Discover SPARK

When Does Autism Begin?

A photo of a mother and toddler

Marina Sarris

Date Revised: September 27, 2022

What if almost everything we assumed about how autism begins is wrong?

For many years, parents and researchers believed that autistic children fell into two general groups, those who were born with autism, and those who developed it in the toddler years.

The first group, which included most children on the spectrum, had developmental delays that began early and became more obvious in the toddler years and beyond.

A smaller group of children, however, seemed to develop typically until sometime after their first birthday, when they regressed into autism. These toddlers lost words and often interest in the world around them, according to their parents. Although less common, a third group may have both early delays and a loss of skills.

In recent years, new studies, conducted in new ways, have some researchers questioning these groupings, and the assumptions about how autism begins.

Two prominent researchers were blunt: newer studies “call into serious question” the idea that regression is uncommon. Instead, regression “may be the rule rather than the exception.”1 According to those and other researchers, regression often begins before most people notice it.

Emerging Ideas about Regression and Autism

How we study regression, and how we define it, matter, says psychologist Robin P. Kochel, Ph.D., a principal investigator for the SPARK autism study at Texas Children’s Hospital in Houston.2  She discusses regression in this recorded SPARK webinar.

With the publication of newer studies, Kochel says, “We went from saying a low proportion of kids with autism have regressions to saying that almost all of them do. Depending on how you define regression and how you measure it, the rates of regression are going to look really different. And that’s a problem.”

Most studies of autism and regression, particularly older ones, relied upon parents’ memories. When did their children begin talking and pointing, for example, and if they lost those skills, when did that happen?

But memories can be faulty, especially when looking back months or years afterward, Kochel says.

In one study, researchers asked parents about the onset of their children’s autism and then asked them the same questions about three years later. A significant number of parents gave different answers at the second time point. Many did not remember a regression they had reported years earlier.3

Researchers also examined home videos of autistic children, looking for clues about when their symptoms began. The researchers spotted both regressions, and early developmental delays, that many of the parents had not noticed or reported.4

At the time, subtle regressions or small delays may not have seemed worrisome or obvious to the parents, Kochel explains.

When Words Disappear

But some skills are considered so noteworthy that families often remember their loss. Children typically begin using words between 12 and 24 months. Just as parents notice each new word, they also often notice when those words vanish.

Traditionally, many studies defined regression, at least in part, as taking place if a child stopped talking. But that is only one symptom of autism, a condition that is also defined by social delays, sensory problems, and repetitive behaviors, such as rocking and hand flapping.

Researchers do not agree on how to define regression. For instance, how long does a child need to have a skill before it can be considered to be lost? What types or combination of skills ─ speech, communication, social, play, or self-help ─ should be counted in determining a regression?

Kochel is working with an international group of autism researchers to reach a consensus on regression that could affect future studies.

In the meantime, research into regression has not waited for a uniform definition to emerge.

Studying Babies Before Autism Symptoms Emerge

Rather than focusing on older autistic children who had regressed, some researchers decided to start at the beginning. They began studying babies who were too young to be diagnosed with autism. They carefully monitored their development from infancy until about age 3, observing or testing them at different time points.

In some of these studies, researchers used recordings of brain activity or scans, or studied children’s eye movements, as the children grew.

These studies are called prospective, because they measure a child’s skills as they emerge, not months or years afterward.

Researchers compared the youngsters who went on to receive an autism diagnosis to those who developed typically. That way, they hoped to see exactly when and how autism began and progressed.

In these studies, researchers enrolled the baby brothers and sisters of older children on the spectrum, because they are more likely to develop autism.5

These studies generally showed typical development in all babies during the first 6 to 10 months of age. The children who would later be diagnosed with autism often had declines or changes in their developmental progress between ages 1 and 2.1 Many of these declineswere subtle, not the overnight or dramatic stories of regression that some parents had told in previous studies that examined regression after it had occurred.

The prospective studies included the siblings of autistic children, so researchers do not know for sure if regression might be different in children who do not have a family history of autism.

One Group of Prospective Studies of Autism and Regression

At University of California Davis, autism researcher Sally Ozonoff, Ph.D., and her team focused on the social skills of infancy: gazing at faces, smiling and making eye contact, and making sounds while looking at an adult. At 6 months of age, the babies in their study had similar levels of those behaviors. But by 12 months, that had changed. The children who would later be diagnosed with autism showed a decline in those skills that continued over the months.1

That was according to observations and ratings of the research team. Did the children’s parents also notice these declines?

