developed for MATLAB, and the continuous decomposition analysis procedure (Benedek & Kaernbach, 2010). In this instance, the phasic skin conductance response (SCR) was utilized as an accurate, high resolution indicator of event-related activity.

In the current analysis, SCR was used as a validation measure for the button presses provided by participants while listening. Following previous research, it was determined that a reliable self-report of chills should be accompanied by observable change in SCR (Grewe et al., 2007). For this reason, an epoch comparison was carried out on the SCR data, such that if the SCR signal was significantly higher in the 4 seconds following a button press compared with a baseline period, then the chills report was retained for analysis; in this case, baseline periods were defined as a random 4-second sample, taken from the 20-second epoch preceding the onset of the specific stimulus being analyzed. As these comparisons (t-tests between means of the baseline period and 4-second epoch after each button press) were carried out for each of the button presses (N = 329), the statistical significance threshold was set to the lower value of p < .001, and significance values were Bonferroni corrected through multiplying each p-value by the total number of comparisons; this was to conservatively accommodate for numerous comparisons and strengthen the analysis. No button presses were retained that were registered before the point of psychoacoustic manipulation in the excerpts; this was to assess chills responses derived from specific manipulations of the stimuli. As a result, filtered button presses (N = 129) served as a primary dependent variable representing the frequency of chills experienced; this filtering process, and exclusion of 200 button presses, further supports previous work showing no systematic effect of button presses on peaks in the skin conductance signal.

Additionally, average amplitudes of the SCR signal were calculated for the 4-second epoch following each validated button press, as an indicator of chills intensity; a 4-second window was selected in accordance with the documented lag structure of skin conductance, with a roughly 2-to 3-second response lag following an event or reported experience (Boucsein, 2012). These data were also normalized (z-scores) within participants to account for differences in response sensitivity across individuals (Khalfa et al., 2002). As a further indicator of chills intensity, the duration of button presses was calculated and averaged within each listening condition; this is derived from a previous suggestion that chills of longer duration may indicate more intense emotional experiences (Craig, 2005).

Results

Frequency of Chills

To calculate the frequency of chills responses across listening conditions as the primary dependent variable, button presses were aggregated and assessed as count data. Due to technical difficulties, no skin conductance data were collected for six participants, and so their button press data were omitted from the analysis. The descriptive statistics for overall chills experiences are presented in Table 2. The maximum number of chills experienced by one participant was 16, whereas only two participants reported experiencing no chills; the mean number of chills reported across participants was 4.60 (SD = 3.94). While repeatedly listening to the same unfamiliar stimulus may introduce repetition effects, the data showed no systematic change in the frequency of chills reported over the course of the experiment (Glósóli: x² [2, 4] = 2.63, p = .26; Ancestral: x² [2, 4] = 0.85, p = .65).

Table 2.
Descriptive statistics of chills reports across conditions (note: no average SCR or duration was calculated for Glósóli high brightness, given only one chills report).

Piece/condition Total chills reported Average normalized SCR Average button press duration (s)

Glósóli

Original 11 .80 6.90

High intensity 29 .33 6.79

Low intensity 6 .21 6.66

High brightness 1 – –

Low brightness 22 .42 8.17

Ancestral

Original 5 .65 5.57

High intensity 10 .19 9.80

Low intensity 11 .21 4.33

High brightness 16 .65 5.83

Low brightness 18 .20 3.10

Notes. SCR = skin conductance response.

To analyze the count data, a general linear mixed-effects model was developed using the lme4 package in R software (Bates et al., 2015), with a Poisson distribution; this was fitted once for loudness-manipulation conditions, and once for brightness-manipulation conditions. In the model, the number of button presses served as the dependent variable, and listening conditions (original, high, and low loudness, or high and low brightness) were fitted as fixed effects; individual participants and the two pieces (Glósóli and Ancestral) were fitted as random effects. For hypothesis testing, likelihood ratios tests were carried out, which utilized an ANOVA to compare the fitted statistical model to a reduced model (no listening condition component); this method enabled the assessment of listening condition effects on the frequency of chills reports through general linear mixed-effects models. For effect size indicators of the fixed effects, marginal R² was calculated following methods from Nakagawa and Schielzith (2013), using the MuMin R package (Barton, 2018).

Results suggest that for loudness manipulations, there was a significant overall effect on chills frequency (x² [2, 7] = 24.43, p = .0001, R² = .15). Further analysis found that increased loudness resulted in more chills for Glósóli (ß = 1.09, SE = .49, z = 2.21, p = .02), which elicited more chills in listeners than any other conditions across the two pieces; decreased loudness reduced the frequency of chills for Glósóli, but this was not significant (ß = −0.60, SE = .50, z = −1.19, p = .23). Interestingly, loudness manipulations had no effect on the control stimulus, Ancestral (increased loudness: ß = −0.09, SE = .43, z = −0.21, p = .82; decreased loudness: ß = 0.12, SE = .54, z = 0.22, p = .82).

For brightness manipulations, there was an overall effect of manipulation on the frequency of chills (x² [2, 7] = 31.85, p < .0001, R² = .34). More specifically, increases in brightness reduced chills across listeners for Glósóli (ß = −2.39, SE = 1.04, z = −2.29, p = .02); decreases in brightness increased the frequency of chills reports, although this was a non-significant trend (ß = 0.89, SE = .56, z = 1.58, p = .11). These effects are in opposition to those of loudness; high loudness increased chills frequency for Glósóli, whereas high brightness decreased chills frequency. Again, no effects were found for any brightness manipulations in the Ancestral piece (increased brightness: ß = 0.57, SE = .58, z = 0.99, p = .31; decreased brightness: ß = 0.49, SE = .38, z = 1.28, p = .19). The frequency of chills across pieces and conditions is visualized in Figure 3.

