Impact of Music on First Pain and Temporal Summation of Second Pain: A Psychophysical Pilot Study

Twenty-eight healthy volunteers participated in this study. To archive a power of 90% and a level of significance of 5%, 28 participants were required to detect an expected difference of 13 points (Guétin et al., 2012) between pairs, for an expected standard deviation equal to 20. One participant failed to complete the test session due to software failure. Hence, 27 participants (14 men, 13 women), aged between 18 and 30 (23.11 ± 3.03 years), were included in the analysis. Participants were recruited using ads posted throughout the campus of the Université de Bretagne Occidentale and on social media. None of them had previously participated to other pain experiments. Exclusion criteria were as follows: any chronic or acute disease, any psychiatric disorder, hearing impairments, current acute pain, and the consumption of any analgesic drugs or central nervous system agents for 24 hours before the experiment. All participants filled out a series of questionnaires to ensure the absence of symptoms of depression or anxiety and pain catastrophizing because those factors are frequently associated with dysfunctional descending pain inhibition. Questionnaires included the French versions of the Beck Depression Inventory (BDI) (Pichot & Lempérière, 1964) and the Pain Catastrophizing Scale (PCS-CF) (French et al., 2005). Participants above the scores of respectively 7 and 23 or greater were not considered for inclusion. State anxiety during the visit was also screened using the Spielberger’s State-Trait Anxiety Inventory (STAI) (Gauthier & Bouchard, 1993). All participants obtained a higher state score than the trait score, indicating that their anxiety level during this quiet test was inferior to their daily anxiety. All subjects were blinded to the hypothesis of the study. The study conformed to the ethical standards established in the Declaration of Helsinki, except for registration in a database. The experimental protocols were approved by the local ethics committee. All participants gave written informed consent prior to inclusion in the study.

Twenty-minute “U” sequences musical pieces were selected from Music Care (music-care.com) database. The “U-based” system corresponds to a therapeutic solution standardized for the management of chronic pain and developed by the company Music Care (Guétin et al., 2012). The “U” sequences are divided into three successive phases with gradual variations in tempo, instrumentation, frequencies, and volume. During the first ten minutes, there is a progressive reduction of the number of instruments (from 10–20 to 2–5), the beats per minute (from 90 to 50 bpm), the volume, and the frequencies. The changes in those musical parameters are assumed to induce a relaxation state. After this phase, the parameters stay constant during another ten minutes with a low tempo of 30 to 40 beats per minute and a maximum of three instruments. Finally, the music gradually increases during 10 minutes to a moderate tempo of 60 to 80 bpm associated with 5 to 10 instruments, to regain an ordinary state of consciousness. All sequences are instrumentally played by professional musicians and can be considered as soothing. Participants had to choose among a set of twenty-one musical genres (rock, classical, jazz, electro, etc.) the one that best suited them. This method ensures some homogeneity in the musical stimuli in terms of dynamic changes, tempo, and arousal while allowing each participant to choose his or her favorite musical genre. The pain threshold was fixed to a medium intensity on the allocate ruler. The self-selected music sequence was delivered through headphones with active noise reduction (Wireless headphone Model AH-NCW5OO, Denon Electronics, USA).

The thermal stimulation paradigm (see Figure 1) is based on an experimental pain model that allows measuring PP, TS, and DNIC in a single session divided into three successive bunches of 120 seconds (Tousignant-Laflamme et al., 2008). This model associates conditioned pain modulation (CPM) with contact-heat induced pain. CPM is a reliable psychophysical experimental measure of the diffuse noxious inhibitory control (DNIC) (Kennedy et al., 2016). CPM paradigms typically imply the combination of two types of noxious stimuli. A test-stimulus is applied before and after a conditioning stimulus. Difference between both test-stimulus is interpreted as CPM-effect that reflects DNIC activation. In this study, dynamic quantitative sensory testing (QST) with noxious heat pain tests was used as test-stimulus and to evoke first and second pain. A 12°C cold pressor test (CPT) was employed as the conditioning stimulus. Figure 1 summarizes the sequential paradigm including two tonic heat pain stimulations (P1 and P3) separated by a 120-second CPT (P2).

