Dr. Karen Parker: The Causes & Treatments for Autism

High-Level Overview

Andrew Huberman talks with Dr. Karen Parker, professor of psychiatry and director of the Social Neurosciences Research Program at Stanford, about the biology of social bonds and what science currently knows about autism. They cover why autism diagnoses have risen, how it is diagnosed, the roles of genes and environment, and the two ancient social neuropeptides, oxytocin and vasopressin. Parker then walks through her lab's research journey: from voles to a new monkey model of naturally occurring social impairment, to spinal fluid studies in children and even infants showing low vasopressin in autism, to a first clinical trial in which vasopressin nasal spray improved social abilities. The conversation closes with the gut microbiome, a failed pharmaceutical trial that did the opposite of her approach, and the discredited vaccine theory.

The tone is warm, careful and hopeful, with Parker emphasizing the heterogeneity of autism, the limits of small studies and the need for rigorous trials, and Huberman pressing on practical questions parents are likely to have. The conversation moves from clinical basics and evolutionary biology to a detailed scientific detective story and then to policy and funding.

The Discussion Explores:

• Why autism rates are rising: better and earlier detection, plus a real increase to 1 in 36 US children
• Diagnosis, genetics and environment: a behavioral diagnosis, 40 to 80% heritability and known risk factors
• What oxytocin and vasopressin do: ancient peptides for birth, bonding, water balance and social behavior
• The oxytocin story: early promise, a subgroup that may benefit and a large negative trial
• Voles and a new monkey model: how vasopressin turns on fathering and predicts low sociability
• Spinal fluid evidence in children and infants: low vasopressin tracks social symptoms and appears before diagnosis
• A vasopressin treatment trial: improved social abilities in children, with a larger trial underway
• The gut, a failed drug and vaccines: vagus-nerve links, Roche's antagonist trial and the Wakefield fraud

Deep Dive into Main Ideas

More Autism, and Better Detection

Asked whether autism is truly increasing, Parker gives a multifactorial answer. Clinicians used to diagnose children at nine or ten; now they can reliably diagnose at two or three, and pediatricians routinely use screeners. But rates have also risen: one in 36 US children now has a diagnosis, up from one in 44 about two years earlier, based on data from the CDC's 11 monitoring sites. Autism is male biased, with roughly three to four boys for every girl. Parker stresses its diversity: "if you've met one kid with autism, you've met one kid with autism."

Autism is a behavioral diagnosis made by an expert, usually a psychiatrist or psychologist, based on two core DSM-5 features: persistent social interaction challenges and restricted, repetitive behaviors. Many people also have anxiety, sensory challenges, seizures or sleep problems. Autism is highly heritable, with estimates from 40 to 80% genetic, and about half is thought to come from many common genetic variants with additive effects; single high-impact genes can produce severe intellectual disability. Autistic traits are continuously distributed in the general population, and people in intense STEM fields like engineering, physics and math tend to score higher on measures such as the Autism Quotient even without a diagnosis. Huberman recalls a residential school near UC Santa Barbara for severely affected children, where Dustin Hoffman studied for Rain Man, and asks whether autism is really many spectrums. Parker agrees that without understanding the biology, it's hard to know how many kinds of autism there are.

Why Autism Is So Hard to Study

Parker explains that the brain is difficult to access in living people: spinal fluid and brain tissue are rarely available, and children with autism often can't tolerate brain scanners. Animal models have their own problems. She says about 50% of drugs that work in animals but fail in human trials can be attributed to poorly chosen animal models, and mice lack much of the complex social cognition and visual orientation that define human social life. Known environmental risk factors include advanced parental age, severe prematurity and maternal illness during pregnancy. Huberman mentions a hypothesis from Yale's Pasko Rakic linking prenatal ultrasound to neuron migration errors in mice, which never went further. Parker argues that because every child's genetic background differs, the field needs genetically defined subgroups, such as those with fragile X, Prader-Willi or Timothy syndrome, to find which treatments work for whom, much as cancer treatment evolved toward molecular profiling.

Oxytocin and Vasopressin: Ancient Social Molecules

The two peptides are each nine amino acids long, differ by only two, and are hundreds of millions of years old, conserved across species for reproduction and social behavior. Because they are so similar, they can bind each other's four receptors, making their effects hard to disentangle. They were first known for effects in the body: oxytocin (Greek for quick birth) for uterine contractions and milk letdown, and vasopressin, also called antidiuretic hormone, for urine output and blood pressure.