Figure 3.
Differences in chills frequency across brightness and loudness manipulations for Ancestral and Glósóli (* = p <.05). The x ² statistics reflect likelihood ratio tests between full and reduced general linear mixed effects models described in the Frequency of Chills results section.

To further account for individual differences in chills reports, an alternative analysis compared the proportion of experiences containing at least one chill report with those containing none, across listening conditions. Fitting the same models above with a binomial distribution (1 for chills, 0 for no chills), results were highly comparable (loudness: x² [2, 7] = 16.67, p = .005, R² = .13; brightness: x² [2, 7] = 15.85, p = .007, R² = .17), with more experiences containing chills reported to increased loudness for Glósóli (ß = 1.99, SE = .94, z = 2.10, p = .03), and less for increased brightness for Glósóli (ß = −2.42, SE = 1.22, z = −1.98, p = .04).

Intensity of Chills

As a secondary dependent variable, the intensity of chills was assessed through SCR amplitudes and chills duration. Starting with SCR amplitudes, as the data were numerical, a linear mixed-effects model was developed; SCR amplitudes were the dependent variable, listening conditions were fitted as fixed effects, while individual participants and the two pieces were fitted as random effects. Importantly, as only one chill was reported for the high brightness condition for Glósóli, this was omitted from the analysis (see Table 2). Results suggest that there was no discernible difference in phasic SCR amplitudes across the reported experiences of chills (x² [3, 11] = 11.66, p = .16); this appears to be the case across all manipulations and conditions for both experimental and control pieces.

The final aspect of chills analyzed was the average duration of reported experiences across listening conditions. However, as the high brightness condition of Glósóli resulted in only one chills report, no reliable value for duration was calculated, and therefore the condition was omitted from analysis (see Table 2). Like SCR amplitude, duration data were numerical and assessed with a linear mixed-effects models; duration was the dependent variable, listening conditions were fixed effects, and individual participants and the two pieces were fitted as random effects. The mixed effects model highlighted no notable differences in chills duration across the different manipulations in the two pieces (x² [3, 11] = 9.43, p = .30).

To summarize results, the psychoacoustic manipulations had significant effects on the frequency of chills reports; however, through the currently used indices of chills intensity, finding suggest that chills did not vary in their intensity across psychoacoustic manipulation conditions.

Self-Report Data

For each listening condition, participants rated their subjective feelings and bodily reactions using self-report scales. Firstly, data were analyzed separately across loudness manipulations and brightness manipulations; secondly, data were compared across experiences, including chills and experiences in which no chills were reported.

In terms of subjective feeling, participants rated their enjoyment of the music, alongside 10 emotional descriptors. Notably, while no effects were found for brightness manipulations on subjective feeling, loudness manipulations had a significant effect on enjoyment ratings (x² [3, 8] = 33.12, p < .001, R² = .19), with increased loudness for Ancestral resulting in reduced reports of enjoyment (ß = −1.55, SE = 0.35, z = −4.38, p < .001). Descriptively, experiences across both Glósóli and Ancestral were similar; both were rated as enjoyable, moving, and energetic; nominal, non-significant trends include experiences with Ancestral being rated as more sad and nostalgic than Glósóli, and feelings of tenderness being more pronounced during Glósóli. When comparing chills and no chills experiences, no differences were found across subjective feelings, suggesting that the overall assessment of the listening experience was consistent, and relatively independent of whether chills were experienced.

Regarding bodily activity, in several cases a response was reported too rarely to perform meaningful statistical analyses; however, in cases where generalized linear mixed-effects models were applicable, no differences were found across loudness or brightness manipulations, or across chills and no chills experiences. The most common bodily responses reported across both Glósóli and Ancestral were feelings of coldness and smiling.

The descriptive statistics and likelihood ratio test statistics pertaining to self-report data are available in the supplementary material (available online).

Empathy and Behavioral Activation/Inhibition

To assess how individual differences may mediate the experience of chills in response to music, trait empathy measurements derived from the IRI were assessed in relation to the frequency of chills reports. As chills frequency data did not follow a Gaussian distribution, spearman rank correlations were carried out to assess relationships with trait empathy; results suggest a small, non-significant relationship between chills frequency and trait empathy (r_s = .20, p = .13); no significant relationships were found for subscales of trait empathy. In addition, behavioral activation (BAS) and inhibition (BIS) characteristics of the listeners was assessed; with the same method, no relationship was found between BAS and chills frequency (r_s = .04, p = .40) or BIS and chills (r_s = −.13, p = .79). As previous research suggested that reward sensitivity was linked to chills (Mori & Iwanaga, 2015), a further analysis was carried out for the sub-facets of BAS; however, no significant relationships were evident, notably between reward sensitivity and frequency of chills reports (r_s = −.19, p = .87). Although it is possible that embedding these psychometric instruments within the experimental procedure influenced chills experiences, no clear difference in chills frequency was found across pre-BIS/BAS, post-BIS/BAS, and post-IRI phases of the experiment (x² [2, 4] = 4.74, p = .09).

Discussion

Existing work has suggested a relationship between increased loudness and the elicitation of chills (Grewe et al., 2007; Panksepp, 1995; Sloboda, 1991). In addition, higher levels of spectral brightness have recently been correlated with continuous ratings of chills intensity (Bannister & Eerola, 2018). These parameters have been explored in studies on emotional expression in music (Eerola, Ferrer & Alluri, 2012; Hailstone et al., 2009; Juslin, 2000; Lange & Frieler, 2018; Laukka et al., 2013), but less so in the context of induced emotion (Egermann et al., 2015; Gingras et al., 2014; Gomez & Danuser, 2007). From a theoretical perspective, it is intuitive to approach loudness increases, and their association with chills, in terms of vigilance theories linked to musical expectations (Huron, 2006); violations of expectations by music, a maladaptive outcome, may tap into the threat signalling fight-or-flight functionality of goosebumps (Darwin, 1872). More specifically, the link between chills, loudness, and crescendos may be explained by auditory looming (Ghazanfar et al., 2002; Neuhoff, 1998, 2001), where a sound increasing in loudness is perceived as an approaching object, and something that demands heightened attention and vigilance given the possibility of threat or danger. A similar process may underlie recent associations between increased brightness and musical chills, although this has yet to be verified in auditory looming literature.