Two contact heat pain tests were performed for 120 seconds each on the thumb side of the forearm, corresponding to the C6 dermatome, by means of a 3-cm² thermode (NeuroSensory Analyser Model TSA II, Medoc Instruments, Israel). The stimulation was administered at a constant temperature during the entire tests. Regarding precision, thermode's temperature remained constant throughout the 120 seconds after a constant rise (0.3°C/s) from the baseline (32°C) to the individually predetermined temperatures. The temperature was generated continuously by the thermode. Participants were blinded to the temperature used. This procedure allows studying two facilitatory pain-related phenomena arising at specific times (Granot et al., 2006; Nielsen & Arendt-Nielsen, 1998; Price & Dubner, 1977). First pain perception is characterized by a peak effect at about 30 seconds after reaching fixed temperature and is reflected by a sharp and brief pain enhancement (referred to as peak of first pain –PP). PP is followed by TS of second pain reflected in progressive pain amplification after about 60 seconds of constant or repetitive noxious heat stimulation. The difference between pain ratings at 120 and 60 seconds is defined as TS amplitude. The contact heat pain tests ensure an intra-subject high reliably of the TS-effect within 1 hour (Kong et al., 2013). Moreover, tonic heat stimulations provide a better reliability for CPM-effect than phasic nociceptive stimuli (Lie et al., 2017).

A cold pressor test (CPT) was used as a conditioning stimulus to elicit a strong and prolonged pain sensation and to trigger DNIC. The conditioning stimulus consisted in the immersion of the left arm (up to the elbow) in circulating cold water maintained at 12°C for 120 seconds. All participants manage to keep the arm in the cooler during the entire 120 seconds of the CPT.

Pain perception was assessed with a computerized visual analog score (CoVAS), which provides real-time continuous measurement of pain intensity. This device allowed extracting a VAS value, ranged from 0 (absolutely no pain) to 100 (worst pain imaginable), for each of the 120 seconds of the test. In addition, participants reported the average pain intensity felt during the former period using a verbal numeric scale (VNS) from 0 to 100.

At the end of the music condition, and to provide further insight about their emotional experience, participants rated their self-selected music piece based on four dimensions with a VAS ranging from 0 to 10: relaxed–tense, very sad–very happy, unpleasant–pleasant, and angry–calm. Participants also used VAS to assess the level of satisfaction and the consistency between the expectation arisen in relation to the music titles and the listening of the music track.

The pilot study used a counterbalanced crossover design with two conditions, one with a self-selected music session (music condition) and one without (no-music condition), while undergoing the thermal stimulation paradigm (see Figure 2). The clusters were defined by presentation order: music condition followed by no-music condition (MN group, n = 13) versus no-music condition followed by music condition (NM group, n = 14). The randomization was pseudo-stratified by sex using a table of random numbers. Each condition lasted for 16 minutes with a 10-minutes rest period between, to allow the recovering of skin receptors. The music sequence was introduced 10 minutes before starting the thermal stimulation paradigm during music condition. In the no-music condition, participants rested in silence for 10 minutes. After the music condition, participants rated their musical sequence.

Before the experiment, participants practiced a pre-test phase in order to get familiarized with the CoVAS and the sensations produced by heat stimulation. A calibration of the thermode temperature was performed to individually adjust pain intensity to 50/100 VAS score. During this pre-test, the thermode was applied on the right forearm. Four phasic heat stimuli (36°C-50°C) were successively delivered to obtained pain threshold, pain tolerance and medium pain scores (50/100). However, the maximum temperature was fixed at 46°C during the tonic heat pain tests to prevent tissue damage. This was not enough to provoke a 50/100 VAS score in 14 participants, with a lowest value of 20.

After this practice block, participants were comfortably lying down. Headphones were positioned with the noise reduction function on. They were given computerized VAS and advised to move the cursor toward the right when they would start to feel pain. The extreme left was defined as “no pain” and the extreme right as “worst pain imaginable.” Before running the thermal stimulation paradigm, participants were told that the thermode temperature will remain stable even thought they could experience some subjective temperature variations.