As part of the study, parents answered questions about their children’s current skills at different times. They were asked whether their youngsters smiled back at them when they smiled, and responded to their name right away, for example.6

Their answers also showed a decline in early social skills, as it was happening. But when asked later to look back on their children’s development, only a small proportion of those same parents said their youngsters had regressed.6

A study of 48,000 children in Norway also found that only 2 percent of the parents who prospectively reported a loss of social communication skills recalled it as being a regression when their children were 3 years old.7

Why Does it Matter?

Figuring out when and how autism emerges is important to understanding the factors that contribute to it. It could also lead to earlier recognition and intervention for autism, before it fully emerges, according to researchers Sally Ozonoff and Ana-Maria Iosif.1

Another reason to study regression is its possible link to medical conditions. Researchers want to know if children with large or sudden losses in skills are also more likely to have seizures or sleep disorders, for example, Kochel says.

A better understanding of regression may also help dispel some concerns about vaccines. Childhood vaccines are often given to toddlers between 15 to 18 months, around the time when autism symptoms may emerge or become more obvious. The timing, along with misinformation from one retracted study, have contributed to the distrust some have of vaccines.

Studies have shown there is no link between vaccines and autism.8-9 But Kochel says she still gets many questions about it. “I completely understand how a parent who witnesses a skill loss can be frantically looking for any information on what possibly caused that,” she says.

More than five years ago, she studied autistic children who did and did not have regressions, as reported by their parents. Her research team found no connection between regression and the vaccines that parents reported their children had received.10

Genes and Regression in Autism

Could genes also influence a regression? Several genetic conditions are linked to both autism and having a loss of skills. Those conditions involve genes such as SYNGAP1, SHANK3, CHD8, and MECP2, according to researcher Kristiina Tammimies, Ph.D.11

In another study, Kochel’s research team studied different categories of genes to see if they were related to regression in autism. They found that children who have variations to genes that are active before they are born are less likely to have a regression than children with variations to genes that are active during later developmental periods.12

One Family’s Experience with Regression

Cason McKee, a boy who participates in the SPARK autism study, seemed to be developing quite typically to his parents, who are both school psychologists. But around 18 months, he gradually began to lose social and speech skills. At his second birthday party, his mother looked around at the other children, and realized that Cason was not acting like them. In that instance, she knew he had autism.

She never focused on the whys of his autism. Years later, she would learn that he has a rare variation to one gene. No one knows yet if that particular gene is linked to regression, but it does contribute to autism and developmental delays.

Like many children who experience a regression, Cason eventually regained skills that had been lost. Today, he is a teen who excels in his classes and loves math, computer programming, Pokémon, and video games. He also plays baseball and tennis.

Interested in joining SPARK? Here’s what you should know.

Photo credit: Getty Images


  • Watch a SPARK webinar about regression and autism with Robin P. Kochel, Ph.D., research psychologist at Baylor College of Medicine and Texas Children’s Hospital.
  • See SPARK articles and webinars about speech and language development.


  1. Ozonoff S. and A-M Iosif Neurosci. Biobehav. Rev. 100, 296-304 (2019) PubMed
  2. Goin-Kochel R.P. Neurosci. Biobehav. Rev. 104, 116-117 (2019) PubMed
  3. Ozonoff S. et al. Autism 22, 891-896 (2018) PubMed
  4. Ozonoff S. et al. J. Am. Acad. Child Adolesc. Psychiatry 50, 796-806 (2011) PubMed
  5. Ozonoff S. et al. Pediatrics 128, e488-e495 (2011) PubMed
  6. Ozonoff S. et al. Autism Res. 11, 788-797 (2018) PubMed
  7. Havdahl A. et al. Paper presented at the International Society for Autism Research meeting, Rotterdam (2018) Abstract
  8. Institute of Medicine. Adverse Effects of Vaccines: Evidence and Causality. Washington: National Academies Press (2011) PubMed
  9. DeStefano F. et al. J. Pediatr. 163, 561-567 (2013) PubMed
  10. Goin-Kochel R.P. et al. Vaccine 34, 1335-1342 (2016) PubMed
  11. Tammimies K. Neurosci. Biobehav. Rev. 102, 208-220 (2019) PubMed
  12. Goin-Kochel R.P. et al. J. Autism Dev. Disord. 47, 3600-3607 (2017) PubMed