The current experiment attempted to causally test theoretical claims made by Huron (2006) regarding musical chills, and the possible role of auditory looming as a mechanism of vigilance. Results suggest that the frequency of chills experiences was significantly affected by both loudness and brightness manipulations. Chills were elicited significantly more frequently when loudness was increased in the experimental stimulus (Glósóli), but not for the control stimulus (Ancestral); an opposite overall effect was found for brightness, with chills elicited less frequently when brightness was increased in Glósóli, while again no changes were found for Ancestral. Across chills reports, no differences were found in terms of chills intensity, measured through skin conductance amplitudes accompanying chills, and chills duration.

Interestingly, when comparing experiences across listening conditions, and comparing chills and no chills experiences, the overall emotional responses were similar; only one difference was found, with increased loudness in Ancestral resulting in decreases in reported enjoyment, although no clear reduction in chills frequency was apparent. This implies that the psychoacoustic manipulations, or presence of chills responses, did not affect the broader emotional experience of listening; this is in contrast with previous work suggesting that musical chills resulted in more moving and intense experiences (Bannister & Eerola, 2018), although it must be noted that this previous work followed a notably different study design. Therefore, it may be interpreted that in numerous cases across the experiment, chills were driven by low-level processes resulting in arousal increases that occasionally may be decoupled from experiences of pleasure or other affective responses. Further research will be required to explore this, but the possibility is of great importance for ongoing work attempting to understand whether the chills construct reflects moments of peak pleasure, or a collection of psychologically distinct responses (Bannister, 2019; Levinson, 2006; Maruskin et al., 2012; Panksepp, 1995; Pelowski et al., 2017).

Regarding individual differences, no relationships were found between the number of chills experienced and behavioral activation, behavioral inhibition, or trait empathy. Overall, the results offer partial support for the vigilance hypothesis of musical chills, but what follows is a discussion aimed at addressing why increasing loudness and brightness may not straightforwardly affect chills experiences, and at highlighting the important interactions between lower-level features, and the underlying musical structure.

Loudness Manipulations

In the current experiment, by increasing loudness at the onset of a previously identified “chills section” in music (Bannister & Eerola, 2018), the frequency of chills reported by participants was significantly increased for the experimental stimulus, Glósóli. However, when loudness was increased for the control stimulus Ancestral, there was no significant increase in occurrences of chills across listeners. No effects were found for indicators of chills intensity (skin conductance amplitude, and chills duration). These results for Glósóli support the vigilance theory of chills, such that salient changes in music, which in this case tap into auditory looming processes, may elicit an unconscious, immediate vigilance response in the face of “threat,” eliciting the evolutionary threat-signalling function of goosebumps; it was shown here that the more pronounced the increase of loudness is for this piece, the more frequent a chills experience is reported. Notably, the reduction of loudness did not affect the chills response when compared with the control condition of Glósóli; one explanation for this is that the manipulation still represents a salient change, a moment that attracts attention and vigilance; this is in some way reflected in recent work that suggested the phasic skin conductance response was more pronounced in response to loudness decreases, as opposed to increases (Olsen & Stevens, 2013). Additionally, this finding may explain the similarity in skin conductance amplitudes across chills experiences in the current study.

An interesting consideration for loudness is the lack of effect found for Ancestral. To explore this, it is important to consider the underlying musical structure of the pieces, an essential factor for qualitatively designating an experimental and control stimulus in the current study. For Glósóli, the excerpt was characterized by a gradual build-up in textural density and dynamics, followed by a sudden dynamic increase and textural change. Notably, this piece was already suited to elicit chills via processes of fear, attention, and vigilance, and loudness manipulations may have directly emphasized or diminished this. On the other hand, Ancestral can be characterized by a standard rock band setup, starting with sparse high vocal melodies, transitioning into a virtuosic guitar solo that remains, for most of the excerpt, as the salient instrument and melody. Interestingly, while this excerpt is also effective at eliciting chills, the introduction of a guitar solo is not easily explained by vigilance theories, suggesting that the piece may be linked to chills through alternative psychological processes. Therefore, the underlying structure of Ancestral does not seem well equipped to elicit chills linked to vigilance, and there may be a disjunction between this structure and the loudness manipulations aimed at activating and emphasizing these processes. The interactions between lower-level and higher-level features of music is an area of notable complexity (McAdams et al., 2004), and indeed, similar conclusions have been made with regards to musical chills, suggesting that while certain moments in music may be effective at eliciting the response, it is likely that these moments are empowered by their relationships with preceding structural developments in the piece (Bannister & Eerola, 2018).

Brightness Manipulations

In comparison to loudness, opposite effects were found for manipulations of brightness, such that increases in brightness resulted in less frequent chills experiences for Glósóli. This is perhaps surprising, considering previous correlations between brightness and musical chills (Bannister & Eerola, 2018), although it must be noted that these relationships have been documented less consistently compared with loudness. However, while increases in brightness may elicit auditory looming processes, such that higher frequencies imply closer proximity, this has currently not been evidenced in auditory looming research, and could be a subtler effect than the dynamic increases produced by approaching entities or objects. Like loudness manipulations, no effects of brightness were found for Ancestral.