Statistical analyses were calculated using Statistica version 13.3 for Windows (TIBCO Software, Inc., 2017). Significant level was set at a p value of less than .05. No significant deviation from normal distribution was detected using Q-Q plots inspection and Kolmogorov-Smirnov test, except for pain ratings at 60 second (VAS₆₀) during P3.

Prior to the study analyses, we determined PP, TS, and CPM scores (Figure 3). PP was individually estimated as the maximum pain rating that occurred during the first 30 seconds of the contact heat pain tests, under each condition and in each period. We calculated TS magnitude with within-period difference scores. TS magnitude was defined as the difference between VAS rating at 120 seconds (VAS₁₂₀) and VAS rating at 60 seconds (VAS₆₀) of the contact heat pain tests, under each condition (with and without music) and in each period (P1 and P3). Paired Student’s t-test were used to ascertain TS in P1 by comparing VAS scores at T₆₀ and T₁₂₀ for each condition. The comparison between VAS₆₀ and VAS₁₂₀ ratings in P3 was tested using Wilcoxon matched-pairs signed rank test. Regarding DNIC activation, CPM was calculated across conditions as the mean VAS score during P1 (VAS P1) minus the mean VAS score during P3 (VAS P3). A three-way mixed ANOVA was conducted to confirm CPM inhibition with order groups (MN or NM) as between-subjects factor and conditions (music and no-music) and periods (P1 and P3) as within-subjects factors.

Demographic and clinical characteristics differences between order groups (MN vs. NM) were detected by independent Student t-test for age, depression, anxiety, and catastrophism. For global perceived pain, a mixed repeated measures ANOVA was conducted with conditions and periods (P1, P2, and P3) as within-subjects factors and order groups as between-subjects factor. Same model was applied for PP and TS variables with two levels of periods (P1 and P3). CPM-effect (VAS P1 - VAS P3) was compared between conditions with a paired Student’s t-test. For all ANOVAs, Greenhouse-Geisser correction was used when Mauchley’s test of sphericity was violate. Post hoc tests were performed with Bonferroni corrections. Partial eta square (η²) was used as the effect size for ANOVA analysis, while Cohen’s d (d) was used for t-tests and Cliff’s Delta (D) for Wilcoxon matched-pairs signed rank test (calculated with the Cliff’s Delta Calculator from Macbeth, Razumiejczyk, & Ledesma, 2011). Partial eta square was interpreted as small effect = 0.01, medium effect = 0.06, large effect = 0.14 and Cohen’s d was interpreted as small effect = 0.2, medium effect = 0.5, large effect = 0.8 in accordance with Cohen’s guidelines (Cohen, 1977). Cliff’s Delta was interpreted as follow: small = 0.11, medium = 0.28, large = 0.43 (Vargha & Delaney, 2000). The results of this randomized study are in compliance with the guidelines of the consortium on the assessment of non-pharmacological treatments.

Demographic and clinical characteristics of participants are summarized in Table 1. The average temperature to evoke a 50/100 score did not differ between order groups nor between sexes. STAI, BDI, and PCS-CF scales did not show any significant differences between the MN and NM groups.

On average, participants rated self-selected musical sequences very relaxing (1.62 ± 1.17), emotionally neutral (5.48 ± 1.66), very pleasant (8.15 ± 1.31), very calm (9.07 ± 0.86), satisfying (6.96 ± 1.94), and congruent with the statement (6.33 ± 2.45).

Figure 4 is a plot showing the averaged pain ratings per second for all participants during the total pain experimental procedure in both conditions. The curves display a pattern consistent with the typical responses to contact heat pain tests and CPM. In each experimental condition, Student’s t-tests confirmed the presence of TS during P1. The last pain ratings (VAS₁₂₀) were significantly increased compared to VAS₆₀ ratings for the no-music condition, (t = 5.13, p < .001, d = 0.99) and the music condition (t = 5.61, p < .001, d = 1.08). VAS₁₂₀ ratings were also significantly higher than VAS₆₀ ratings during P3 in the no-music condition (z = 4.47, p < .001, D = 0.42) and the music condition (z = 4.43, p < .001, D = 0.34). Furthermore, ANOVA revealed a significant reduction in heat pain intensity following the CPT, reflecting CPM inhibition. The average pain ratings between P1 and P3 were significantly different, F(1, 25) = 8.75, p < .007, η² = 0.81. Post hoc tests confirmed a significant higher VAS scores during the first tonic heat phase in both no-music (p < .02) and music conditions (p < .03).

