Singing and the Body

Grape et al. 2003 — the pilot: amateurs vs. professionals

Christina Grape and colleagues' 2003 paper in Integrative Physiological & Behavioral Science is the foundational empirical study. Sixteen singers — eight amateur, eight professional, all with at least six months of weekly lessons — each completed a single singing lesson while the researchers recorded ECG, drew blood, ran visual-analogue mood scales, and conducted a semi-structured interview about the experience.

The physiological results sketch a pattern that subsequent studies refine but do not reverse. Serum oxytocin rose in both groups after the lesson — the first time the oxytocin signal appeared in this literature. Cortisol and prolactin showed sex-specific responses (rising in men, falling in women), suggesting the endocrine response to singing is sex-modulated in ways no single number captures. Heart-rate variability — specifically the total power and the low- and high-frequency bands — rose during singing in professionals but not in amateurs. The authors read this as the "heart-brain connection" of trained singers: the body has learned to coordinate cardio-respiratory work in a way the amateur body has not. TNF-alpha, an inflammatory cytokine, rose in professionals and fell in amateurs after the lesson, hinting that the same activity is registered as training stress by one group and as relief by the other.

The affective results are equally interesting. Both groups felt more energetic and more relaxed after the lesson. But only amateurs reported increased joy and elatedness. The interviews supplied the likely explanation: professionals approached the lesson achievement-oriented (technique, vocal apparatus, bodily control), while amateurs framed it as self-expression and emotional release. The same activity produces different affect when it is held under different intentions. This is the first hint, in the singing literature, that what the singing is for matters as much as the singing.

Kreutz et al. 2004 — singing vs. listening

Gunter Kreutz and colleagues separated the producing from the receiving. A choir was randomized: half sang the same piece; half listened to it through headphones with the same duration of exposure. Saliva samples for cortisol and secretory IgA (an antibody marker of mucosal immunity) were collected before and after, alongside mood ratings.

The producers showed two things the listeners did not. Secretory IgA rose — the immune system marked the act of singing with an antibody bump. Cortisol dropped, and mood improved. The listeners, exposed to the identical music for the identical time, showed neither immune nor cortisol changes of comparable magnitude. The most efficient summary is the one many singing teachers had been giving for decades without the assay numbers: you have to do it for it to do it to you. Music is not, in this study, a chemical that diffuses into a passive recipient. It is a behavior that changes its agent.

Fancourt et al. 2016 — the clinical population

Daisy Fancourt and colleagues extended the work to a clinically meaningful sample: cancer patients and their caregivers participating in group-singing programs run by a charity. The study took saliva samples before and after a one-hour group-singing session and measured cortisol, a panel of cytokines (including IL-6, IL-17, IL-2, IFN-gamma), beta-endorphin, oxytocin, and mood.

The results extended the Kreutz and Grape findings to a population under genuine chronic stress. Cortisol dropped significantly. Several inflammatory cytokines decreased. Beta-endorphin and oxytocin shifted in directions consistent with affiliative and rewarding experience. Mood improved across the group. The clinical implication is the one that has driven the subsequent growth of singing programs in cancer care: a one-hour intervention with no pharmacology produces measurable changes in inflammation, stress, and mood in a population for whom those three variables are tightly coupled to disease course and quality of life.

Bullack et al. 2018 — the Kreutz group's longer-form replication

Antje Bullack, Carolin Gass, Urs M. Nater, and Gunter Kreutz returned to the singing-vs.-listening question fourteen years after the original Kreutz 2004 study, with a refined design published in Frontiers in Behavioral Neuroscience. A mixed amateur choir completed two sessions one week apart — one a normal 60-minute singing rehearsal, one a 60-minute listening condition with the same choral repertoire piped through speakers while participants sat passively. Saliva samples for secretory IgA and cortisol, plus positive- and negative-affect scales, were collected immediately before each session and again 60 minutes later.

The singing condition reproduced the 2004 findings cleanly. Positive affect rose, negative affect fell, and secretory IgA increased after the hour of singing. The pattern is now well-established: active group vocal production produces a converging shift in mood and mucosal immunity within an hour.