Given previous correlations between brightness and musical chills, differing trends of effect between the two pieces, and the seeming disconnection between brightness and vigilance processes underlying chills, an alternative explanation may be required, namely in the form of linking chills to social processing. Importantly, while Huron (2006) provides an intuitive, testable theory of musical chills, it is not suited to explain the wide variety of musical features linked to the phenomenon (Bannister, 2018), nor aesthetic chills with film and poetry (Wassiliwizky, Koelsch et al., 2017; Wassiliwizky et al., 2015). An alternative theory is that chills are elicited by social bonding and empathy processes (Panksepp, 1995); an example of this is the kama muta experience, in which strong, moving emotional responses accompanied by goosebumps are elicited by intensified communal sharing relations (Fiske et al., 2017). Although conjecture, it may be that social bonding taps into thermoregulatory functions of goosebumps, as opposed to threat-signalling functions; existing research suggests overlap between thermoregulatory and social processes in the brain (Inagaki & Eisenberger, 2013; Panksepp, 1998), which may have occurred through evolution by repeatedly sharing body heat closely with other conspecifics to survive in cold conditions (IJzerman et al., 2015).

In the social bonding context, brightness might imply closer proximity between oneself and another source or entity; but rather than activating vigilance processes (Kovacevich & Huron, 2018), this proximity may be social, a form of intimate or communal closeness between oneself and another. A striking example of this is found in the related ASMR phenomenon, described as a relaxing, tingling sensation originating in the scalp and spreading down the neck, spine and arms (Barratt & Davis, 2015). Common elicitors of ASMR include people speaking quietly and closely into a sensitive microphone or performing actions that produce “crisp” sounds such as tapping fingernails or crunching metallic foil (Barratt & Davis, 2015), whispering (Andersen, 2015), and sibilant sounds in speech such as “s”, “sh,” and “sk” (Kovacevich & Huron, 2018). These sounds linked to ASMR may be characterized by higher levels of brightness, possibly portraying a certain social intimacy, proximity and connection, although this has yet to be empirically tested. However, by assessing brightness and musical chills from this perspective, a scenario may be occurring that is comparable to loudness: It is possible that Ancestral is better suited to elicit chills through social bonding processes, especially when contextualizing the guitar solo as a “super-expressive” sound (Juslin, 2001), reflecting expressions of the human voice that occupy a privileged position of communicative salience (Belin et al., 2000). Furthermore, in previous studies utilizing Ancestral (Bannister, 2019; Bannister & Eerola, 2018), the piece has been described as eliciting “being moved” responses, an emotional concept linked to prosocial cues and social bonding (Menninghaus et al., 2015). In contrast, Glósóli may not be well suited to engaging social processes. Instead, if Glósóli is predisposed to elicit chills through auditory looming and vigilance mechanisms, then increases of brightness may diminish these processes by enhancing auditory characteristics linked to social proximity and intimacy, conflicting with vigilance mechanisms and the underlying musical structure. The current data, with increased loudness enhancing chills and increased brightness suppressing chills for Glósóli, suggest that this could be the case. However, this is presently speculative, and it remains important to note that no clear, significant differences were found for any manipulations in Ancestral. Consequently, there may be alternative parsimonious hypotheses regarding this data, including effects of aesthetic judgments of music in chills experiences (Juslin, 2013), such that psychoacoustic manipulations may be perceived to either compliment a piece, or instead seem out of place, depending on the style and genre of the music. It will be essential to replicate and extend the current paradigm to explore these possibilities.

Limitations and Conclusions

There are several limitations worth highlighting in the current study. Firstly, given the novel approach of manipulating psychoacoustic parameters in real musical stimuli to affect induced emotions, a level of control was difficult to extend across all features in the auditory stimuli such as roughness and flux; as a result, it is not trivial to confirm the effects of loudness in complete isolation. However, the relationships between loudness, roughness and flux might reflect ecologically valid perceptual experiences, and these parameters contribute in similar ways to valence and arousal expressions in some music (Eerola, 2011); also, in the case of loudness and roughness, both aspects have been linked to chills (Grewe et al., 2007; Nagel et al., 2008), and both may be applicable to ideas of vigilance mechanisms (Arnal et al., 2015; Ghazanfar et al., 2002; Huron, 2006). Importantly, the study demonstrated that loudness and brightness could be manipulated independently of each other, suggesting that separable perceptual composites were assessed in relation to musical chills; furthermore, inferences can still be made from the current results regarding the direct effects of loudness and brightness on musical chills. A related point to consider is how effects of psychoacoustic parameters may differ across instruments and in vocal aspects such as lyrics. However, while there may be an issue of domain specificity across auditory and linguistic modes and psychoacoustic manipulation effects, the lyrical content in the stimuli was minimal, offering strong control over possible domain-specific interactions. It will be important for future research to consider these variations between psychoacoustics and emotion that may be mediated by different musical features, instruments, and structures.

Secondly, as only two pieces of music were utilized in the current experiment, further research will be required to generalize the findings across a broader selection of structural and psychoacoustic contexts. It remains a complex task to elicit and manipulate chills in experimental settings, but some possible progressions could be to work with high quality MIDI excerpts, or originally compose musical stimuli, which would allow for greater control of manipulations; a necessary caveat is whether artificial or originally composed music would be sufficient to induce strong experiences such as chills. Therefore, it was deemed important to develop the current, novel paradigm using unfamiliar real pieces of music extensively validated as effective elicitors of chills (Bannister, 2018; Bannister & Eerola, 2018), as a platform for future more generalizable investigations; from here, it will be crucial to replicate and extend the existing data across diverse musical stimuli, psychoacoustic manipulations and listener populations.