Table 2 depicts mean VAS and VNS scores during P1, P2, and P3 by conditions. Pain ratings differed significantly across conditions; VAS: F(1, 25) = 53.48, p < .001, η² = 0.68; VNS: F(1, 25) = 54.01, p < .001, η² = 0.68, and across periods; VAS: F(1, 25) = 32.95, p < .001, η² = 0.57; VNS: F(1, 25) = 39.63, p < .001, η² = 0.61. Bonferroni corrections confirmed that the averaged pain perception was significantly higher during no-music condition than during music condition (VAS: p < .002, VNS: p < .001). Pain ratings were significantly higher during the CPT than during the first and the second tonic heat pain tests (VAS: p < .001, VNS: p < .001). There was no order effect. No significant interaction effect was found between condition and period, F(1, 25) < 0.44, p > .75). Post hoc tests showed that pain values in music condition were significantly lower than those in the no-music condition at P1, P2, and P3 (VAS: p < .001; VNS: p < .001), (see Figure 5a and 5b). In addition, average pain scores were significantly lower in P3 than in P1 for the two conditions (Figure 5b). However, this difference reached significance only for VNS (Music: p < .001; No-music: p < .02). Differences in mean VAS values between P1 and P3 were only significant when P2 was removed from the analysis.

PP ratings differed significantly across conditions, F(1, 25) = 16.75, p < .001, η² = 0.40, and across periods, F(1, 25) = 19.44, p < .001, η² = 0.44. No order effect was found, F(1, 25) = 0.21, p > .65. During the PP phase, the maximum pain intensity score was different between conditions (p < .001). The effect of condition was significant during P1 and P3 (P1: p < .001; P3: p < .001), where on average no-music condition was associated with rates higher (P1: 52.03 ± 5.10; P3: 37.63 ± 4.52) than during the music condition (P1: 40.88 ± 4.21; P3: 26.85 ± 4.28) (mean ± SE). In addition, a significant effect of period was found for both no-music and music conditions (ps < .001), suggesting the analgesic effect of CPM-inhibition on PP.

TS amplitude (mean ± SE) was found at 17.22 ± 3.36 VAS units for P1 and 17.89 ± 3.00 for P3 in the no-music condition. In the music condition, the magnitude of TS at P1 and P3 was 10.56 ± 1.88 and 10.11 ± 1.77 VAS units, respectively. The three-way ANOVA revealed a main effect of condition, F(1, 25) = 15.82, p < .001, η² = 0.34, with higher TS amplitude for the no-music condition than the music condition (p < .001). Post hoc tests demonstrated that TS magnitudes values were significantly decreased in the music condition for P1 (p < .04) and P3 (p < .03) (see Figure 5c). No significant order effect, F(1, 25) = 0.08, p = .78), neither period effect, F(1, 25) = 0.018, p = .89, were found. Interaction between condition and period was not significant, F(1, 25) = 0.10, p = .75.

No difference in CPM inhibition (VAS P1 – VAS P3) was found between the two conditions (p = .88).

In this pilot study, we aimed to determine if music-induced analgesia may involve inhibitory descending pain pathways and whether the CPM paradigm associated with tonic heat pain stimuli was well suited to evaluate such involvement. The global activation pattern of the pain-related responses to endogenous pain mechanisms observed in this study is consistent with the initial results, accounting for the peak of first pain (PP), temporal summation of second pain (TS), and diffuse noxious endogenous control (DNIC)-effect (Tousignant-Laflamme et al., 2008). Psychophysically, our results show an inhibitory effect of music listening on the behavioral correlates of first and second pain, regardless of CPM-effect. The CPM paradigm used in this study appears to be a reliable and convenient tool to study central pain modulation, since the lack of music modulation on DNIC activation can be interpret as an interesting result. The outcomes suggest the contribution of descending pain-modulatory systems in the analgesic effect of music and support the independency of DNIC feedback loop to the “top-down” descending inhibitory pain pathways.