The listening condition produced the methodologically important result. Cortisol fell — passive exposure to choral music does regulate the stress axis — but, paradoxically, negative affect rose rather than fell. Listening alone in a research setting was experienced as somewhat dysphoric even as the body's stress hormones came down. The endocrine and the subjective dissociated cleanly. This matters for three reasons. First, it shows that "cortisol down equals feel better" is not a reliable shortcut. Second, it locates the affective payload on the active side — what your body does, not what your body is exposed to. Third, it foreshadows the cleaner dissociation that Good & Russo would isolate four years later with oxytocin as the affect-tracking variable.

Good & Russo 2022 — the cleanest dissection

Arla Good and Frank A. Russo's 2022 pilot in Psychology of Music is the within-subjects study the previous literature needed. Every participant sang in both conditions: once in a group, once alone, with the same material. Saliva samples for cortisol and oxytocin and self-report mood were taken before and after each session. Because each person serves as their own control, the design isolates the social variable cleanly.

The dissociation is sharp. Cortisol dropped in both conditions — solo and group singing both regulate the stress axis, consistent with Kreutz's finding that the act of producing the sound is what does it. Oxytocin rose only after the group condition. Mood improved only after the group condition. Most importantly, across participants the magnitude of the mood improvement correlated with the magnitude of the oxytocin rise, and did not correlate with the magnitude of the cortisol drop. So even though cortisol fell in both, the falling cortisol is not what made people feel better. The oxytocin is. And oxytocin needs the group.

Schlaug et al. 2005 — beyond singing: instrumental training and developmental plasticity

The five studies above record what happens to the body during a single hour of singing. Gottfried Schlaug, Andrea Norton, Katie Overy and Ellen Winner's 2005 review in the Annals of the New York Academy of Sciences records what happens to the body over years — specifically, what instrumental music training does to the developing brain of a child. It is the long-timescale complement to the acute-endocrine work above, and the trilogy uses it for the same purpose: to support the claim that the body is tunable through music.

Structural changes in adult musicians. The review synthesizes a decade of prior neuroimaging that established the baseline: musicians who began training early differ structurally from non-musicians in identifiable ways. The corpus callosum — especially the mid-body fibers that coordinate the two hemispheres for bimanual action — is enlarged in early-trained musicians, consistent with the demand instrumental playing places on interhemispheric coordination. Gray matter is enlarged in motor cortex, premotor areas, auditory cortex (including Heschl's gyrus, where pitch is first cortically represented), and parts of the cerebellum. fMRI shows that during melodic and rhythmic tasks, musicians recruit auditory-motor networks more efficiently and with more focused activation than non-musicians do.

Children, longitudinally. Schlaug's contribution is to show that these adult differences are at least partly caused by training, not merely preselected. His longitudinal cohort of 5–7-year-olds beginning instrumental lessons was matched against non-musician controls at baseline and followed forward. After about fourteen months of lessons, the trained children showed measurable gray-matter increases in motor and auditory regions and corresponding gains in fine motor control and melodic discrimination. The behavioural gains are concentrated in skills closely related to what the training actually exercised — near transfer — with broader cognitive transfer effects emerging more slowly and more modestly.

Cross-sectional convergence in older children. Children aged roughly 9–11 who had been studying instruments for several years performed better than non-musicians on standardized tests of vocabulary, motor coordination, and music-discrimination tasks, and their fMRI patterns during musical tasks were closer to those of adult musicians than to those of untrained age-mates. Selection-effect cautions apply — the children who stick with instruments differ from those who do not in ways that the cross-section cannot fully control — but the longitudinal arm of the same paper is what supplies the causal weight.

Why music is a multimodal training ground. Schlaug's conceptual contribution is the framing of instrumental training as a uniquely demanding cognitive workout: it requires mapping a symbolic notation system to fine sequential motor output, with continuous multisensory feedback (auditory, proprioceptive, visual), under hierarchical temporal constraints (phrases, meter), sustained attention, working and long-term memory load, and — in ensemble playing — a social-coordination dimension that returns us to the singing literature. Few activities engage so many systems simultaneously. The transfer story Schlaug proposes is that the systems shared with non-musical cognition — temporal processing, attention, working memory, auditory discrimination — are what get strengthened, which is why modest gains appear in reading, vocabulary, and certain mathematical operations after several years of training.

What the paper does not claim. The authors are explicit about not overselling. The far-transfer effects (the "music makes kids smarter" headline) are real but modest, require sustained training to appear, and are smaller than the popular literature implies. Music training is not a universal IQ booster. What it reliably does is induce regionally specific brain changes in auditory-motor networks and produce near-transfer gains in skills the training directly exercises — with the more diffuse cognitive benefits accumulating slowly across years.