Finally, there are procedural aspects of the current paradigm that could be refined. For example, providing subjective feeling scales multiple times in the same order after each stimulus may facilitate automated behaviors from participants. Familiarity is also important to control when causally manipulating aspects of musical stimuli; however, there is no consistent picture on how familiarity affects musical chills (Benedek & Kaernbach, 2011; Laeng et al., 2016; Nusbaum et al., 2014), and this was largely controlled in the current experiment using unfamiliar stimuli and pseudorandomized presentation orders. Finally, administering psychometric instruments during breaks in the experimental procedure may influence the subsequent emotional experiences with music; however, no evidence for this was found in the current work. Further causal research on chills should consider developing improved aspects of the experimental procedure to accurately capture and preserve strong emotional experiences that best resemble those encountered in everyday life.

To conclude, the current study presents novel evidence for processes of auditory looming and vigilance underlying the musical chills response, providing a first causal test of any theory of the phenomenon. However, while there is now causal evidence for vigilance processes underlying chills, shown through increased loudness resulting in a higher occurrence of chills across listeners, this was only the case for one excerpt in the experiment, suggesting two interrelated key points: Firstly, there are important interactions between psychoacoustic, lower-level parameters and higher-level structural features (such as crescendos) that need to be considered and explored; and secondly, that fear and vigilance processes may only underlie a certain proportion of musical chills responses, with others possibly linked to social and empathic processes, which further interact with the psychoacoustic and structural intra-musical relationships. Future work should attempt to develop more sophisticated approaches to psychoacoustic manipulations in emotion induction paradigms; additionally, further research should delineate the theoretical discourse surrounding chills, and investigate the different potential routes of induction, and how these relate to different musical and psychoacoustic features, and emotional arousal more broadly.

Supplemental material

Supplemental Material, supplementary_materials - A Vigilance Explanation of Musical Chills? Effects of Loudness and Brightness Manipulations

Supplemental Material, supplementary_materials for A Vigilance Explanation of Musical Chills? Effects of Loudness and Brightness Manipulations by Scott Bannister in Music & Science

Piece/condition	RMS	Brightness	Centroid	Entropy	Roughness	Flatness
Glósóli
Original	−.13	.00	−.02	.001	−.20	−.11
High intensity	.98	.00	−.02	.001	1.04	.82
Low intensity	−.65	−.04	−.04	−.009	−.52	−.55
High brightness	−.13	.19	.47	.06	−.05	−.08
Low brightness	−.05	−.16	−.38	−.06	−.26	−.07
Ancestral
Original	−.11	−.004	−.03	.004	−.22	−.11
High intensity	1.07	−.004	−.03	.004	1.19	1.05
Low intensity	−.68	−.007	−.03	−.003	−.58	−.67
High brightness	−.14	.55	.68	.07	−.11	−.09
Low brightness	−.12	−.54	−.58	−.08	−.26	−.16

Piece/condition	Total chills reported	Average normalized SCR	Average button press duration (s)
Glósóli
Original	11	.80	6.90
High intensity	29	.33	6.79
Low intensity	6	.21	6.66
High brightness	1	–	–
Low brightness	22	.42	8.17
Ancestral
Original	5	.65	5.57
High intensity	10	.19	9.80
Low intensity	11	.21	4.33
High brightness	16	.65	5.83
Low brightness	18	.20	3.10

Footnotes

Author contribution

The author conceptualised and designed the experiment, carried out data collection, performed data analysis, wrote the original draft of the manuscript, revised the manuscript, and managed the overall project.

Data availability statement

The raw self-report data, pre-validated chills button press data, and normalized, detrended phasic skin conductance time-series data are submitted alongside the manuscript, in .xlsx format.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This research was funded by an Arts and Humanities Research Council (https://ahrc.ukri.org) doctoral studentship awarded to SB (Grant number AH/L503927/1), through the Northern Bridge Doctoral Training Partnership (). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

ORCID iD

Scott Bannister

Action editor

Daniela Sammler, Max Planck Institute for Human Cognitive and Brain Sciences.

Peer review

Matthew E. Sachs, Columbia University, Department of Psychology.

Manuela Marin, University of Innsbruck, Department of Psychology.

One anonymous Reviewer.

Supplemental material

Supplemental material for this article is available online.

References

Algoe

Haidt

(2009). Witnessing excellence in action: The ‘other-praising’ emotions of elevation, gratitude, and admiration. The Journal of Positive Psychology, 4(2), 105–127.

Andersen

(2015). Now you’ve got the shiveries: Affect, intimacy, and the ASMR whisper community. Television and New Media, 16(8), 683–700. http://doi.org/10.1177/1527476414556184.

Arnal

Flinker

Kleinschmidt

Giraud

A.-L.

Poeppel

(2015). Human screams occupy a privileged niche in the communication soundscape. Current Biology, 25(15), 2051–2056.

Bach

Neuhoff

Perrig

Seifritz

(2009). Looming sounds as warning signals: The function of motion cues. International Journal of Psychophysiology, 74(1), 28–33.

Bach

Schächinger

Neuhoff

Esposito

Di Salle

Lehmann

Herdener

Scheffler

Seifritz

(2008). Rising sound intensity: An intrinsic warning cue activating the amygdala. Cerebral Cortex, 18(1), 145–150.

Balkwill

Thompson

(1999). A cross-cultural investigation of the perception of emotion in music: Psychophysical and cultural cues. Music Perception, 17(1), 43–64.

Balkwill

Thompson

Matsunaga

(2004). Recognition of emotion in Japanese, Western, and Hindustani music by Japanese listeners. Japanese Psychological Research, 46(4), 337–349.

Bannister

(2018). A survey into the experience of musically induced chills: Emotions, situations and music. Psychology of Music, 48(2), 297–314. Doi: 10.1177/0305735618798024.

Bannister

(2019). Distinct varieties of aesthetic chills in response to multimedia. PLoS ONE, 14(11), e0224974. Doi: 10.1371/journal.pone.0224974.

10.