The participants chose musical pieces they rated as relaxed, emotionally neutral (neither happy nor sad), pleasant, and calm, confirming that the music choice ensured optimal conditions for music-induced analgesia (Good & Ahn, 2008; Roy et al., 2007). Participants reported significantly less pain with music compared to no music, confirming previous studies (Gardner et al., 1960; Good, 1996; Good & Ahn, 2008; Guétin et al., 2012). Pain ratings were approximately reduced by 10 units on the VAS and the VNS by music. It corresponded to about a 30% reduction of pain. It occurred during heat pain stimulations in this and other studies (Dobek et al., 2014; Roy et al., 2007). The present data indicates a music-induced analgesia triggered by cold nociceptive stimulation similar to what was observed for heat stimulations with a 18% reduction for all participants and without gender difference. However, other authors found analgesia to be effective on pain induced by cold pressure only in reducing pain unpleasantness (Garcia & Hand, 2016) or only in women (Mitchell & MacDonald, 2006). Differences in cold-water temperature may explain this discrepancy since those two studies used a 4 or 5°C temperature to elicit pain, in place of 12°C in the present work. Although 12°C was found sufficient to trigger a DNIC effect, pain perception is known to increase while the CPT temperature decreases (Mitchell, MacDonald, & Brodie, 2004; Tousignant-Laflamme et al., 2008). Music analgesia could be ineffective to reduce most severe pain. The same emotional brain circuits, including the amygdala, are known to be activated by both music and pain stimulations (Blood & Zatorre, 2001; Koelsch, Fritz, Cramon, Muler, & Friederici, 2006; Neugebauer, 2015). Sharp pain could allocate some cerebral regions involved in affective process exclusively to nociceptive inputs, rendering music analgesia inefficient.

Psychophysical paradigms, such as TS of second pain and CPM, have been widely used to study endogenous pain facilitation and inhibition in healthy and clinical populations (Granot, Granovsky, Sprecher, Nir, & Yarnitsky, 2006; Granot et al., 2008; Horn-Hoffman, Kunza, Maddena, Schanbela, & Lautenbachera, 2018; Julien, Goffaux, Arsenault, & Marchand, 2005; Mackey et al., 2017; Nielsen & Arendt-Nielsen, 1998; Potvin et al., 2007; Price et al., 1977; Price et al., 2002; Tousignant-Laflamme et al., 2008). Along with functional imaging (Dobek et al., 2014) and the RIII reflex (Roy et al., 2011; Ruscheweyh, Kreusch, Albers, Sommer, & Marziniak, 2011), this study promotes psychophysical testing to evaluate the effect of music on inhibitory descending pain systems.

On one hand, we elicited peripheral and spinal pain facilitation with noxious heat stimulations in order to evaluate the effect of music listening on first and second pain responses. Acute pain perception is divided into two phases characterized by distinct temporal responses (Nielsen & Arendt-Nielsen, 1998; Ploner, Gross, Timmermann, & Schnitzler, 2002; Price & Dubner, 1977). First pain, which is also called fast-pain, is described as a sharp, pricking, and well-localized sensation. As we expected, this study demonstrated that music-induced analgesia induces an inhibitory effect on the peak of first pain (PP). During the first 30^th seconds of the heat tests, maximum pain ratings were reduced by approximately 25% during the music session. PP is a first pain-related phenomenon that correlates to A-delta fibers activity (Tousignant-Laflamme et al., 2008; Treede et al., 1995), along with the spinal nociceptive withdrawal reflex (the RIII reflex; Dowman, 1991; Willer, Boureau, & Be-Fessard, 1978). A-delta fibers transmitted nociceptive information to central transmission neurons with a fast conduction velocity. Roy, Lebuis, Hugueville, Peretz, and Rainville (2012) investigated the modulatory effects of the emotional and arousal dimensions of music on the RIII reflex. RIII amplitude was found to decrease during the presentation of musical excerpts. Larger physiological effects were associated with pleasant-relaxing musical stimuli (Roy et al., 2012). Our results support, in accordance with this previous study, an inhibitory effect of music on pain signals conducted by A-delta fibers through descending modulation. Because music was judged as very pleasant and relaxing in the current study, both emotions and arousal induced by music could initiate the activation of the inhibitory pain pathways (Roy et al., 2012).