The convergent pattern

Read in sequence, the six studies sort the body-music relationship into four components, each operating on its own timescale.

• The stress-regulation component — cortisol drop — activates with vocal production itself. Solo or group. Kreutz showed it, Good & Russo confirmed it. This is the part of the singing-high that the shower singer is getting.
• The immune-system component — secretory IgA rise — also activates with vocal production. Demonstrated cleanly by Kreutz. The producing body marks the act with mucosal-antibody activity; the listening body does not.
• The affiliative component — oxytocin rise, mood lift, sense of bonding — activates only when other people are also singing. Grape 2003 saw the oxytocin signal; Good & Russo 2022 nailed down that the group is required and that this is the variable that tracks mood. This is the part that the choir member, the drumming circle, the congregational chant is getting that the shower singer is not.
• The developmental-plasticity component — structural changes in motor and auditory cortex, larger corpus callosum, near-transfer gains in fine motor control and pitch discrimination, modest far-transfer gains in vocabulary and motor coordination — activates with sustained instrumental training over years. Schlaug 2005 is the canonical reference. This is what the body becomes when music is not just a one-hour event but a long-term practice.

Two further empirical facts cross-cut these components. Sex modulates the cortisol/prolactin response (Grape 2003) — the same behavior produces different endocrine signatures depending on hormonal substrate. And intention modulates the affective response (the Grape interview material): the same lesson read as achievement vs. as self-expression produces different mood outcomes. The body is responsive not only to what is being done but to what the agent takes the doing to be for.

Bullack 2018's most useful contribution to this picture is the demonstration that endocrine and subjective measures can dissociate cleanly. Cortisol fell with passive listening but negative affect rose. The shortcut "cortisol down = feel better" is wrong. What feels good is what your body actively does, in the presence of others doing it with you.

Why this matters for the trilogy

Four points the trilogy uses directly.

First, the chord chapters of Limen are about singing-together, not singing. The five-study convergence supplies the empirical support for the distinction. Resonance within a single body has a measurable endocrine signature (cortisol regulation, IgA). Resonance between bodies has its own signature (oxytocin, mood, the felt sense of belonging) that the single body does not produce alone. The chord, in the trilogy's grammar, is the second thing. The choir is doing something the soloist is not, and the body knows the difference.

Second, the social mechanism has the structural signature the receiver model predicts. The trilogy's larger claim is that biological tissue is tuned to a relational structure — to between-ness as a physical fact, not as a metaphor. Oxytocin is the endocrine variable the body uses to register that between-ness. The fact that it rises selectively in the group condition, and that mood tracks it, is consistent with a body whose affective state is partly determined by whether other bodies are present and in sync. This is not magical — it is the predictable behavior of an organism evolved for affiliation. But it is also exactly what a receiver-of-relationship needs to look like at the endocrine level.

Third, the prescriptive consequence: the trilogy's clinical sympathy for ensemble music as a therapeutic modality rests on the fact that the active ingredient is the ensemble, not the music. Music-therapy programs that prioritize solo work as a stress-reduction measure are getting one arm of the effect (cortisol, IgA). Programs that prioritize coordinated group production — choirs, drumming circles, the kind of collective vocalization the contemplative traditions of every continent converged on independently — are getting both arms. The Fancourt cancer-population data and the Good & Russo dissociation together are the clearest argument in the recent literature for why that matters and which biomarker tracks it.

A fourth point, quieter and more speculative: every contemplative tradition that the trilogy takes seriously — Carmelite chant, Sufi dhikr, Vedic mantra, Tibetan chö, the Eastern Orthodox obikhod — is built around group vocalization, often sustained, often near the lower end of the comfortable vocal range, almost always coordinated. The traditions converged independently on a practice whose endocrine fingerprint we are only now learning to read. The convergence is not coincidence. It is what an organism with a tuned receiver does when it is trying to listen together.

The five studies: Grape 2003 (Springer), Kreutz 2004 (PubMed), Fancourt 2016 (ecancer), Bullack et al. 2018 (Frontiers), Good & Russo 2022 (SAGE). For the cochlea-as-resonator side of the singing story see the Manoussaki cochlea papers; for the central-nervous-system side see the Oster binaural-beats explainer. For the larger picture in which singing is one of several embodied modalities by which the receiver is tuned, see the Synthesis.