Bannister

Eerola

(2018). Suppressing the chills: Effects of musical manipulation on the chills response. Frontiers in Psychology, 9, 2046. Doi: 10.3389/fpsyg.2018.02046.

11.

Barratt

Davis

(2015). Autonomous sensory meridian response (ASMR): A flow-like mental state. PeerJ, 3, e851.

12.

Barton

(2018). Mumin: Multi-model inference (R Package, version 1.42.1).

13.

Bates

Mächler

Bolker

Walker

(2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67(1), 1–48.

14.

Belin

Zatorre

Lafaille

Ahad

Pike

(2000). Voice-selective areas in human auditory cortex. Nature, 403(6767), 309–312.

15.

Benedek

Kaernbach

(2010). A continuous measure of phasic electrodermal activity. Journal of Neuroscience Methods, 190, 80–91.

16.

Benedek

Kaernbach

(2011). Physiological correlates and emotional specificity of human piloerection. Biological Psychology, 86(3), 320–329.

17.

Blood

Zatorre

(2001). Intensely pleasurable responses to music correlate with activity in brain regions implicated in reward and emotion. Proceedings of the National Academy of Sciences, 98(20), 11818–11823.

18.

Boucsein

(2012). Electrodermal activity. Plenum Press.

19.

Carver

White

(1994). Behavioural inhibition, behavioural activation, and affective responses to impending reward and punishment: The BIS/BAS scales. Journal of Personality and Social Psychology, 67(2), 319–333.

20.

Colver

El-Alayli

(2016). Getting aesthetic chills from music: The connection between openness to experience and frisson. Psychology of Music, 44(3), 413–427. Doi: 10.1177/0305735615572358.

21.

Coutinho

Cangelosi

(2011). Musical emotions: Predicting second-by-second subjective feelings of emotion from low-level psychoacoustic features and physiological measurements. Emotion, 11(4), 921–937.

22.

Craig

(2005). An exploratory study of physiological changes during “chills” induced by music. Musicae Scientiae, 9(2), 273–287.

23.

Darwin

(1872). The expression of emotions in man and animals. University of Chicago Press.

24.

Davis

(1983). Measuring individual differences in empathy: Evidence for a multidimensional approach. Journal of Personality and Social Psychology, 44(1), 113–126.

25.

Dean

Bailes

Schubert

(2011). Acoustic intensity causes perceived changes in arousal levels in music: An experimental investigation. PLoS ONE, 6(4), e18591.

26.

Eerola

(2011). Are the emotions expressed in music genre-specific? An audio-based evaluation of datasets spanning classical, film, pop and mixed genres. Journal of New Music Research, 40(4), 349–366. Doi: 10.1080/09298215.2011.602195.

27.

Eerola

Ferrer

Alluri

(2012). Timbre and affect dimensions: Evidence from affect and similarity ratings and acoustic correlates of isolated instrument sounds. Music Perception, 30(1), 49–70.

28.

Eerola

Vuoskoski

Kautiainen

(2016). Being moved by unfamiliar sad music is associated with high empathy. Frontiers in Psychology, 7, 1176.

29.

Egermann

Fernando

Chuen

McAdams

(2015). Music induces universal emotion-related psychophysiological responses: Comparing Canadian listeners to Congolese Pygmies. Frontiers in Psychology, 5, 1341.

30.

Egermann

Sutherland

Grewe

Nagel

Kopiez

Altenmüller

(2011). Does music listening in a social context alter experience? A physiological and psychological perspective on emotion. Musicae Scientiae, 15(3), 307–323.

31.

Fiske

Seibt

Schubert

(2017). The sudden devotion emotion: Kama Muta and the cultural practices whose function Is to evoke it. Emotion Review, 11(1), 74–86.

32.

Fletcher

Munson

(1933). Loudness, its definition, measurement and calculation. Bell System Technical Journal, 12(4), 377–430.

33.

Fritz

Jentschke

Gosselin

Sammler

Peretz

Turner

Friederici

Koelsch

(2009). Universal recognition of three basic emotions in music. Current Biology, 19(7), 573–576. http://doi.org/10.1016/j.cub.2009.02.058.

34.

Gabrielsson

Lindström

(2010). The role of structure in the musical expression of emotions. In Juslin

Sloboda

(Eds.), Handbook of music and emotion: Theory, research, applications (pp. 367–400). Oxford University Press.

35.

Ghazanfar

Neuhoff

Logothetis

(2002). Auditory looming perception in rhesus monkeys. Proceedings of the National Academy of Sciences, 99(24), 15755–15757.

36.

Gingras

Marin

Fitch

(2014). Beyond intensity: Spectral features effectively predict music-induced subjective arousal. The Quarterly Journal of Experimental Psychology, 67(7), 1428–1446.

37.

Gomez

Danuser

(2007). Relationships between musical structure and psychophysiological measures of emotion. Emotion, 7(2), 377–387.

38.

Grassi

Darwin

(2006). The subjective duration of ramped and damped sounds. Perception and Psychophysics, 68(8), 1382–1392.

39.

Grewe

Nagel

Kopiez

Altenmüller

(2007). Listening to music as a re-creative process: Physiological, psychological, and psychoacoustical correlates of chills and strong emotions. Music Perception, 24(3), 297–314.

40.

Groarke

Hogan

(2016). Enhancing wellbeing: An emerging model of the adaptive functions of music listening. Psychology of Music, 44(4), 769–791.

41.

Guhn

Hamm

Zentner

(2007). Physiological and musico-acoustic correlates of the chill response. Music Perception, 24(5), 473–484.

42.

Hailstone

Omar

Henley

Frost

Kenward

Warren

(2009). It’s not what you play, it’s how you play it: Timbre affects perception of emotion in music. Quarterly Journal of Experimental Psychology, 62(11), 2141–2155.

43.