In addition, music listening was also found to affect second pain transmission. Second pain sensation is described as longer-lasting, burning, and more diffuse than first pain (Ploner et al., 2002). The amplitude of the temporal summation of second pain was reduced by about 40% when participants were listening to music. TS is the behavioral correlate of wind-up of spinal cord neurons (Chen, Shomojo, Svensson, & Arendt-Neilsen, 2000; Granot et al., 2006; Nielsen & Arendt-Nielsen, 1998; Potvin et al., 2007). It corresponds to an increase of pain perception associated with repetitive or persistent stimulations of nociceptive C-fibers (Nielsen & Arendt-Nielsen, 1998). Extended activation of those primary fibers enhances excitability of dorsal horn neurons (Li, Simone, & Larson, 1999), resulting in pain progressive amplification in association with changes in cerebral dynamics (Chen et al., 2000).

The concomitant impact of music listening on spinal pain transmission mediated by afferent A- and C-fibers strongly suggests the activation of supraspinal descending systems, involved in local spinal inhibition (Bingel et al., 2007; Calvino & Grilo, 2006; Cheng et al., 2015; Fields et al., 1977; Fields, Bry, Hentall, & Zorman, 1983). This is consistent with the implementation of descending inhibition pain pathways that blocked the transmission of nerve impulses from nociceptive non-specific dorsal horn neurons (Bingel et al., 2007; Fields et al., 1977). We propose that music listening could strengthen descending pain modulation by cerebral mechanisms that integrated psychological and physiological factors that are discussed below.

Our finding regarding spinal facilitatory pain mechanisms are at variance with those of Ruscheweyl et al. (2011) reporting no inhibition of PP and TS by music. This apparent discrepancy may be due to differences in goals and designs between the two studies. The purpose of this latter study was to assess whether PP and TS were modulated by several distraction tasks including music listening. In this context, participants were exclusively under music stimulation during the electrical pain procedure. In the current study, music was listened to for 10 minutes before the thermal stimulation paradigm, rendering the stimulus not only distracting but mainly as an inducer of emotions and relaxation. Furthermore, participants were asked to continually assess pain during the thermal stimulation paradigm so that their attention was intentionally maintained toward painful stimulus, supporting the primacy of pain processing over any concurrent tasks (Miron, Duncan, & Bushnell, 1989). Thus, the effect of music distraction cannot be considered as a major factor in the interpretation of our results. The comparison between these two studies suggests that central modulation of pain by music depends on the activation of cerebral areas involved in emotional and/or relaxation processes.

On the other hand, inhibitory CPM-effect was well activated by CPM paradigm as expected, but not reinforced by music listening. This result is in evident opposition with the hypothesis that music could strengthen inhibitory descending pathways including DNIC. However, studies in rats demonstrated that this lower brainstem mediated inhibitory mechanism was relying on a spinal-supraspinal-spinal circuitry regardless of cortical brain activity (Bouhassira, Villanueva, Bing, & Le Bars, 1992; Patel & Dickenson, 2019; Villanueva & Le Bars, 1995). This result suggests that efferent inhibitory fibers involved in DNIC and music analgesia rely on distinct descending funiculars. We propose that CPM may be independent from emotion and/or relaxation as it is for distraction (Moont, Pud, Sprecher, Sharvit, & Yarnitsky, 2010). Therefore, cerebral process elicited by music stimulation could specifically target brainstem structures involved in general descending inhibitory pain pathways, irrespective of DNIC feedback loop.