A 1971 paper by zoologist Peter Klopfer, "Mother Love: What Turns It On?", proposed that oxytocin released into the brain during birth and nursing triggers maternal bonding. In sheep and goats, whose babies can run with the herd within an hour, passage through the birth canal activates oxytocin, and blocking its receptors prevents bonding. But Parker stresses bonding is species specific; primates live in extended family groups, share care and readily adopt. She shares that after a C-section, near-fatal blood clots and difficulty breastfeeding, she still loved her son intensely. In humans, intranasal oxytocin has been shown to dampen the amygdala's response to fearful images, part of why it was nicknamed the love or trust hormone, labels both think are far too simple. Parker notes that loss of social connection underlies many disorders, from autism to addiction and depression.

The Oxytocin Trials

Parker describes Lorna Wing's 1979 attempt to subtype the social features of autism, from socially avoidant children to those who are "active but odd," interested in others but struggling to read cues. Early studies gave a single dose of intranasal oxytocin (typically 24 international units), mostly to verbal males, and found improvements on tasks like reading emotions from pictures of eyes. Currently, only two FDA-approved drugs treat autism, both antipsychotics that address irritability, and none treat its core features.

In mouse models of several genetic syndromes, oxytocin in the hypothalamus is reduced, and giving oxytocin during development can restore oxytocin neuron numbers. In Parker's trial with Stanford child psychiatrist Antonio Hardan, children received oxytocin twice a day for four weeks, and those with the lowest baseline blood oxytocin benefited most. A larger multi-site phase 3 trial found no benefit, but Parker notes oxytocin degrades quickly and requires meticulous handling that is hard to maintain across sites. Once a big trial fails, she says, funding dries up, even though oxytocin appears relatively safe in children. She also cites Adam Guastella's work suggesting oxytocin may work best in young children. An earlier study of about 200 children found blood oxytocin was not a marker of autism, but lower levels tracked greater social difficulty in everyone, autistic or not. Huberman stresses the urgency parents feel during early windows of plasticity, and Parker notes clinic wait times of 12 to 18 months and later diagnoses in poorer areas, arguing for biomarker tests to prioritize at-risk children. On MDMA or psilocybin, both agree there would be reluctance to give psychedelics to children, though modified compounds might be explored.

Voles and the Vasopressin Light Switch

Parker says oxytocin always got its "day in the sun while vasopressin was left behind. In the 1990s, researchers showed vasopressin was key to male social behavior in monogamous prairie voles, promoting pair bonding with a mate and fathering, in contrast to non-monogamous montane voles in which males provide no care. As a graduate student, Parker noticed that meadow voles, normally non-monogamous, lived with females and helped raise pups under short, winter-like day lengths. When she infused vasopressin directly into the brains of these males, it was like turning on a light switch: they gathered the pups into a nest and huddled over them, while placebo did nothing. Huberman notes it shows latent circuits can be unmasked rather than built from scratch. Parker was later drawn into autism research as parent advocates, Autism Speaks and the Simons Foundation began funding the field, quoting her mother that chance favors the prepared mind."

A Monkey Model and Spinal Fluid Clues

To overcome the limits of mouse work, Parker developed a model at the California National Primate Research Center using rhesus macaques living in large outdoor groups. Her team adapted the Social Responsiveness Scale for monkeys and found that animals who spend much time alone carry more autism-like traits: they groom less, make fewer social overtures and don't return lip smacks, the macaque friendly gesture that Huberman demonstrates. Parker stresses these monkeys don't have autism but model its social features. Measuring many neurochemical systems in blood and spinal fluid, a statistical classifier sorted monkeys into low and high social with 93% accuracy, and the key driver was vasopressin in cerebrospinal fluid, not blood. Levels were stable within individuals and tracked grooming, replicated in a second group.

For human spinal fluid, which no one would fund, Parker went around colleagues, placing hot pink stickers on lumbar puncture trays so she'd be called whenever children had spinal taps for medical reasons, collecting leftover fluid with consent. In seven children with autism and seven without, CSF vasopressin alone classified 13 of 14 correctly; oxytocin showed no difference. A replication with NIH samples from Susan Swedo's research, now including girls, found lower vasopressin in autism in both sexes, and lower levels predicted greater social, but not repetitive, symptom severity.