Halpern

Blake

Hillenbrand

(1986). Psychoacoustics of a chilling sound. Perception and Psychophysics, 39(2), 77–80.

44.

Harrison

Loui

(2014). Thrills, chills, frissons, and skin orgasms: Toward an integrative model of transcendent psychophysiological experiences in music. Frontiers in Psychology, 5, 790.

45.

Huron

(2006). Sweet anticipation: Music and the psychology of expectation. The MIT Press.

46.

IJzerman

Coan

Wagemans

Missler

Beest

Lindenberg

Tops

(2015). A theory of social thermoregulation in human primates. Frontiers in Psychology, 6, 464.

47.

Ilie

Thompson

(2006). A comparison of acoustic cues in music and speech for three dimensions of affect. Music Perception, 23(4), 319–330.

48.

Inagaki

Eisenberger

(2013). Shared neural mechanisms underlying social warmth and physical warmth. Psychological Science, 24(11), 2272–2280.

49.

Jäncke

Leipold

Burkhard

(2018). The neural underpinnings of music listening under different attention conditions. Neuroreport, 29, 594–604.

50.

Jorm

Christensen

Henderson

Jacomb

Korten

Rodgers

(1998). Using the BIS/BAS scales to measure behavioural inhibition and behavioural activation: Factor structure, validity and norms in a large community sample. Personality and Individual Differences, 26(1), 49–58.

51.

Juslin

(2000). Cue utilisation in the communication of emotion in music performance. Journal of Experimental Psychology: Human Perception and Performance, 26(6), 1797–1813.

52.

Juslin

(2001). Communicating emotion in music performance: A review and a theoretical framework. In Juslin

Sloboda

(Eds.), Music and emotion: Theory and research (pp. 309–337). Oxford University Press.

53.

Juslin

(2013). From everyday emotions to aesthetic emotions: Towards a unified theory of musical emotions. Physics of Life Reviews, 10(3), 235–266.

54.

Juslin

Laukka

(2003). Communication of emotions in vocal expression and music performance: Different channels, same code? Psychological Bulletin, 129(5), 770–814.

55.

Juslin

Timmers

(2010). Expression and communication of emotion in music performance. In Juslin

Sloboda

(Eds.), Handbook of music and emotion: Theory, research, applications (pp. 453–489). Oxford University Press.

56.

Juslin

Västfjäll

(2008). Emotional responses to music: The need to consider underlying mechanisms. Behavioural and Brain Sciences, 31(5), 559–575. Doi: 10.1017/S0140525X08005293.

57.

Khalfa

Peretz

Blondin

J.-P.

Manon

(2002). Event-related skin conductance responses to musical emotions in humans. Neuroscience Letters, 328(2), 145–149.

58.

Kovacevich

Huron

(2018). Two studies of autonomous sensory meridian response (ASMR): The relationship between ASMR and music-induced frisson. Empirical Musicology Review, 13(1-2), 39–63.

59.

Laeng

Eidet

Sulutvedt

Panksepp

(2016). Music chills: The eye pupil as a mirror to music’s soul. Consciousness and Cognition, 44, 161–178.

60.

Lange

Frieler

(2018). Challenges and opportunities of predicting musical emotions with perceptual and automatized features. Music Perception: An Interdisciplinary Journal, 36(2), 217–242. http://doi.org/10.1525/mp.2018.36.2.217.

61.

Lartillot

Toiviainen

Eerola

(2008). A Matlab toolbox for music information retrieval. In Preisach

Burkhardt

Schmidt-Thieme

Decker

(Eds.), Data analysis: Machine learning and applications (pp. 261–268). Springer.

62.

Laukka

Eerola

Thingujam

Yamasaki

Beller

(2013). Universal and culture-specific factors in the recognition and performance of musical affect expressions. Emotion, 13(3), 434–449.

63.

Laukka

Juslin

Bresin

(2005). A dimensional approach to vocal expression of emotion. Cognition and Emotion, 19(5), 633–653.

64.

Laukka

Quick

(2013). Emotional and motivational uses of music in sports and exercise: A questionnaire study among athletes. Psychology of Music, 41(2), 198–215. http://doi.org/10.1177/0305735611422507.

65.

Leman

Vermeulen

De Voogdt

Moelants

Lessafre

(2005). Prediction of musical affect using a combination of acoustic structural cues. Journal of New Music Research, 34(1), 39–67.

66.

Levinson

(2006). Contemplating art: Essays in aesthetics. Oxford University Press.

67.

Lippman

Greenwood

(2012). A song to remember: Emerging adults recall memorable music. Journal of Adolescent Research, 27(6), 751–774. http://doi.org/10.1177/0743558412447853.

68.

Lonsdale

North

(2011). Why do we listen to music? A uses and gratifications analysis. British Journal of Psychology, 102(1), 108–134. http://doi.org/10.1348/000712610X506831.

69.

Maruskin

Thrash

Elliot

(2012). The chills as a psychological construct: Content universe, factor structure, affective composition, elicitors, trait antecedents, and consequences. Journal of Personality and Social Psychology, 103(1), 135–157.

70.

McCrae

(2007). Aesthetic chills as a universal marker of openness to experience. Motivation and Emotion, 31(1), 5–11. Doi: 10.1007/s11031-007-9053-1.

71.

McAdams

(2013). Musical timbre perception. In Deutsch

(Ed.), The psychology of music (pp. 35–67). Academic Press.

72.

McAdams

Douglas

Vempala

(2017). Perception and modeling of affective qualities of musical instrument sounds across pitch registers. Frontiers in Psychology, 8, 153.

73.

McAdams

Vines

Vieillard

Smith

Reynolds

(2004). Influences of large-scale form on continuous ratings in response to a contemporary piece in a live concert setting. Music Perception, 22(2), 297–350.

74.

