The research team, led by Ellen Rankins and B. Caitlin Peters, recruited 72 autistic children between the ages of 6 and 11 and randomly assigned them to one of two ten-week programs. Both groups received identical occupational therapy built around a modified Zones of Regulation curriculum, which teaches emotional awareness and individually tailored self-regulation tools. The only meaningful difference was the setting: one group worked in an indoor clinic space, while the other performed the same 45 minutes of intervention activities while mounted on a horse in an indoor arena, complete with a horse leader, side walker, and a certified therapeutic riding instructor overseeing safety and animal welfare. Crucially, children in the equine group rode the same horse each week, a design choice that would later prove important for interpreting the hormone data.

What sets this trial apart from earlier work is its intensive biological sampling. Rather than measuring stress markers only before and after the intervention, the researchers collected saliva four times during every weekly session: upon arrival, roughly fifteen minutes into the activities, immediately after the session ended, and again twenty minutes later. That final timing matters because cortisol changes peak between five and twenty minutes after an event, so the last sample captures the hormone’s delayed response. The team then assayed two biomarkers that tell complementary stories about the body’s stress systems. Cortisol, the endocrine product of the hypothalamic-pituitary-adrenal axis, tends to rise when people feel distressed, overwhelmed, or unable to cope with novelty. Salivary alpha-amylase, a surrogate marker of sympathetic nervous system activity, rises instead during states of attention, effort, and engagement — regardless of whether the underlying emotion is positive or negative.

The feasibility results alone are noteworthy. Collecting biological samples from autistic children during active therapy is notoriously difficult, and previous studies had only managed sampling before and after horse interactions. Here, the team achieved valid samples in 99 percent of the 504 planned collections, far exceeding their 85 percent benchmark. Children held a small swab under their tongue for one minute while a researcher chatted about food to encourage salivation. Samples were frozen, shipped on dry ice, and analyzed with high-sensitivity enzyme immunoassays run in duplicate. The near-complete dataset means future studies can confidently sample during — not just around — human-equine interactions, opening a window on what actually happens physiologically while a child is on horseback.

The cortisol findings were striking. Across both groups, cortisol declined by roughly 25 percent from the start to the end of each session, a pattern partly attributable to the hormone’s natural afternoon rhythm, since all sessions ran between 1:00 and 5:30 in the afternoon. But the equine group showed an additional 23 percent reduction relative to the clinic group at the twenty-minute post-session mark, a difference that held up after statistical adjustment for adaptive behavior, clock time, seasonal effects, and changes in volunteers, horses, or therapists. The researchers applied a conservative standard, counting only changes exceeding 0.02 micrograms per deciliter and 10 percent as meaningful, and the equine effect cleared both bars. In plain terms: the same therapy produced measurably deeper relaxation when a horse was part of it.

Just as revealing was what did not change. Pre-session cortisol concentrations remained stable across all ten weeks in both groups, contradicting the researchers’ hypothesis that repeated horse exposure would progressively lower baseline stress hormones. The team suggests the intervention’s effects on the HPA axis may be dose-dependent — one hour per week may simply be too little to shift baseline physiology — or that diurnal cortisol sampling across the whole day would be needed to capture such changes. Meanwhile, alpha-amylase told a different story entirely. It spiked by about 22 percent during sessions in both groups, then recovered to near baseline within twenty minutes, with no significant difference between the clinic and equine conditions. Across the ten weeks, pre-session alpha-amylase trended upward, with the equine group showing an estimated 31 percent greater increase that fell just short of statistical significance.

Read together, these two hormonal profiles sketch a remarkably specific psychological state. Rising sympathetic arousal paired with falling cortisol is the physiological signature of engaged, positive-affect experience — the state of a child who is excited, attentive, and motivated rather than frightened or overwhelmed. This asymmetry echoes earlier developmental research showing that children with higher alpha-amylase and moderate-to-low cortisol tend to show better executive function and academic performance. The authors propose that the therapy sessions, whether on horseback or in the clinic, presented genuine cognitive challenges that temporarily elevated arousal, while the equine environment added a layer of calm that suppressed the distress signal. The horse may act as a sensory and social buffer: its rhythmic movement provides sensorimotor enrichment, and interactions with animals — a phenomenon sometimes explained through the Biophilia Hypothesis, the idea that humans are innately drawn to living things — have been shown to calm children across species from guinea pigs to dogs.

The exploratory analyses hint that these biological signatures connect to real clinical gains. Within-session cortisol reductions were meaningfully associated with ten-week improvements in social functioning, particularly social cognition, while across-week cortisol declines tracked improvements in social functioning and reduced dysphoria. The researchers speculate that the intrinsically rewarding, less verbally demanding nature of human-horse interaction may have helped children persist through socially demanding tasks — reading a horse’s reactions and adjusting their own behavior — that build social-cognitive skills. Within-session alpha-amylase changes also showed meaningful, if non-significant, correlations with improvements in emotional reactivity and irritability. Because the study was powered for its primary behavioral outcomes rather than these biomarker associations, the findings are best treated as a roadmap for larger trials rather than definitive proof of mechanism.

The trial was not without limitations. Dropout was far higher in the clinic group — 32 percent versus 6 percent in the equine group — which is itself telling about which setting families preferred, but it complicates between-group comparisons. Small effect sizes left some analyses underpowered, and alpha-amylase was measured at only one post-session timepoint, preventing any characterization of recovery trajectories. The authors also note that alpha-amylase varies by season, with lower concentrations in summer, underscoring the need for careful covariate control in future salivary bioscience work. Still, the study, registered at clinicaltrials.gov as NCT05846932, marks a turning point for a field that has long relied on caregiver questionnaires vulnerable to social desirability bias and lacking any theory of how change actually occurs.

The broader implications reach beyond the barn. By demonstrating that an enrichment experience can produce a distinct, theoretically meaningful physiological profile — calm valence with high engagement — the study lends empirical weight to the framework of biological sensitivity to context, which holds that children most affected by adversity may also benefit most from enriching environments. If future research confirms that hormone responses predict who thrives in equine-assisted therapy, clinicians could one day use simple salivary screening to match children to the interventions most likely to help them, or to identify who needs additional sessions. For now, the image that emerges from the data is vivid and quietly profound: an autistic child, mounted on a familiar horse, physiologically calm yet fully alert, learning to regulate emotions in a state that their own biology confirms is both exciting and safe. The magic, it turns out, was measurable all along.

Subject of Research: Effects of equine-assisted occupational therapy on cortisol and sympathetic nervous system activity in autistic children

Article Title: Effects of Incorporating Horses into Occupational Therapy on Adrenocortical and Sympathetic Nervous System Reactivity and Regulation in Autistic Children: A Randomized Controlled Trial

Article References: Rankins, E. M., Aichele, S., Granger, D. A., Hepburn, S., Gabriels, R., Schmid, A. A., & Peters, B. C. (2026). Effects of Incorporating Horses into Occupational Therapy on Adrenocortical and Sympathetic Nervous System Reactivity and Regulation in Autistic Children: A Randomized Controlled Trial. Journal of Autism and Developmental Disorders. https://doi.org/10.1007/s10803-026-07540-9

Image Credits: AI Generated

DOI: 10.1007/s10803-026-07540-9

Keywords: autism, equine-assisted therapy, occupational therapy, cortisol, salivary alpha-amylase, HPA axis, sympathetic nervous system, emotion regulation, randomized controlled trial, hippotherapy, human-animal interaction, salivary bioscience