Then came what she calls "liquid gold: John Constantino's freezer of spinal fluid taken from infants for medical reasons. After he traced about 2,000 paper records, the team found that infants later diagnosed with autism already had low CSF vasopressin, while oxytocin didn't differ. This suggests low vasopressin comes before, not after, years of social difficulty, and could help identify at-risk children early. Parker's working hypothesis is reduced vasopressin production, which her lab is testing in postmortem tissue. She describes spinal fluid as the kitchen sink of the brain." Because too little vasopressin causes central diabetes insipidus, with thirst, frequent urination and bedwetting, her team is also investigating whether a subset of autistic children show these signs. She mentions that in rats acupuncture can release vasopressin, but warns strongly against people obtaining vasopressin on their own, given its effects on blood pressure.

The Vasopressin Trial

With Hardan, Parker ran what she calls a first-in-class randomized, double-blind, placebo-controlled trial of intranasal vasopressin twice a day for four weeks in children aged six to twelve, with 17 on the drug and 13 on placebo. Using the Social Responsiveness Scale as the primary outcome, plus clinician ratings and lab tasks like identifying emotions in faces, vasopressin-treated children improved in social abilities on parent report, clinician evaluation and their own test performance. A subset also showed less anxiety and fewer repetitive behaviors, and no increase in aggression was seen. She recounts a father who was astonished to find his son making small talk with a stranger in a grocery store aisle. Placebo children could then receive the drug in an open-label extension. She stresses it was a small pilot and that a larger trial is being completed this year. The mechanism is unknown, possibly greater social motivation or attention to social cues, and she hopes future brain imaging and tracers will show where it acts. Given the apparent safety and need, she says "it actually seems unethical not to move forward" rigorously.

The Gut Connection and the Roche Trial

Huberman asks about the microbiome. Parker describes mouse studies in which giving probiotics to genetically modified mice with social deficits normalized their behavior and increased oxytocin, and in a recent study vasopressin, in the hypothalamus. The effect depended on the vagus nerve, which connects the gut directly to the hypothalamic nuclei that make these peptides; cutting it blocked the rescue. She would like to test an ear-clip vagus nerve stimulator in autistic individuals. Fecal transplant trials in autism are ongoing, Huberman notes.

By contrast, Roche tested balovaptan, a drug that blocks the V1A vasopressin receptor, most linked to social behavior. Its trial missed its primary outcome and a later trial was stopped after a futility analysis. Parker never received a convincing rationale for blocking the system; she thinks the animal evidence of vasopressin-induced aggression came from solitary hamsters and territorial prairie vole fathers, contexts where aggression is species-appropriate. Huberman sees the failed antagonist as indirectly supporting her approach.

Vaccines and Autism

Parker explains that Andrew Wakefield's paper claiming vaccines caused autism was retracted and he lost his medical license after evidence he fabricated data. Many studies since have found no relationship, and she says the scientists and physicians she knows vaccinate their own children. Huberman acknowledges listeners' concerns about the growing number of childhood vaccines and the difficulty parents have getting clear guidance. Parker calls the money spent chasing fraudulent data a tragedy, and adds that fear around the topic made many researchers avoid studying immunology and autism at all, even though some parents report immune dysregulation in their children that deserves evidence-based investigation.

Conclusion

Karen Parker's work illustrates how a disorder defined only by behavior can start to be understood biologically. Autism is common, highly heritable and extremely varied, which is why treatments that help some children can look like failures in large, mixed trials. Oxytocin may still help a subgroup, but her research points more strongly to vasopressin: low spinal fluid levels mark socially impaired monkeys, track social symptom severity in children of both sexes and appear in infants before diagnosis, and a small trial of vasopressin nasal spray improved social abilities. Larger replication, better biomarkers, genetically defined subgroups and more funding for novel ideas are the next steps. As Huberman puts it at the close, "this is how diseases are cured."

This is a summary of "Dr. Karen Parker: The Causes & Treatments for Autism" from Huberman Lab, giving you the GIST so you can decide if you want the REST. Listen to the full episode.