New research finds children exposed to “safe” levels of three EPA-monitored air pollutants show altered developmental processes in neural networks crucial to some social and cognitive functions.

Between 1955 and 1970, the United States passed five laws to monitor and regulate the country’s air pollution. Each act built upon the last and, in 1970, Congress passed the Clean Air Act, which established guidelines for “safe” levels of ambient air pollution, some of which are still used today.

But a new study suggests those standards may be too high, and as a result, children across the U.S. exposed to “safe” levels of pollution may be experiencing adverse effects to their brain development.

The study, published earlier this week in Environment International, hinges on an analysis of two sets of MRIs of children’s brains from one of the largest long-term studies of brain health and child development in the country.

Researchers from the Keck School of Medicine at USC worked with 9,497 participants in the study, establishing a baseline MRI when they were between ages nine and 10; two years later, the researchers conducted another MRI, looking for changes in brain function over time. Once they had each participant’s baseline and second MRI, the researchers ran environmental data from Environmental Protection Agency (EPA) air monitors through a model to help establish the air quality within a half mile radius of each child’s home.

What they found came as a mild surprise: children exposed to legally acceptable levels of fine particulate matter pollution (PM2.5), nitrogen dioxide (NO2) and ground-level ozone (O3) across the U.S. showed signs of altered brain connectivity in areas crucial to adolescent development.

Adolescence is “this period of vulnerability,” said Megan Herting, an associate professor of public health sciences at USC and the study’s senior author, that has “kind of been overlooked as an important time period in which air pollution might modify how the brain is growing and changing.”

To test that hypothesis, the research team established the local environmental pollution levels of particulate pollution, like soot, along with nitrogen dioxide and ozone for each participant, accounting for the season, proximity to roads and wind patterns. The researchers controlled for a participant’s sex, race, parental education level, household income and location.

Once they had each MRI and the environmental data, the team ran analyses to determine how neural networks in each participant’s brain were affected by the levels of pollution from PM2.5, NO2 and O3 in their surroundings.

They were specifically interested in brain functions that together “give rise to all sorts of cognitive and emotional functions that we use everyday,” said Devyn Cotter, a Ph.D. candidate in the department of public health sciences at USC, and the lead author on this study. The “salience network” governs how the brain switches its attention in response to external stimuli; the “frontoparietal network” affects executive functions like memory and processing speed, and the default mode network governs how people daydream.

If any of these networks were to deviate from their usual development, by forming too many or too few connections, “that could put you at higher risk down the line for emerging psycho pathologies,” said Cotter, potentially jeopardizing an adolescent’s future mental health.

“A lot of previous studies have looked at higher levels of air pollution,” said Herting, but the pollution levels in each participant’s environment “fell well below what EPA is setting as a safe threshold.” In the case of NO2, which is typically found in the emissions from cars, trucks and power plants burning fossil fuels, the average number of pollutants participants were exposed to was nearly 65 percent less concentrated than the standard set by the EPA.

The agency has not updated its NO2 annual average standard since it was first created in 1971.

These findings show “the long shadow that air pollution leaves on our health early in our lives,” said Robbie Parks, an assistant professor of environmental health sciences at Columbia University, who was not involved in the study. He called the analysis “robust,” and noted the study was one of the first to attempt to measure the direct mechanism by which pollution affects the brain.

The findings add momentum to the idea that “air pollution concentrations should be pushed even lower,” Parks said.

Marianthi-Anna Kioumourtzoglou, an associate professor and colleague of Parks’ at Columbia’s School of Environmental Health Sciences, agrees. “We need stricter standards,” said Kioumourtzoglou, who was also not involved in the study. “We need to protect children’s brains because that’s their future.”

Kioumourtzoglou isn’t so sure this study will prompt the EPA to lower its “safe” pollution standards, a term she’s weary of because for many developmental outcomes “there is no safe level of pollution.” She said that creating these standards is inherently complicated for the agency, and that “unfortunately,” sometimes “people’s health is not necessarily their first priority.”

For Herting and Cotton, their work raises even more avenues for scientific inquiry, such as identifying any potential metals or chemicals in the particulate matter affecting adolescent brain development and then asking “to what degree those are having the neurotoxic effects long term,” said Herting.

Still, Herting said, one thing is clear: “We need cleaner air even beyond what we thought was working before.”

In 2021, the World Health Organization lowered its standard for acceptable levels of PM2.5—bits of pollution smaller than 2.5 microns, or about one-thirtieth the diameter of a human hair—to 5 micrograms per cubic meter of air. The EPA’s current standard is 12 micrograms per cubic meter of air; earlier this year, the agency proposed lowering that number to between 9 and 10, “reflecting the latest health data and scientific evidence.”

It is unclear how the EPA will interpret the results from this latest study, for which it provided a portion of the funding.