I had the following article in the ICMPC 11 proceedings.


An often cited conclusion that musical chills are mediated by endogenous opioids (endorphins) is based on an experiment that showed the opioid antagonist naloxone reduced the chills rate of music in some subjects. However, we find some experimental problems with its methods, results and conclusion. Dr. Goldstein's experiment with musical chills and naloxone used 10 subjects, all music chill responders, and found that 3 had significant chill reduction related to naloxone. He did not provide the result showing if the other 7 had any reduction at all, and the assumption would be that they had no reduction in chills. Naloxone is a highly competitive opioid antagonist, thus this result is odd if music chills are from endorphins.

An experimental confound, not known in 1980, is endogenous opioids are partly responsible for natural relief from chronic pain. If the three subjects who showed chill reduction had chronic pain relieved by endorphins, then Naloxone would have restored the pain. The pain increase, which need not be perceptible, could influence the ability to experience chills. If 7 of 10 subjects had no reduction in chills, then the opposite conclusion, that chills do not arise from endorphins, could be made. We conclude the results of the 1980 music chills and Naloxone experiment should not be used to claim that musical chills are mediated by endogenous opioids. 


  • In many people, listening to music may induce goose-bumps or chills – we will refer to them as music chills. Music chills are reported to be very pleasant. In fact, music chills have been shown as connected to definite brain pleasure center activations, the nucleus accumbens and the VTA [Menon2005]. It is often claimed that music chills are mediated by endogenous opioids, endorphins, based on an experimental result by Dr. Goldstein [Goldstein1980] in 1980. His paper has 2 parts: a survey of people's experience with chills, and an experiment exploring the effect of an opioid antagonist, naloxone, on subjects experiencing musical chills. It is of course the experiment with naloxone that is used for supporting the claim that endorphins mediate music chills.

    The claim that endorphins mediate music chills does feel seductive; it sounds very reasonable given the euphorigenic effects of opioids, and this reasonableness may explain the strong embellishment many other researchers have given to those results. Unfortunately, a critical examination of Goldstein's results do not actually support the claim that endorphins mediate music chills, and in fact, the results may well support the opposite conclusion – that endorphins are not involved in the pathway for music chills.

    This is an important issue to resolve since any understanding of the origins of music must eventually account for the music chill phenomena. If music chills are an endorphin euphoria or are generated as part of a neural pathway using a strong reward from endorphins, then this may imply music has a strong intrinsic hedonic value. Furthermore, since music chills seem to require an emotional response to music, it may imply that it is the emotional response peak that has hedonic value. This hedonic reward contrasts with a motivation reward, usually associated with dopamine [Berridge2008]. Perhaps instead, a music chill may be a dopamine reward for motivation and a release of epinephrine for the physical aspects of the chill (the physical aspects are piloerection, skin conductance change, and heart rate change). Resolving the music chill reward basis is important because two very different musical evolutionary paths may be needed to explain an endorphin reward vs. a dopamine with a side of epinephrine reward.

     The results from Dr. Goldstein's naloxone experiment are highly statistically significant; that is, p<0.01 that naloxone had no affect on chill production by music. Since naloxone totally blocks the euphorigenic effects of opioids and naloxone blocked music-chills at a statistically significant level, the assumed conclusion that endorphins mediate music-chills seems natural. But there is a problem. While it is absolutely true that, in his experimental protocol, naloxone strongly interfered with music-chills in some subjects - to be precise, perhaps only three of ten subjects - having the effect occur on well less than half of the subjects is problematic for claiming endorphins mediate music-chills. We discuss why a confound, the endogenous pain control system, could explain the response of the three subjects, and, more importantly, that if seven subjects were unaffected, then why the conclusion stating that endorphins mediate music-chills is unlikely.


  • Most of his article covered a very useful survey about chills, the activity sources of chills, and the populations who experience chills. The naloxone experiment's methods, results and discussion are a small part of the paper which we cover here.

    For the naloxone experiment, ten subjects were selected who had strong chill responses to music. The selected subjects were strong responders – the only data detailed in the paper, a single 6 minute trial of one subject, averaged one chill per 14 seconds. Each subject provided the music that induced chills for themselves. Subjects indicated the presence of a music chill by raising 1 to 3 fingers, the number of which indicated the intensity of the chill. He then says, "In preliminary experiments, saline or naloxone hydrochloride (10mg/ml) was injected intravenously in a double-blind manner, between two auditions of the same musical passage. Thrill scores, based on frequency, intensity, and duration, were compared for the preinjection and postinjection audition." p126. This constitutes all the method section for the naloxone experiment.

    In the results section for the naloxone experiment, he unfortunately provides no data. He first provides a few observations about the variability between subjects concerning where in musical passages subjects would experience chills. On the naloxone part of the experiment, the totality of his presented results consists of about 90 words, and follows:

    In 3 of 10 subjects, thrills were significantly attenuated by naloxone as compared to saline. For 19, 10, and 9 sessions (with approximately equal number of naloxone and saline injections), the data on these three subjects yielded one-tailed p values of 0.0015, 0.032, and 0.038, respectively. As shown by Fisher, the null hypothesis in this case --that naloxone had no effect in any subject-- can be rejected at p<.01. We have shown elsewhere that double-blind injections of naloxone can not be distinguished from saline by any interoceptive cues”. pp127-128.

