A meta-analysis of 17 randomized controlled trials found intranasal oxytocin had no significant effect on emotion recognition, a non-significant moderate effect on empathy, and only a small significant effect on theory of mind in people with neurodevelopmental disorders.
Hedges' g = 0.08 for emotion recognition (no effect)Despite widespread enthusiasm for oxytocin as a social cognition enhancer, pooled data from 17 randomized trials showed essentially zero improvement in the ability to recognize emotions in people with neurodevelopmental disorders
What the researchers found
Across 17 RCTs with 466 participants with neurodevelopmental disorders, intranasal oxytocin showed:
- Emotion recognition: no significant effect (Hedges' g = 0.08 — essentially zero)
- Empathy: moderate but non-significant effect (Hedges' g = 0.49)
- Theory of mind: small, significant effect (Hedges' g = 0.21)
Meta-regression found that none of the moderating variables — diagnosis type, participant age, oxytocin dose, or frequency of administration — predicted treatment response. This uniformly modest-to-null result across different conditions and dosing approaches suggests a fundamental limitation rather than a design problem that could be solved with optimization.
Why it matters
Intranasal oxytocin has been one of the most-hyped potential treatments for autism and related conditions, generating enormous public interest and research investment. This meta-analysis provides a needed reality check: the aggregate evidence from well-designed trials shows minimal clinical benefit. For families and clinicians, this is important because oxytocin nasal sprays are sometimes obtained off-label with high hopes and significant expense. The results suggest that more research is needed before intranasal oxytocin can be recommended as a treatment for social cognition deficits.
How the study worked
Systematic search of Medline, PsychINFO, and Scopus for randomized controlled trials of intranasal oxytocin on social cognition in neurodevelopmental disorders, published through July 2017. Seventeen studies met inclusion criteria (466 total participants). Meta-analysis used a random-effects model with Hedges' g as the effect size measure. Meta-regression examined potential moderators including diagnosis, age, dose, and frequency.
What this study cannot tell us
The included studies were heterogeneous in design, NDD diagnosis (autism spectrum disorder, intellectual disability, Prader-Willi syndrome, etc.), outcome measures, and oxytocin dosing protocols. The total sample of 466 participants across 17 studies yields relatively low statistical power for subgroup analyses. Studies published after July 2017 are not included. The meta-analysis could not assess long-term effects, as most studies used single-dose or short-term protocols. Publication bias, while not formally assessed in the abstract, could affect results.
How to read the evidence
This is a meta-analysis of randomized controlled trials — one of the highest levels of evidence in clinical research. The systematic approach, use of random-effects modeling, and meta-regression strengthen the conclusions. The largely null findings are robust and unlikely to be overturned by additional small studies.
When this study was published
Published in 2018 with studies through July 2017, this meta-analysis is now about 8 years old. Subsequent trials have generally confirmed these modest findings, though some newer studies have explored different dosing regimens, longer treatment durations, and biomarker-guided patient selection.
The bigger picture
The oxytocin-for-autism hypothesis has been a dominant theme in translational neuroscience for over a decade, driven by oxytocin's role as a 'social' neuropeptide and early single-study positive results. This meta-analysis is part of a growing body of work that has tempered initial enthusiasm. The disconnect between oxytocin's clear role in social behavior (from animal studies) and its weak clinical effects may reflect several issues: poor brain penetration via intranasal delivery, individual variability in the oxytocin system, or the oversimplification of treating complex social cognitive deficits with a single neuropeptide.
Questions still open
- Would chronic high-dose oxytocin administration produce different results than the single-dose or short-term protocols used in most included studies?
- Could genetic variation in the oxytocin receptor (OXTR) gene identify a subgroup of patients who respond to intranasal oxytocin?
- Should research pivot toward oxytocin receptor agonists with better brain penetration rather than the native peptide delivered intranasally?
Common questions
Does oxytocin nasal spray help with autism?
If oxytocin is the 'love hormone,' why doesn't it help with social skills?
Read the original research
Intranasal oxytocin, social cognition and neurodevelopmental disorders: A meta-analysis.
Psychoneuroendocrinology, 87, 9-19
Citation
Keech, Britney; Crowe, Simon; Hocking, Darren R. (2018). Intranasal oxytocin, social cognition and neurodevelopmental disorders: A meta-analysis.. Psychoneuroendocrinology, 87, 9-19. https://doi.org/10.1016/j.psyneuen.2017.09.022