Melchers

Haas

Reuter

Bischoff

Montag

(2016). Similar personality patterns are associated with empathy in four different countries. Frontiers in Psychology, 7, 290.

75.

Menninghaus

Wagner

Hanich

Wassiliwizky

Kuehnast

Jacobsen

(2015). Towards a psychological construct of being moved. PLoS ONE, 10(6), e0128451.

76.

Meyer

(1956). Emotion and meaning in music. University of Chicago Press.

77.

Mori

Iwanaga

(2015). General reward sensitivity predicts intensity of music-evoked chills. Music Perception, 32(5), 484–492.

78.

Nagel

Kopiez

Grewe

Altenmüller

(2008). Psychoacoustical correlates of musically induced chills. Musicae Scientiae, 12, 101–113.

79.

Nakagawa

Schielzeth

(2013). A general and simple method for obtaining R2 from general linear mixed-effects models. Methods in Ecology and Evolution, 4(2), 133–142.

80.

Neuhoff

(1998). Perceptual bias for rising tones. Nature, 395(6698), 123–124.

81.

Neuhoff

(2001). An adaptive bias in the perception of looming auditory motion. Ecological Psychology, 13(2), 87–110.

82.

Neuhoff

Planisek

Seifritz

(2009). Adaptive sex differences in auditory motion perception: Looming sounds are special. Journal of Experimental Psychology: Human Perception and Performance, 35(1), 225–234.

83.

Nusbaum

Silvia

(2011). Shivers and timbres: Personality and the experience of chills from music. Social Psychological and Personality Science, 2(2), 199–204. Doi: 10.1177/1948550610386810.

84.

Nusbaum

Silvia

Beaty

Burgin

Hodges

Kwapil

(2014). Listening between the notes: Aesthetic chills in everyday music listening. Psychology of Aesthetics, Creativity, and the Arts, 8(1), 104–109.

85.

Olsen

Stevens

(2013). Psychophysiological response to acoustic intensity change in a musical chord. Journal of Psychophysiology, 27(1), 16–26.

86.

Panksepp

(1995). The emotional sources of “chills” induced by music. Music Perception, 13(2), 171–207.

87.

Panksepp

(1998). Affective neuroscience: The foundations of human and animal emotions. Oxford University Press.

88.

Pearce

Ruiz

Kapasi

Wiggins

Bhattacharya

(2010). Unsupervised statistical learning underpins computational, behavioural, and neural manifestations of musical expectation. NeuroImage, 50(1), 302–313.

89.

Pelowski

Markey

Forster

Gerger

Leder

(2017). Move me, astonish me…delight my eyes and brain: The Vienna Integrated Model of top-down and bottom-up processes in Art Perception (VIMAP) and corresponding affective, evaluative, and neurophysiological correlates. Physics of Life Reviews, 21, 80–125.

90.

Plomp

Levelt

(1965). Tonal consonance and critical bandwidth. Journal of the Acoustical Society of America, 38(4), 548–560.

91.

Rickard

(2004). Intense emotional responses to music: A test of the physiological arousal hypothesis. Psychology of Music, 32(4), 371–388.

92.

Ries

Schlauch

DiGiovanni

(2008). The role of temporal-masking patterns in the determination of subjective duration and loudness for ramped and damped sounds. Journal of the Acoustical Society of America, 124(6), 3772–3783.

93.

Salimpoor

Benovoy

Larcher

Dagher

Zatorre

(2011). Anatomically distinct dopamine release during anticipation and experience of peak emotion to music. Nature Neuroscience, 14(2), 257–262.

94.

Salimpoor

Benovoy

Longo

Cooperstock

Zatorre

(2009). The rewarding aspects of music listening are related to degree of emotional arousal. PLoS ONE, 4(10), e7487. Doi: 10.1371/journal.pone.0007487.

95.

Schäfer

Sedlmeier

Städtler

Huron

(2013). The psychological functions of music listening. Frontiers in Psychology, 4, 511. http://doi.org/10.3389/fpsyg.2013.00511.

96.

Schubert

(2004). Modelling perceived emotion with continuous musical features. Music Perception, 21(4), 561–585.

97.

Sloboda

(1991). Music structure and emotional response: Some empirical findings. Psychology of Music, 19(2), 110–120.

98.

Steinbeis

Koelsch

Sloboda

(2006). The role of harmonic expectancy violations in musical emotions: Evidence from subjective, physiological, and neural responses. Journal of Cognitive Neuroscience, 18(8), 1380–1393.

99.

Tajadura-Jiménez

Väljamäe

Asutay

Västfjäll

(2010). Embodied auditory perception: The emotional impact of approaching and receding sound sources. Emotion, 10(2), 216–229.

100.

Taruffi

Küssner

(2019). A review of music-evoked visual mental imagery: Conceptual issues, relation to emotion, and functional outcome. Psychomusicology: Music, Mind, and Brain, 29(2-3), 62–74.

101.

Wassiliwizky

Jacobsen

Heinrich

Schneiderbauer

Menninghaus

(2017). Tears falling on goosebumps: Co-occurrence of emotional lacrimation and emotional piloerection indicates a psychophysiological climax in emotional arousal. Frontiers in Psychology, 8, 41.

102.

Wassiliwizky

Koelsch

Wagner

Jacobsen

Menninghaus

(2017). The emotional power of poetry: Neural circuitry, psychophysiology and compositional principles. Social Cognitive and Affective Neuroscience, 12(8), 1229–1240.

103.

Wassiliwizky

Wagner

Jacobsen

Menninghaus

(2015). Art-elicited chills indicate states of being moved. Psychology of Aesthetics, Creativity, and the Arts, 9(4), 405–416.

104.

Zentner

Grandjean

Scherer

(2008). Emotions evoked by the sound of music: Characterization, classification, and measurement. Emotion, 8(4), 494–521.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.12 MB