The activation of descending inhibitory pathways in response to pleasurable music is consistent with imaging studies in humans that showed that pleasant music activates brain and brainstem areas such as the prefrontal cortex, the amygdala, and the periaqueductal gray, which are assumed to play a role in descending pain modulation (Blood & Zatorre, 2001; Dobek et al., 2014; Koelsch et al., 2006; Millan, 2002; Roy et al., 2009). Neuro-endocrinological activities related to music or pain processes overlap suggesting, once more, the implication of central modulation in the antinociceptive effect of music (Blood & Zatorre, 2001; Dobek et al., 2014; Evers & Suhr, 2000; Goldstein, 1980; Koelsch et al., 2006; Neugebauer, 2015; Salimpoor, Benovoy, Larcher, Dagher, & Zatorre, 2011). Pleasurable experiences of music induce release of some of the neurotransmitters such as opioid, dopamine, and serotonin (Evers & Suhr, 2000; Goldstein, 1980; Salimpoor et al., 2011), which are associated with descending inhibitory pain modulation. The descending pain-modulation pathways originate from the periaqueductal gray, which contains populations of opioidergic and dopaminergic receptors (Fields, 2000; Wood, 2008). The opioid and dopamine neurons of the periaqueductal gray interact, contributing together to the activation of the descending serotonin track that project to the spinal dorsal horn, and inhibits A and C-fibers-mediated transmission (Doly, Fischer, Brisorgueil, Vergé, & Conrath, 2005; Pierce, Xie, Levine, & Peroutka, 1996). Recent functional magnetic resonance imaging responses in the amygdala and the periaqueductal gray were associated with a greater reduction of the dorsal horn responses in participants who received local heat pain stimulation when listening to music (Dobek et al., 2014). In conclusion, those results taken together are consistent with the hypothesis that emotions and relaxation induced by music could enhance the central neuromodulation of pain signaling from limbic and paralimbic regions to spinal dorsal horn.

Some limitations of our pilot study should be emphasized. The maximal temperature generated by the thermode was targeted at 46°C. This value was not enough to provoke in 14 participants a 50/100 VAS score. This may have limited the amplitude of the endogenous pain mechanisms. In particular, the absence of music modulation induced by CPM could be partly explained by an insufficient TS activation that would have hidden the role of DNIC in music analgesia.

The question of sexual difference in music-induced analgesia was not placed at the start of the study. For example, the possible role of female hormonal status (Iacovides, Avidon, & Baker, 2015) was not questioned. Therefore, our results, obtained with a small group, cannot be considered at variance with the higher sensitivity to experimental pain reported in healthy woman compared with men (Fillingim & Ness, 2000).

Finally, we recommend that this preliminary study be replicated with a larger sample size to include older healthy participants. The inclusion of additional variables is also recommended, such as music training, sound level, and stress assessment. Those potential factors could contribute to music-induced analgesia through distraction or engagement processes. Moreover, it would have been valuable to have some indicators of stress level before and after the thermal stimulation paradigm in each condition. The State-Trait Anxiety Inventory was the only criterion considered. In the future, the inclusion of at least a perceived stress scale before, between and after sessions would be valuable. In addition, the musical sequences used were soothing, instrumental, and new to the participants. These particular pieces of music question the impact of the arousal and the significance attached by the individual to the music type. Future studies should examine the reproducibility of the results with self-selected preferred music, which should involve a high self-commitment and emotional engagement.

The current study shows the inhibitory effect of music on pain-related responses to spinal facilitatory pain mechanisms, consistent with the activation of descending inhibitory pain pathways. Given its universal and financial accessibility, we widely recommend music listening as a complementary approach to manage acute pain, in order to reinforce the descending pain modulation system.

MC conceived the study design and conducted the recruitment and the outcomes assessments under the supervision of CB. ALFB provides her help and expertise in the experimental evaluations. MC analyzed the data. MC and CB wrote the paper. MC prepared figures and/or tables. CB, AW, ALFB, SG, BQ and LM reviewed drafts of the paper. All authors approved the final version of this paper.

We thank Music Care for the access to Music Care database. This study was funded by the APICIL foundation (reference 1084.17). The APICIL was not involved with the data collection or analysis. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed. No writing assistance was utilized in the production of this manuscript.