     In the discussion of the naloxone experiment, he merely says, "Preliminary experiments suggest that in some people thrills elicited by music may be attenuated by naloxone, a specific antagonist at the opiate (endorphin) receptors." p128


  • Unfortunately, we must make a two assumptions about the experiment and the results.

    The first assumption is the dosing of the naloxone. Naloxone (AKA narcan) is a competitive opioid antagonist, which means it displaces opioids from opioid receptors and binds to the receptor without activating the receptor. Its effective medical use is treating opioid overdoses to restore breathing. Naloxone, in a sufficient dose (less than 4 mg), will induce within minutes withdrawal symptoms in an opioid addict. A 10 mg dose is the maximum clinically used dose; if breathing is not restored, the assumption is opioids are not the cause. A 2mg dose also removes any pain control given by the opioids. In many naloxone experiments that are testing the possible involvement of endorphins, 10 mg is a common dose.

    Although drug density was given at 10mg/ml, the actual dosing was not explicit in this experiment, however, in both prior and later published experiments (not music chills experiments), Dr Goldstein used 10mg naloxone doses. Thus, in this music chill experiment, the dose seems likely to be 10 mg although it is ambiguous.

    The second assumption concerns the data of the other 7 subjects. Had they as a group experienced a reduction of music chills but where each individual did not have a significant reduction? That statistic would have been a very important result and one that would have been detected if it existed. Without other evidence, we must assume that the other seven subjects did not have a reduction in music chills with naloxone.

    Given the following factors, (1) the capability of naloxone as a competitive opioid antagonist, (2) the use of a medically significant dose of naloxone, and (3) 70% of the subjects did not have a reduction in music chills, then the conclusion of the experiment could be that endorphins do not mediate the music chill response.

    If endorphins are not mediating the music chill response, then for the 3 subjects who showed a strong reduction in music chills, an alternate explanation is needed. One explanation is, because naloxone negates the endorphins created by the body's endogenous pain control system [Bruehl2004], a slight chronic pain sensation increased in those 3 subjects and the pain interfered with experiencing the chills. The pain need not be very noticeable to interfere.

     The human endogenous pain control system is complex; it covers brain structures, spinal areas and even peripheral nerves, and some of the pain control is endorphin based. Acute pain can be reduced by placebo effect in the presence of stress via endorphins and thus the pain relief effect is removed by naloxone [Grevert1983, Zubieta2009] . Ironically, Dr Goldstein discovered that effect in 1983. Chronic pain control by the endogenous pain control system also uses endorphins, even at the peripheral level [Stein2003]. The pain of a small healing cut, for example, seems controlled by opioid secretion for peripheral pain control. The conclusion is clear, if the subjects had any minor chronic pain that was controlled by the endogenous pain control system, naloxone would have restored the sensation of that pain.


  • Dr. Goldstein's work was preliminary, as stated by him. The experimental results had

    statistical significance but something may have seemed odd to him. He probably wished this result to be true because it made sense, and furthermore, the placebo pain reduction and chronic pain control via endorphins was not known at the time. This paper should not reflect on the rest of his amazing career of research and discovery; a career so impressive his biography was part of an issue of Annual Review of Pharmacology and Toxicology from Annual Reviews in 1997.

    Endogenous opioids, such as the endorphins, mediate many neural pathways in the body and brain; endorphins mediate both a particular reward pathway and the endogenous pain control system. The endogenous pain control system was not understood when Dr. Goldstein's experiment was performed, but the opioid reward pathway was. By blocking the opioid receptor sites with naloxone thereby stopping both the reward pathway and the endogenous pain relief pathway, the music chill reduction effect occurred in only three of ten subjects. For music-chills, those results are more consistent with attributing naloxone causing chronic pain restoration than they are with naloxone blocking the opiate reward pathway.

    The naloxone experiment with music chills by Dr. Goldstein cannot be used to claim that endorphins mediate music chills in some direct pathway. If the assumptions are true about the other 7 subjects, then the opposite could be asserted with greater probability; that endorphins are not in a direct pathway for music chills.


  • Berridge, K.&Kringelbach, M. (2008). Affective neuroscience of pleasure:

    reward in humans and animals., Psychopharmacology 199-3, 457-480

    Bruehl, S., Chung, O. Y., Burns, J. W.&Biridepalli, S. ( 2003). The association between anger expression and chronic pain intensity: evidence for partial mediation by endogenous opioid dysfunction., Pain, 106-3,. 317--324.

    Bruehl, S., Chung, O. Y., Ward, P.&Johnson, B. (2004 ). Endogenous opioids and chronic pain intensity: interactions with level of disability., Clin J Pain, 20:5, 283--292.

    Goldstein, A. ( 1980). "Thrills in Response to Music and Other Stimuli," Physiological Psychology, 8, 126-129.

    Grevert, P., Albert, L. H.&Goldstein, A. (1983). Partial antagonism of placebo analgesia by naloxone., Pain, 16:2, 129--143.

    Menon,V.&Levitin, D. J. (2005). The rewards of music listening: response and physiological connectivity of the mesolimbic system., Neuroimage, 28:1, 175--184.

    Stein, C., Schäfer, M.&Machelska, H. (2003). Attacking pain at its source: new perspectives on opioids., Nat Med, 9:8, 1003--1008.

    Zubieta, J.-K.&Stohler, C. S. (2009). Neurobiological mechanisms of placebo responses., Ann N Y Acad Sci, 1156,198--210.